diff --git "a/12894267.csv" "b/12894267.csv" deleted file mode 100644--- "a/12894267.csv" +++ /dev/null @@ -1,7123 +0,0 @@ -issuekey,created,title,description,storypoints -31070496,2020-02-20 10:49:45.821,Decouple package requirements from docker init,"### Problem to Solve - -Having different package version installed than required in the script makes problems. We need a solution, where the executable scripts (such as data processing operations, models) define what requirements they have. - -This should ideally not be as a separate requirements file but withing the script as (possibly) annotations. - -We should start by drawing how this could be done. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",1 -31067701,2020-02-20 10:07:00.866,Data visualization pipeline fails,"### Observed behaviour - -Running a data visualization pipeline fails because of this error: - -python: can't open file '/epf/model/tsne.py': [Errno 2] No such file or directory - - -### Expected behaviour - - - -### Steps to reproduce - -Go to env environment, go to cats dogs project, launch a data visualization pipeline. - -http://ec2-3-126-88-77.eu-central-1.compute.amazonaws.com:10080/mlreef/cats-dogs/commit/2988631f693d975c7523da4655aaefe9c7876884 - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -move the actual script paths out of the code -(`web/src/functions/pipeLinesHelpers.js` Line 104) - -https://gitlab.com/mlreef/frontend/-/blob/develop/web/src/functions/pipeLinesHelpers.js#L104 - -and into the data model (`web/src/dataTypes.js`) - - - -The source to path in the epf repo is this: /src/visualisation/tsne.py - - - -Additional Notes -===================== - - -relates to:",1 -31059812,2020-02-20 07:49:53.678,Automatic page refresh,"### Problem to Solve - -We need to include a better page refresh mechanism to improve UX. - -There are some views that need automatic refreshing: -* Experiment overview (after an experiment was executed) - possibly others (such as data instance, data visualization) -* Training log view (during training) - -Step 1: Discuss with @andres46 what mechanisms we could use for different views. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",8 -31058731,2020-02-20 07:06:02.411,GPU libraries missing,"### Observed behaviour - -![image](/uploads/9ad15a80e52589cafc30c071dc1e5f9f/image.png) -### Expected behaviour - -No erros. - - -### Steps to reproduce - -http://ec2-3-126-88-77.eu-central-1.compute.amazonaws.com:10080/mlreef/cats-dogs/-/jobs/36 - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -* https://gitlab.com/mlreef/frontend/-/blob/develop/src/bin/startup.sh -* https://www.tensorflow.org/install/gpu#linux_setup - -The basic infrastructure on the Develop environment seems to be installed correctly … but maybe not exactly the correct version combination - -* http://ec2-3-126-88-77.eu-central-1.compute.amazonaws.com:10080/mlreef/testo-rainer/-/jobs/50 - -**Installation** -During startup the ec2 instance executes the following script which installs the Nvidia drivers: -* https://gitlab.com/mlreef/frontend/-/blob/develop/src/terraform/develop-startup.sh",2 -31057600,2020-02-20 06:37:06.888,Create better dropdown from param with list,"### Problem to Solve - -If a param from pipelines has predefined values, as described in a list, we need to represent these better in the frontend. Currently, a dropdown is only shown if it is a boolean. - -We need to include list as an option in the epf (annotation, parsing). - -We need to show the possible list values in the frontend as a dropdown. - - -### Technical Solution -For the first iteration (Merge Request) of this implementation, it is perfectly fine to use standard browser popups - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -Link to XD: https://xd.adobe.com/view/5e5706b9-f556-4c88-7106-5edc92fef339-2866/ - -![image](/uploads/5d246a56f4a847a2c7455addbf886192/image.png)",3 -31056117,2020-02-20 05:50:51.249,Provide param description on mouse-over,"### Problem to Solve - -As a user I want to better understand what parameters really do and what they are for. Therefore, a mouse-over on the params question mark shall provide this from the EPF (param annotation). - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -Link to XD: https://xd.adobe.com/view/5e5706b9-f556-4c88-7106-5edc92fef339-2866/ - -![image](/uploads/9436e5427f73adc99fb8c244c34c16e4/image.png)",2 -31033460,2020-02-19 16:12:09.328,Setup domains,"### Problem to Solve - -We need to set up the following domains and subdomains: - -* [x] mlreef.com = application -* [x] about.mlreef.com = website -* [ ] blog.mlreef.com = blog page -* [x] docs.mlreef.com = documentation - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",2 -31032605,2020-02-19 15:53:02.163,WIP: Create new website,"### Problem to Solve - -The goal is to create the new website. - -### Technical Solution - -* We can use cookiebot to manage our cookies and cookie-popups: https://www.cookiebot.com/ - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD Landingpage: https://xd.adobe.com/view/3494c225-ed97-4d87-47d0-0c746e8c67a6-42a5/ - -Link to XD Landingpage Mobile: https://xd.adobe.com/view/d8dd7007-47b0-4d69-7b8d-bc30e6488c88-aeec/ - -Link to XD Subpages_Template: https://xd.adobe.com/view/923ee175-b21b-4b55-489d-6fea316a1ec9-e40f/",16 -31031801,2020-02-19 15:31:28.384,Render docs pages,"### Problem to Solve - -We want to have the docs pages rendered from md files. This should be the design. - -Question: I would like to add a menu as in varian b. How could we structure the docs to achieve this? - -Note: There are 3 frames, all scrollable separately from each other. The middle one, which one will scroll the most, will not affect the other ones. - - -### Technical Solution - -The docs pages need to be routed to: docs.mlreef.com - - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/6b222fdd-685b-44ad-66bd-dbb2dbe392ce-1d13/ - -![Docs_variant_b](/uploads/faa8fcf15727eecabe1f2f273fccdb35/Docs_variant_b.png) - -![Docs_variant_a](/uploads/2341fdc214a25e61531b7b142a0ad11d/Docs_variant_a.png)",8 -31020923,2020-02-19 12:00:38.363,New experiments still not showing up in experiement overview (running),"### Observed behaviour - -When creating a new experiment in dev the new experiment is not shown in under ""running"" but the pipeline is running on gitlab backend. - -### Expected behaviour - -After training, the new entry should inmediately show up and give the user the feedback that the experiment pipeline successfully started. - -### Steps to reproduce - -In dev, open cats and dogs, open experiments and start a new experiment. After that return to experiment overview. Even after refreshing the page no result. - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",2 -30987351,2020-02-18 17:37:17.416,Integrate tabular data capabilities,"### Problem to Solve - -As a User, I want to be able to see and interact with tabular data in MLReef. - -Scope for this Iteration: Simple Tabular Data View - -See a very good practice here: https://www.kaggle.com/datasets/digvijaysinhgohil/imdb-dataset-toprated-films-18982022 - -### Technical Solution - -We would implement a POC with Dash, to create the plot for the file experiment.json and a csv file with different types of data such as integer, double, string, date etc. Just to check if we can create a visualization without adding to much weight to our framework. - -To check how it should look like: -* https://dash-gallery.plotly.host/dash-salesforce-crm/ -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -* Basic level: Just showing data , upload a file a show the plot in the webpage -* Good level: Use API's to get the data and parameters to adjust the plot -* Advanced level: Accept new code from the user to create the plot, or adjust the parameters with a menu.",9 -30971617,2020-02-18 14:18:47.847,Show correct time stamps and time order,"### Observed behaviour - -Currently the times pipelines were executed (experiments, data visualization and data pipelines) seems always to be ""just now"". In addition, the list of experiments, data visualizations and data instances should always be ordered by time, the newest being on top of the list. - -### Expected behaviour - -The time should be shown like: - -- 3 seconds ago -- 5 minutes ago -- 1 hour ago -- 2 days ago - -The expires in ... can for now be deleted (we will implement this feature in the future) - -### Steps to reproduce - -Open dev environmen, open data instances, view how they are represented. - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -![image](/uploads/a1916f5280a09e0574173d0cc7fe0179/image.png) - -relates to:",1 -30940922,2020-02-17 18:24:42.788,Remove whiteout on repository load,"### Observed behaviour -While navigating back to the repository / project main page from a subpage, there is a short whiteout while the repo is loading again. - -### Expected behaviour -No complete whiteout, at least the static elements of the UI should remain visible - - -### Steps to reproduce -* Open a Project, -* Navigate to the ""Experiments"" section -* Click on ""Data"" to get back to the repository - - -### Solution - - -Additional Notes -===================== - -* Screencast Video: ![WhiteoutOnRepoLoad](/uploads/3627269d64010371ed7180b46925ee9f/WhiteoutOnRepoLoad.mov)",4 -30779301,2020-02-14 14:09:55.208,Share pics (no nudes),"### Problem to Solve -As a MLReefer I want to recieve images from @flohin in the Lost City. - - -### Technical Solution -Open Camera and shoot, selfies prefered. - -Technical equipment needed: - -* [ ] Florian -* [ ] His whip -* [ ] Selfie stick - -### Terminology - -If possible, reenact this: - -![image](/uploads/919d9abf35e5a19cf44c4f215f66c0f2/image.png) - - -Additional Notes -=====================",1 -30768057,2020-02-14 11:28:36.673,Experiments: model info extended,"### Problem to Solve -In the experiments executive summary the Algorithm (soon will be renamed to model) represents the info what model was used for training. - -Given the existence of code repositories we can now complete the information for this view. - -Relates to prior ticket: frontend #261 and MR !304 - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -![image](/uploads/90051310f293a417f000386331a7ab96/image.png)",4 -30712352,2020-02-13 11:01:13.975,Deleting data sets does not work,"### Observed behaviour - -Deleting a data instance does not work (the button does not react) - -### Expected behaviour - -It should delete the data instance and the underlying data-pipeline branch. - -### Steps to reproduce - -Go to the dev environment, -* open cats and dogs project -* go to data sets -* press the delete button to the modal view. - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution - -* [ ] Get the delete function to work -* [ ] Adapt the view to the new design, link to XD: https://xd.adobe.com/spec/98f972a6-7f0d-4c5f-469a-65c191a35b3a-eec8/ - -* [ ] Add at least one test - - - -Additional Notes -===================== - -![image](/uploads/c87efaaa8c94ea82a357789f6c895537/image.png) - - -relates to:",4 -30688982,2020-02-12 19:00:38.478,Separate Data Pipeline input and output data,"### Observed behaviour - -When creating a pipeline (eg. data processing pipeline, data visualization or experiment), the resulting branch that is created contains from its start all files present in the origin branch. - - -### Expected behaviour - -We are currently using branches to store the output of a branch. As long as we have this system, the origin files should not be in the newly created pipeline branch (such as data-pipeline/XXX branch). - - -### Steps to reproduce - -Open dev environment, start a processing pipeline, navigate to data instance (or in gitlab backend) to see the content of this branch. - - -### Terminology - -@rainerkern do you think this ticket makes sense in the perspective of (hopefully) switching to a better storing mechanism (buckets)? - - -### Solution -The technical implementation of this can look similar to how Maven does it. In Maven, there exists a general output folder and every script generates a separate folder within that. Subsequent scripts can then easily access the previous results. - -We can (and probably should) do something similar. For example, if a pipeline consists of three scripts: - -1. Every Data Pipeline script per default needs an input and an output folder. -2. The EPF (`mlreef.yaml`) takes the original `data` folder as input for the first script in the pipeline and directs the output to a temporary location (e.g. `output/1`) -3. The EPF uses `output/1` as input for the second script and directs its output to `output/2` -4. Then the EPF uses `output/2` as input for the third script and directs its output to `output/3` -5. Then, at last, the `data` folder is cleared and the contents from `output/3` are moved to `data` and subsequently persisted as the Data Instance - -**Mixing Data:** -If the user decides that he wants to original input data and the output data mixed together, in step `5` the input data is **NOT** deleted from the `data` folder - -This functionality can also be added for each step. So the data from each step is propagated to the output folder - -@cpmlreef create designs in frontend. - -**Storage:** -This description is still completely agnostic of the type of storage that will be used; be it S3, FTP, or GIT.",6 -30688756,2020-02-12 18:51:28.271,Navigate in folder tree in select data modal,"### Problem to Solve - -As a User I want to be able to navigate in the folder structure to select files within folders. This is currently not possible, as only the very parent view of the data repository is shown (and not clickable = no navigation). - -### Technical Solution - -The function should look the same as the folder tree navigation in the data repository view. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -Link to XD: https://xd.adobe.com/spec/88d8a453-2b64-43bf-71a6-a2a48c4ef2b7-fee9/ - -![image](/uploads/658fbb758ad1edf5dad206e69b9d92ea/image.png)",4 -30688493,2020-02-12 18:38:49.715,Make modal views sexy from size,"### Observed behaviour - -Currently, nearly all modal views are not in a ""propper"" size. Either they are to big, not adapting to their content, to streched or to little. - -We need to make all of them sexy! ;) - -### Expected behaviour - -The modals need to be correct in size as of their design. - -### Steps to reproduce - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution - -This ticket relates very well to this one: frontend #251 - -* [ ] Add at least one test - - - -Additional Notes -===================== - -Modal in select data: - -![image](/uploads/d7afaecf7e302d4c8329a3c2bfdd22a1/image.png) - -Execute pipeline modal: -Either to streched - -![image](/uploads/931eb6335e2f9482f7fca0e4e3a30447/image.png) - -or too little - -![image](/uploads/c4699a492eefab6a33d9b5dd72f9ed6e/image.png) - - -relates to:",4 -30688319,2020-02-12 18:29:03.662,"File/Folder name in modal data select breakes to ""early""","### Observed behaviour - -The space for file and folder names in the modal view to select files / folders is to short, therefore the name breaks very ""fast"". - -### Expected behaviour - - - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution - -Change the space for names to it fits at least 30 characters for all modal views: - -* [ ] Data visualization select data modal -* [ ] Data processing pipeline select modal -* [ ] Experiment pipeline select modal - -Additional Notes -===================== - -![image](/uploads/0bda59649abd80b0adfaf3817f9d6f40/image.png) - -relates to:",1 -30687751,2020-02-12 18:05:21.031,Deleted files have wrong lable in image diff view,"### Observed behaviour - -The diff view when files where deleted has currently the lable ""source file"" on it. This view exists for example in commit detailed views, when creating new Merge Requests or when reviewing MRs in the Changes tab. - -### Expected behaviour - -It should state ""deleted"". - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -![image](/uploads/7eceb98a58b06270075acd1df769932a/image.png) - -relates to:",1 -30614037,2020-02-11 08:00:57.696,Create merge request link from branch view does not work,"### Observed behaviour - -The link from the branch overview to create a new merge request from a branch does not work. - -### Expected behaviour - -The button links to create new MR, where the information is already pre-filled (branch source name, diff views, commit history). - -### Steps to reproduce - -In develop env, retail project, branches. - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -![image](/uploads/70ac809aac673e1b5a2208b7b022573d/image.png) - -relates to:",1 -30597571,2020-02-10 18:23:42.073,Limit character count on file names to 40,"### Problem to Solve -We have to limit the amount of characters displayed in the data repository to 40, as more will do breaks in line. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -![image](/uploads/321e2b1f59f054050a9b085d8686c2db/image.png)",1 -30589644,2020-02-10 14:53:39.178,Link branch name to branch in new merge request view,"### Problem to Solve -As a user i want to be able to review the branch by clicking on the branch name that I want to create a MR from. To do this, simply link the branch name to the corresponding branch in the data repository overview. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -By clicking on branch name -![image](/uploads/6326cc47759c2b44b43056d5606e1771/image.png) - -it links to this: - -![image](/uploads/0fcccceeba7334e893dd60177c6dcd30/image.png)",1 -30467418,2020-02-06 13:05:37.615,Forking a project repository doesnt seem to work,"### Observed behaviour - -When forking a project nothing happens. - -### Expected behaviour - -After pressing the fork button in a ""foreign"" repository, the repo shows a message being forked and then the user lands on the forked repo overview page. - -There are some logics that need to be considered when forking a project: -* [x] Forking a project is only possible, if the project is not the user´s own project. If it is the user´s own project, the project button is hidden away. -* [ ] Forking a foreign project puts the project into ""my projects"" tab -* [ ] A project from the same group can be forked -* [ ] A project from the same group will appear in the ""my projects"" tab - -Relates to: frontend#357 - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",4 -30436937,2020-02-05 17:36:36.759,Allow launching of multiple experiments in parallel," -### Expected behaviour - -For example drag and drop the same algorithm several times with different hyper-parameters set by the user. -The pipeline will run the experiments in parallel. The user can compare the results of all the experiments. - -Other case can be drag and drop different algorithms with hyper-parameters set by the user. -The pipeline will run the experiments in parallel The user can compare which model is the best for the dataset. - - -### Steps to reproduce - -Go into develop environment, open a new experiment pipeline, copy and adjust parameters for the models and execute the pipeline. To test open the experiment overview page or in our gitlab the repo CI/CD tab. - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",10 -30435961,2020-02-05 16:58:05.244,Pipeline warning: Fork project to gain access to functions,"### Problem to Solve - -There are some functions that need to be ""dissabled"" for projects that are not owned by the user. - -For example, a user will not be able to start an experiment within an open project by another user. - -If the user would want to do this, he first must fork the project and then create own experiments. The same accounts many different functions, such as creating new files, uploading files, etc. - -**This ticket needs design elements and detailed description on what functions are concerned!** - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",10 -30435514,2020-02-05 16:42:21.678,Differenitate between data instance and normal branches in dropdown,"### Problem to Solve - -**Completed** data instances should show up in dropdown menus within the pipeline views as graphically separated from ""normal"" data repo branches. - -Currently, all data instances are filtered out in the dropdown views. - -Note: Only ""normal"" branches shall be shown in the data repo overview dropdown (as it is currently is handled)! - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/spec/604591b4-1d03-4f45-4bf8-3da3b2759d8c-1674/ - -![Select_Data_Experiment_PPL_B](/uploads/022843b247ddba3a744ff07d99cb5162/Select_Data_Experiment_PPL_B.png)",2 -30339033,2020-02-03 20:28:33.325,Setup PROD Deployment Pipeline,"### Problem to Solve -Setup the deployment of a productive environment in the _Infrastructure_ repository. - -Design a release pipeline which deploys **exactly** the same binaries to the prod environment that was deployed to the test environments earlier. - -The Layout consists of three repositories: _Frontend_, _Backend_ and _Infrastructure_. Also in this setup, there is no need for an extra _develop_ branch in the Frontend repo. - -* Development in bug/feature branch - * Pipeline builds Frontend docker for branch - * Pipeline deploys to dynamic _feature-branch-environment_ with new Frontend container and the Backend container referenced in the _docker-compose.yml_ - -so far nothing new here. - -* Merge to master (get rid of the develop branch for now) - * Pipeline deploys to static _develop-environment_ with the Backend container referenced in the _docker-compose.yml_ - * Triggers _Infrastructure_ pipeline - * Take _frontend:latest_ and _backend:latest_ and tag them _frontend:master-$pipline-id_ and _backend:master-$pipline-id_ respectively - * Provide a manual job which runs the infrastructure pipeline to _frontend:master-$pipline-id_ and _backend:master-$pipline-id_ to production - -* Infrastructure Pipeline - * Provide a Job which deploys a specific set of docker images to PROD - - -**Additional:** -If we save the exact docker image tags which were used during a run for Frontend and Backend images, this would allow us to trigger the Infrastructure Pipeline from the Frontend Project as well as the backend Project and only change out the new parts; A Frontend trigger would update only the Frontend on the PROD environment without touching the backend. - - -### Technical Solution - -* [ ] If useful: Add documentation to the readme or the code - - -### Terminology -* **Instance/Environment**: A separate ""server"" running on EC2 hosting a complete, independent installation of all microservices and databases of MLReef -* **DEV/develop Environment:** The Environment used by developers for development. Can break at any time. -* **TEST Environment:** A more stable environment used for User Acceptance Testing. Should be quite stable already. `DOES NOT CURRENTLY EXIST` -* **PROD/production:** The one and only real production environment -* **staging:** Not a real Environment but rather a temporary place to stage new versions for production deployment. `DOES NOT CURRENTLY EXIST` -* **PipelineId:** A globally unique integer, automatically generated by Gitlab, reasonably human-readable, and increases in number. - - -Additional Notes -------------------- -Download Artifacts from other (most recent) pipelines [Gitlab ticket](https://gitlab.com/gitlab-org/gitlab-runner/issues/2656) - -",10 -30338847,2020-02-03 20:22:26.700,Better AMI (Amazon Machine Image) for Gitlab Runners,"### Problem to Solve -Gitlab Runners can be deployed in two different modes. The single runner which just accepts a job and executes it directly. And the multi-runner; which uses a separate driver to create AWS EC2 instances and execute jobs on them. This would offer us much more parallel computing power. - -The driver uses a configurable Amazon Machine Image (AMI) to create the new EC2 instance. - -In the Frontend project, the multi runner configuration exists, but the single runner config is used. It is located inside deployment script script -> [src/bin/deploy.sh](https://gitlab.com/mlreef/frontend/-/blob/develop/src/bin/deploy.sh) - -In order to be able to harness the power of GPUs in (future) elastically deployed Gitlab runners we need an AMI which already contains NVIDIA Cuda drivers. - -### Technical Solution -**Preferably find** or otherwise create a more specialised AMI (Amazon Machine Image) which already has the following software preinstalled: -* Docker in the newest version -* Nvidia Cuda drivers - -look to the [ec2 startup script](https://gitlab.com/mlreef/frontend/-/blob/develop/src/bin/startup.sh) for guidance - -Additionally: -* [ ] Add documentation to the readme or the code",5 -30333722,2020-02-03 17:37:38.025,Make Develop instance more accessible for developers,"### Problem to Solve -As a Developer, I like to use my locally running frontend while connected to the `develop` instance of the backend and Gitlab. - -Currently, when doing so, CORS errors are thrown by the remote backend. - -### Technical Solution -* [x] Add documentation in the Frontend's getting started guide how to connect your local instance to the DEVELOP infrastructure -* [x] Implement CORS headers in the Backend #280 - - -Additional Notes -===================== -Error, when trying to connect to the aws instance. -![image](/uploads/85cff25740227bf4a54406a64f14ff82/image.png)",1 -30171137,2020-01-31 22:54:25.498,Integration and Proxy as MicroService-Holder,"### Problem to Solve - -In the end, we want to have a working infrastructure hosted on AWS with a certain URL. -Some ports are publicly available, others must be present for maintenance (ssh). - -We have to assert the following is fixed: -* frontend is available as HTTP/HTTPS -* backend is available as HTTP/HTTPS -* gitlab is available as HTTP/HTTPS -* No cors violations are happening -* frontend can talk to backend & gitlab -* backend can talk to gitlab - -I propose the following **Solution A**: -* have the reverse proxy as the single domain entry point **mlreef.com** -* proxies mlreef.com/api to backend:8080/api -* proxies mlreef.com/gitlab to gitlab:80/gitlab with relativeRootUrl /gitlab -* configure gitlab to use the relativeRootUrl /gitlab to maintain proper redirects (mlreef.com/gitlab/user/signin, e.g.) -* frontend uses just /gitlab as the gitlab root URL, with the proxy active URLs get a proper rewrite. - * Attention: In my understanding, the API would be on /gitlab/api/v4 instead of OTHER_HOST:PORT/api/v4, be aware of that extra /gitlab in that URL -* frontend uses just /api as the service root URL, with the proxy active URLs get a proper rewrite. -* All services can run with http. The reverse proxy can be configured to act as an https endpoint - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [x] Discuss this and read into reverse-proxying -* [x] Set gitlab relativeRootUrl in docker-compose -* [x] fix nginx config: add /gitlab to gitlab:/gitlab instead of /gitlab to gitlab:/ -* [x] adapt prodution frontend to just use /gitlab and dont need any root url variables (Also: CORS!) -* [x] frontend should be able to choose different url when run in development - -Additional Notes -===================== -* https://stackoverflow.com/questions/35756663/api-gateway-vs-reverse-proxy -* https://stackoverflow.com/questions/43548363/using-api-gateway-as-reverse-proxy-for-web-applications -* https://stackoverflow.com/questions/50333994/node-js-api-gateway-nginx-as-reverse-proxy",4 -30124272,2020-01-30 17:55:53.876,Include status icons on running pipeline summary cards,"### Problem to Solve -As a user I want to easily observe if a pipeline is still running. The easiest way is to include a spinner in the ""running"" pipeline cards (see XD link). - -This needs to be done for: -* [ ] Data visualizations -* [ ] Data instances -* [ ] Experiments - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -XD link: https://xd.adobe.com/view/209888ea-007c-4cd4-6cff-e246d446ebd9-56a9/ - -![image](/uploads/50bf9eea6060afad77dc509e0fb59c5b/image.png)",6 -30122553,2020-01-30 16:54:07.598,Experiment Detail: introduce dynamic data to exp. detail view,"### Problem to Solve -In the experiment detail view there are for alpha 1 three tabs available: details, training log and files. The details view shows a top level information on the experiment execution. Much of the current data is static. - -This ticket aims at making the content dynamic for: - -**Metadata** -* [x] Experiment name: Here the name of the experiment needs to be shown (big-dolphin_30012020 without the experiment/ ) -* [x] ID: This needs to be a UID. As an experiment is currently also a branch, lets use the UID from gitlabs branch. -* [x] Created: The time format seems complicated. Let's use this format: 30.01.2020 - 15:24:34 -* [ ] Training time: Taken from gitlab pipeline runner in the format: 07:32:15 -* [x] Completed: If still running write ---, if canceled write ---, if failed write ---, if completed, write start 30.01.2020 - 18:25:13 -* [x] Owner: user_name who started the experiment - -**Source summary** -* [ ] Data source: write and link the data source, either a branch or data instance. The link goes directly to the data instance detail view or the data repo view with the selected branch. - -**Model parameters** -* [ ] List all standard and advanced parameters with the values and, if available, the description of the parameter. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",4 -30113054,2020-01-30 12:23:19.691,Upload a file in data repo,"### Problem to Solve -As a user i want to be able to upload a new file to my repository via an UI interface. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/spec/254df67a-f28e-4fff-5703-2a4183dde1e6-537c/ - -![upload_a_file_A](/uploads/1aea4bd6d86da8b54bdc360c50d20f0f/upload_a_file_A.png) - -![upload_a_file_A-1](/uploads/dfeaee4cc24e9b97b56f1bf68257475b/upload_a_file_A-1.png) - -![upload_a_file_A-2](/uploads/96cbc4cebfdd46bc06e9c797ab14c12f/upload_a_file_A-2.png)",12 -30108050,2020-01-30 11:19:36.389,Backend: Redesign Experiment domain model,"### Problem to Solve - - -There was an error/misunderstanding in the backend data model. - -It is not: Run experiments which are pipelines on data, but run DataPipelines on DataRepository to DataInstance with cleansed Data. -Run Experiment (just choose files and Model) to create Experiment output - -* Pipeline-DataInstance: DataRepository + DataPipeline-Configuration -> (DataInstance - Execution) -> DataInstance -* Adhoc-DataInstance: DataRepository + DataPipeline-Configuration -> (DataInstance - Execution) -> DataInstance -* DataRepository + DataInstance + Experiment-Configuration -> (Experiment - Execution) -> Experiment - -An Experiment is both an Experiment-Pipeline-Configuration and its Experiment-Instance. The output can be a model an some Files - -Experiment = ExperimentPipeline + ExperimentInstance -* has a name and generated slug -* has experiment.mlreef.yaml -* points to DataInstance or DataRepository -a) DataInstance - * choose files - * configuration is transient, configure Model directly -b) DataRepository+branch - * choose files - * configure Model directly, - * just info about files and branch name - * no info about DataProcessors preProcessing - - -### Technical Solution - -* [x] Refactor Experiment -* [x] - must not include preprocessing pipeline -* [x] - must point to a DataInstance -* [x] - needs a name -* [x] think about filename experiment.mlreef.yml -* [x] remove dataOperations config",4 -30107502,2020-01-30 11:07:31.987,Experiments: Training info,"### Problem to Solve -In the experiments executive summary view the Training tab shows on what parameters the experiment pipeline was run. The info can be extracted from the experiment/xxx yaml file. We currently have two classes of parameters - -a) required or standard parameters: these are required for running the experiment and are explicitly tracked in the yaml file - -b) advanced parameters: these have preset values which are defined in the data datatypes.js (location FE web?). If the values have been changed while running a parameter, these values need to be overwritten by the values in the experiment/xxx yaml file. - -> For now we simply can list all parameters (standard and advanced) each in one paragraph - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -![image](/uploads/3da9c9ebdf27cce0736309561b8e0dfa/image.png)",6 -30057280,2020-01-29 08:11:13.884,Insights / Show jobs list,"### Problem to Solve -As a user I want to see all past and current jobs. A job is for now a pipeline that hast started. - - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/07843c33-3b31-43b5-6d0f-10684160726c-71b5/ - -![Insights_Jobs_B](/uploads/c4ca7499aa28fab05221af7dce02bb50/Insights_Jobs_B.png) - -![Insights_Jobs_A](/uploads/37662b248f8cbed035a7d1c2b497b816/Insights_Jobs_A.png) - -![Insights_Jobs_C](/uploads/eb3fc7964eefc8d092292152a9ff1f62/Insights_Jobs_C.png)",4 -30030133,2020-01-28 15:33:36.170,Adapt pipeline modal (loading view),"### Problem to Solve -The current **last modal view** for creating data processing, data visualization and experiments is almost entirely static. - -Until further technical solutions are created, we will give a more lively feedback by integrating a indetermined progress bar and links to the new data isntances, data visualizations and experiments. - -* [ ] Adapt experiment modal to show experiment created with key infos and dismiss others, (e.g.hyperopt) (@cpmlreef adapt view) -* [ ] Adapt view for visualization + processing **singular only pipelines** - -### Technical Solution - -For the progress bar, use an inderterminate progress bar: https://material-ui.com/components/progress/ - -Backend won't be ready for sending the percentage so, the indeterminate is the one viable solution for the moment. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD for **data processing** modal: https://xd.adobe.com/spec/c0c2c46c-388c-4c39-518b-3af4fb97ecc9-b6f3/ - -Link to XD for **data visualization** modal: https://xd.adobe.com/view/528d7eb2-3422-4fc3-6a33-8c4ba4aaa4cb-24bc/ - -Link to XD for **experiment** modal: https://xd.adobe.com/spec/43c770a4-40c4-4b40-5917-6849c1913037-35f6/ - -![Processing_Data_C](/uploads/f22980632b95bd83fd9520f45f54834a/Processing_Data_C.png)",4 -30028944,2020-01-28 15:06:43.105,Limit commit message in commit overview,"### Observed behaviour -The commit message is to long and therefore, distorts the representation of commits in the commits overview (see additional notes). - -### Expected behaviour -We need to limit the commit message to 40 characters in order to avoid this error. - -### Steps to reproduce -Go to Sign language classifier, then to commits in the blue nav bar. - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test -* A viable solution would be to add a collapsible text so that user can see all the commit msg - - - -Additional Notes -===================== -![image](/uploads/b62b53444e58aff9a4c94f45a4baa92b/image.png) - -relates to:",2 -30018957,2020-01-28 10:43:50.392,Move delete project function to settings/general,"### Problem to Solve -As a user (admin/owner of a repository) I want to be able to delete a project. Currently, this function is present but in the ""wrong"" section (in project overview page). - -This page will change completely with the new marketplace #183 - -A new view will be created by @cpmlreef for this feature. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",4 -30013315,2020-01-28 09:59:49.640,Integrate navbar counters,"### Problem to Solve -As a user I want to be able to see the number of branches, commits, visualizations and so on in the data repository overview - blue navbar. - -These are the counters (marked with x are done, the others we would need to create): - -* [x] Commits -* [x] Branches -* [x] Visualizations -* [x] Data Instances -* [ ] Contributors - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -Additional Notes -===================== -includes / duplicates #138 - -Project stats: ![image](/uploads/658e6294e83369c58543dc4812dcaada/image.png) - - -Screenshot: - -![image](/uploads/d275ec41e4139baa3a7211afa05f3638/image.png)",4 -29998266,2020-01-27 21:01:52.186,Backend: Add Mokko,"### Problem to Solve - -Low Prio ticket: refactoring tests to use a more kotlin-style approach with Mockk instead of Mockito - -### Technical Solution - -* [ ] Add library -* [ ] Rework some tests - - -### Terminology -* https://mockk.io/",1 -29975372,2020-01-27 11:47:17.757,Select + deselect function in data select modal (all pipelines),"### Problem to Solve -As a user I want to be able to easily select or deselect possible files in the data select modal view in pipelines: - -""Select"" should only select all files and folders within the current folder. - -Meaning, if the user navigates inside the `data/` folder and then clicks *Select all*, all the files and folders within the `data/` are selected but none outside. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -![image](/uploads/7284263b84ba4df68d46169bebdad619/image.png)",2 -29975041,2020-01-27 11:36:40.529,Display images in Readme viewer,"### Problem to Solve -As a user I want to be able to see images in the readme. This is currently not possible as ""quote from @SaathvikT"": -*This was an issue I had at the beginning as well because gitlab displays images in readme by parsing them in a different way i.e. it throws them as ""img"" tags to the frontend but these ""img"" tags are converted to text (so they are not really images, if it makes sense).* - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",10 -29956773,2020-01-26 20:34:52.311,Http 500 Errors in Data Populator,"### Observed behaviour -Gitlab throws an HTTP 500 error on the first API call after installation - - -### Expected behaviour -Gitlab answers with 201 Created - - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",2 -29873087,2020-01-23 08:11:30.161,Backend: Setup integrationTests for whole stack including gitlab,"### Problem to Solve - - -### Technical Solution - - -* [x] Add integrationTest gradle task -* [x] Add at least one integration test -* [ ] ~~Replace Jedis with lettuce due to version conflicts~~ -* [x] Setup postgres, redis as easy TestContainers -* [x] Setup postgres, redis, gitlab (via testcontainers docker-compose) -* [x] Run real integration tests against gitlab - -### Terminology - - -Additional Notes -===================== - -* https://www.testcontainers.org/ -* https://www.testcontainers.org/modules/docker_compose/",4 -29679223,2020-01-16 16:41:34.905,Create user documentation (Alpha 1),"### Problem to Solve -As a *Alpha 1* user I want to have a basic documentation about how MLReef works. The user documentation should be accessible through our website and be based on markdown (hosted in the frontend repository) including a table of content. - - -### Technical Solution -These are the steps to create the documentation - -* [x] Create concept and first layout -* [x] Create main page -* [x] Create subpage structure -* [x] Create instructions for main pipelines and functions -* [ ] Incorporate md to new website - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",3 -29628396,2020-01-15 12:27:21.719,Data visualization - Overview,"### Problem to Solve -The output of a data visualization pipeline needs to be stored in a seperate view (similar to data instance view). - -![Data_Instances_Overview_A](/uploads/733feddb0d545bffc50c0de28e6a0bbe/Data_Instances_Overview_A.png) - -Link to XD: https://xd.adobe.com/spec/b52ce94c-57ce-4470-5b78-d2188589dd5a-c31d/ - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",2 -29550110,2020-01-13 16:06:46.165,Real Infrastructure with Data persistence,"### Problem to Solve -""Real"" Infrastructure with persistent database and data storage. -",10 -29451455,2020-01-09 17:22:34.505,CSS: Wrong selection button for data types,"### Observed behaviour -The button for selection the data type is only clickable at the checkmark, but not the whole button and the style is not 100% right to the specification. -![image](/uploads/0856494a020788c4cc62c4f16d22ad67/image.png) - - -### Expected behaviour -The button should be completely clickable and be of this style: - -![image](/uploads/5333421fbe74e764f728d7714ea29f7d/image.png) - -See link to XD file: https://xd.adobe.com/view/c1bef4a8-42a9-40b0-6669-00305c4adbd7-b78f/ - - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -Link to XD file: https://xd.adobe.com/view/c1bef4a8-42a9-40b0-6669-00305c4adbd7-b78f/",2 -29279212,2020-01-03 16:02:27.568,Rest-Endpoint: CRUD DataProcessors,"### Problem to Solve - -Add DataProcessors CRUD and connect with CodeProjects -Can be created with REST and through python parsing. - -Unique in ""slug"" and ""branch"", must be versioned/scoped as data-op-slug:branchname - -### Technical Solution - -* [x] CREATE DataProcessors via REST -* [x] Retrieve all DataProcessors via Rest: limit to VISIBILITY public and viewable by",10 -29278062,2020-01-03 15:05:36.212,Rest-Endpoint: CRUD DataProjects,"### Problem to Solve - -As a User I want to retrieve all my DataProjects via MLReef. -Therefore, the REST has to provide basic CRUD for DataProjects - -* Retrieve all my (viewable) DataProjects -* Create, with gitlab project creation and EPF-Bot -* Delete and Update, with gitlab creation - -### Technical Solution - -* [x] Add endpoints -* [x] Add gitlab connection for create and delete -* [x] Add rest docs",6 -29272673,2020-01-03 12:20:21.578,EPF Can load Data Operation and Algorithms from MLReef repositories,"### Problem to Solve -As a User, I want to be able to run community and custom data operations in my pipelines. In order to do that the EPF has to be able to execute Data Operations and Model Algorithms from any public, team, and private repository hosted on MLReef. - -These operations must either be actively included in the EPF during the publishing process, or downloaded and executed by the EPF on the fly, or of course a combination of the two. - - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - -* [x] R&D if pip is useful or an overkill -* [ ] assert that git clone works for public repos -* [x] ~~R&D: add concept how backend sets permissions of EPF-Bot to allow git clone~~ -* [x] make EPF use a token for a EPF-Bot user -* [x] ~~create an EPF-Bot user for every DataProject~~ -* [x] give EPF-Bot the needed permissions on used DataProcessors -* [ ] add docu",40 -29272586,2020-01-03 12:15:10.683,R&D Gitlab runners and azure driver,"### Problem to Solve - -We might switch to Azure, or support it at least. -We need to assert that Gitlab runners work with Azure. - -### Technical Solution - -* [ ] Research azure and gitlab runners -* [ ] Add docu and some examples",1 -29272533,2020-01-03 12:12:24.983,Find testing data set,"### Problem to Solve -We want to test GIT / GITLFS on local gitlab instance. For this, we need to find a good data set at least above 20GB. -Ideally, this data set can be kept for Alpha 1 usage. - - -### Technical Solution - - -Additional Notes -===================== -Link: https://www.kaggle.com/romainpessia/artificial-lunar-rocky-landscape-dataset - -Link: https://www.kaggle.com/paultimothymooney/pain-pills-in-the-usa - -Link: https://www.kaggle.com/diyer22/retail-product-checkout-dataset#20180824-13-36-25-2.jpg - -CERN: http://opendata.cern.ch/record/43",1 -29272345,2020-01-03 12:02:25.416,Kubernetes: Research and concept with Runners and GPUs,"### Problem to Solve - -Evaluate and research topics around kubernetes: - -* [x] Check if Gitlab CI runners can be spawned via Kubernetes -* [x] Check if Kubernetes and GPU machines work together -* [x] does Kubernetes have a concept of ""nodes"", e.g. ""run DB on node A and web on node B""? -* [x] does Kubernetes have a concept of node types, e.g. ""GPU node"" or ""fast node""? - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Provide some research docu -* [ ] If useful: Add documentation to the readme",4 -29272239,2020-01-03 11:56:54.475,"ReverseProxy: Solve CORS, Routing and SSL violations","### Problem to Solve -Our three services run at different ports, and thus technically on different hosts. -* frontend port 80 -* Gitlab port 10080 -* backend port 20080 - -Modern browsers have protection against Cross-origin resource sharing across different hosts. We have also seen some SSL problems with this setup. - -This has to be solved. - - -### Technical Solution - -Probably a Reverse proxy will solve any CORS issues we are currently facing - - -Additional Notes -===================== - * Pretty supported project: https://github.com/jwilder/nginx-proxy - * https://www.freecodecamp.org/news/docker-nginx-letsencrypt-easy-secure-reverse-proxy-40165ba3aee2/ - * https://serverfault.com/questions/229945/what-are-the-differences-between-haproxy-and-ngnix-in-reverse-proxy-mode - * ngnix - * haproxy - * Quarcus - * https://security.stackexchange.com/questions/48347/documented-best-practices-for-reverse-proxy-implementation",4 -29258352,2020-01-03 09:35:28.211,EPF-Annotations: Use key-value formats for parameters and dataProcessors in python files,"### Problem to Solve - -Assert that ```@parameter(name=""value"")``` instead of ```@parameter(""value"")``` is used again. - -### Technical Solution - -* [ ] Add doc",1 -28920925,2019-12-19 14:58:47.022,Get rid of warnings in the log,"### Description -There are currently a lot of warnings in the tests log which make difficult the process of reading the tests report - -### Technical solution -Get rid of those warnings by validating properly the props that the components receive in the tests - - -![image](/uploads/e840157dcd747775e80c7f5f17a1a3f4/image.png)",1 -28306029,2019-12-11 10:39:43.085,Adapt Market Place View to new style,"### Technical solution - -Adapt the current ""project-overview"" page to the new look: - -![Market_Place_alpha1](/uploads/bd943ba0f1b94a1cd39bb539a58d2b1b/Market_Place_alpha1.png) - -Link to XD: https://xd.adobe.com/view/5d9dcfd2-3bfb-4ee0-5069-8c9e2f0a5b7b-2bac/ - -Please use only the second artboard (no highlighted projects). - -### Questions - -1. As you can see, there is a header nav-bar. Currenlty, we only have the ML projects, but hopefully soon we will be able to have the others as well. Does it make sense to code this navbar now, as we only have data repositories for Alpha 1?",10 -27052804,2019-11-14 14:04:38.759,Failing test for Fileview,"### Problem to solve - -The tests are currently not working for the fileView set, the error is because of the configuration of the test which is not receiving a store object. - -### Technical solution -Utilize the newly added store factory to retrieve a store object, which provides an initial state and lets the test run. - -![image](/uploads/ed6fff5bb9e52e7fc60ae4e9d19a4990/image.png)",2 -27031862,2019-11-13 21:02:47.388,Commits view error & weird behaviour,"### Problem to solve -![image](/uploads/8529c6a8666e2103e869b004452e3fad/image.png) -![image](/uploads/2f6505a8cb0c5fc04dd6844dd1aab943/image.png) -When branches dropdown is clicked in the commits view, link changes weirdly, it should be reviewed and corrected. -The exception is raised when an option is clicked. - -### Technical solution -Control the exception by checking the nullity of the prop or look for the root of the error. - - -#### notes: -Wait until MR for #147 is merged for avoiding conflicts",2 -26931420,2019-11-11 10:53:10.949,Load previous Data Operation configuration,"### Problem to Solve -As a Data Scientist, I want to be able to load the data operation configuration from a previous experiment. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme or the code - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation - -relates to: #150 ",1 -26930908,2019-11-11 10:37:00.516,Online Textfile Editor,,5 -26927289,2019-11-11 10:13:32.842,Artifact download from experiment pipelines,"relates to: [epf#48](https://gitlab.com/mlreef/epf/issues/48) - -blocked by: #188 - - -### Feature -As a user I want to access the output files of a trained model. These are noremally the model itself and other relevant files, such as CSV files measuring the training performance. - -As a user i want to discover the files and also be able to delete the files (in case i do need to free storage space). - -The user can access the files via the FE through the experiment details view: - - -### Technical solution - -The API provided by gitlab **should** be enough for implementing this functionality. - - -### Open questions - -![Experiment_Details_Files](/uploads/204dd81d1602e396e48bc330cda82bf7/Experiment_Details_Files.png) - -Link to XD: https://xd.adobe.com/view/acf7fd8b-01b4-4903-65a6-c348a4417cb6-dac6/ - - -- We still need to define, if deleting the files in this view is a good option. Further, maybe it makes more sense to have a delete button somewhere else to delete the entire experiment (including all files). ",1 -26901721,2019-11-10 16:51:46.301,Delete a single File,"### Feature - -The delete function is accessed in the file view directly, as shown here: - -![image](/uploads/fb5d91c9f6382daa656b39ca483614fe/image.png) - -Clicking on delete will open this modal view (very similar on how GitLab does it): - -![Delete_File](/uploads/9f49da32f6b94f41c1b3a23e6516bb4e/Delete_File.png) - -When the user writes a different target branch (existing or completely new one) the selector ""start new MR... appears"": - -![Delete_File_A](/uploads/402ea7cee0253d9386f02dbb471123c1/Delete_File_A.png) - -Link to XD file: https://xd.adobe.com/spec/d4e2618b-1cab-4d2d-7224-75835533f6d2-a2d3/ - -### Technical solution - -All variants have to be included in this ticket: - -- [x] target branch is the same = simple commit -- [x] new target branch = creates new branch -- [x] new target branch + start new MR selected = creates new branch + starts new MR",1 -26901193,2019-11-10 16:09:00.329,Show Merge Request Bar when appropriate,"### Feature - -This additional information bar shows the user what commit status his current branch has: - -![image](/uploads/b0ab7d90ce8d1a75c4f0ed92e38a5584/image.png) - -This is very similar to how GitLab does it in the Branch menu: - -![image](/uploads/04e1a7c07e12a84685387d35122bbda0/image.png) - -This bar does never show up when user is in ""Master"".",2 -26901173,2019-11-10 16:07:28.629,Create New ML Project,"### Problem to solve -We need to be able to create a new ML project (repository). - -This form is very similar to how GitLab does it. We, in addition, add the ""choose data type"" function, as this will define, in what category the project can be found in the market place. - -As of now I think the best way to do this is by adding tags to the project: e.g. image, video as tags. - - -### Technical solution - -* [ ] Update the [handbook page](https://gitlab.com/mlreef/www-mlreef-com/blob/master/handbook/engineering/sops/create-gitlab-project.md) on creating projects (delete the Machine Learning projects section) - - -Additional Notes -=================== - -![New_project](/uploads/035d6f663da385a314e3edc3b71dacb3/New_project.png) - -Link to XD: https://xd.adobe.com/view/c1bef4a8-42a9-40b0-6669-00305c4adbd7-b78f/ -",1 -26901103,2019-11-10 16:02:13.115,Clicking on the avatar icon in nav bar,"### Observed Error - -When clicking in master branch on the top right avatar icon, the following message appeared: - -![image](/uploads/0c8c687c6b625987cb402b4e72452774/image.png)",1 -26901031,2019-11-10 15:56:09.408,"Data Visualization ""Drag & Drop"" Field","### Problem to solve - -Currently, the data visualization pipeline does NOT allow any pipeline creation, as the data visualizations can NOT be drag&droped to the center field. - -![image](/uploads/cfb067a392b778a8ae766b12ba6923af/image.png) - - -### Technical Solution -Use the same functionality provided in the data pipeline and experiment pipeline views. - -![image](/uploads/45ae88f09dfafe4d7dc0def06f6c7a8a/image.png)",1 -26900989,2019-11-10 15:52:09.648,Show experiment information in experiments executive summary view,"### Problem to Solve - -One of the most important parts in MLReef is experiment reproducibility. This is shown in this view: - -![image](/uploads/f67bf553f771df7fd645545463c31917/image.png) - -For Alpha 1, we need to be able to show the information for: - -- [ ] Data -- [ ] Training - -The information in the middle Tab ""Algorithm"" will only display the name of the algorithm used (Title). - -The information that will be displayed is, in the first iteration, NOT clickable (meaning it does not provide further information, e.g. when clicking on ""123 data files"" the user does not get the data files shown. This will be solved in subsequent tickets. - -This ticket needs to be coordinated with the development of our EPF (@flohin, @vaibhavmehrotra). - -Please also provide instructions (@all) if this ticket should be subdivided in smaller tickets. - -### Technical Solution - -Please provide here the technical solution for this ticket. Especially, how is the information stored and how can the FE read it.",5 -26900759,2019-11-10 15:27:36.110,Add Filtering Functionality,"### Feature - -Currently, the filter buttons do not work - lets change that ;) - -![image](/uploads/6a49110980ddfacbb37eb776748b16fa/image.png) - -This tickets is for: - -- [ ] Data Instances -- [ ] Data Visualizations -- [ ] Experiments",1 -26900638,2019-11-10 15:16:15.140,Validate pipeline input earlier,"### Problem to solve - -Currently, the FE tests the input variables in a pipeline in the modal view after pressing ""execute"": - -![image](/uploads/1714c59b014c20e3cf0b25648bbb88e1/image.png) - -The test should start one step before, as there the feedback of incorrect input variables is best: - -![image](/uploads/667eada8118c00a4d74d9673014e4962/image.png)",1 -26900582,2019-11-10 15:12:34.883,Load Pipeline Configuration,"### Problem to Solve -As a Data Scientist, I want to be able to load the experiment configuration from a previous experiment. - -An executed data pipeline will create a config file describing what exact components it used (the `mlreef.yml`) - -To recreate a data pipeline a user shall simply click the ""view pipeline"" button in the data instance view and be directed to an automatically pre-filled data pipeline view like this one, where all parameters and operations are selected according to the saved config file: - - -### Technical Implementation - -* [x] Add at least one unit test -* [ ] If useful: Add documentation to the readme or the code - -* There are two alternatives for solving this issue: -1. Parse the .mlreef.yml file to get the operations, folder selected and parameters entered by the user. -2. Persist pipelines related information in a database(maybe a snippet), so that the pipeline view can be rebuilt from that source. - -Additional Notes -==================== -![image](/uploads/9f0f822dfed388a010b52d36d7767deb/image.png) - -![image](/uploads/a681cfa2d9855b981041da3b9cd293c6/image.png) - -relates to: #164 ",5 -26900532,2019-11-10 15:07:28.548,Get user or group name in Repo View,"### Problem to solve - -Currently, the path to a project always starts with ""group name"". -![image](/uploads/1aac491135b9d6099f4b0c324eeacd2b/image.png) - -Instead, it should show the name of the author of this repository (either a user or a group).",1 -26900506,2019-11-10 15:04:47.914,Combine same experiments cards,"### Problem to Solve -This ticket is very similar to ticket #119 - Combine same cards. - -Same experiment cards in the data instance overview should be grouped together, looking like this (for expired cards - same for any other label): - -![image](/uploads/b5f0d04cadbbbcab2b3a64be99c4fc92/image.png) - -Currently same experiment cards are not grouped and it looks like this: - -![image](/uploads/8887b092affb8046f9672b80facdfa81/image.png) - -### Technical Solution -Similar solution as in commits overview - here same aged (in days) commits are grouped togther.",6 -26885667,2019-11-09 14:20:40.452,Commit Details with Diff view on Images,"### Feature - -A commit with images can have either/and these scenarios: -1. Add an image -2. Delete an image -3. Change an existing image (Diff view!) - -This commit detail view should look like this, dependent on what scenario the commit has: - -![Commits_Details_A](/uploads/8b3514a76cdd53f34f13d93573b9a2fa/Commits_Details_A.png) - -Link to XD file: https://xd.adobe.com/spec/98efca21-bcd1-40e6-5430-8855fd3efe7d-eede/",5 -26884917,2019-11-09 13:08:32.804,Limit Commit message to GUI size,"### Problem to solve - -Currently, the last commit message is to long, which takes much of the available space and squeezes the profile avatar as shown here: - -![image](/uploads/8e6bc8c463c8801b393900b6eab5b9f6/image.png) - -### Technical solution - -Limit to a maximum of 40 characters the commit message.",1 -26867257,2019-11-08 18:06:18.459,Create new Branch,"Simple form to create of a new branch from an existing branch. - -Link to XD file: https://xd.adobe.com/spec/254df67a-f28e-4fff-5703-2a4183dde1e6-537c/ - -![Data_Rep_Overview_A___1](/uploads/1d775b9ee93647257390fdee96ad444a/Data_Rep_Overview_A___1.png) - -![Data_Rep_Overview_A](/uploads/4fd340b8a94ab2bda9d85df7ce4fa732/Data_Rep_Overview_A.png)",1 -26699377,2019-11-05 08:49:21.984,Login in Frontend via Gitlab/Backend: Research and concept,"### Problem to Solve - -Users need to login via their own gitlab account. -Login mask consists of username/password, which should somehow lead to a valid authentification with gitlab. -Gitlab is now our ""user store"", therefore, direct authentication is encouraged. - -### Technical Solution - -* [x] Research, if we can forward credentials to gitlab and retrieve the token -* [x] Research only, if we can create a token via gitlab api - - -See the following tickets: - -* Backend-Issue: https://gitlab.com/mlreef/backend/issues/18 -* Backend-MR: https://gitlab.com/mlreef/backend/merge_requests/16 -* Fronted-Issue: https://gitlab.com/mlreef/frontend/issues/167",5 -25368474,2019-09-27 07:37:27.973,Forked Project does not appear in the My Projects List,"### Observed behaviour -After forking a Project, the new project does not appear immediately in the ""my-projects"" overview - - -### Expected behaviour -After clicking the fork button the new project should appear in the my projects list - - -### Steps to reproduce -* Login to master -* Go to explore projects -* Click on either Ham10000, SAR-ESA, or Demo -* Click Fork -* Go back to the my-projects view - - -### Solution -* [ ] Add at least one unit test -* [x] Add documentation to the readme or the code where it makes sense - - - -Additional Notes -===================== - - -relates to:",1 -23931221,2019-08-19 17:27:35.133,Data Visualization - Pipeline,"### Problem to solve - -Create the execution process and modal view for executing data visualization pipelines. - -![Data_Visualization_A](/uploads/bcb87b276d97bd82d2e6c761c0d24d3e/Data_Visualization_A.png) - -![Data_visualization_C](/uploads/0da9032a7d218a6132cc6328affd3c2a/Data_visualization_C.png) - -![Data_visualization_B](/uploads/257d143b17b626e21045ff0708cfc99d/Data_visualization_B.png) - -Link to XD: https://xd.adobe.com/view/528d7eb2-3422-4fc3-6a33-8c4ba4aaa4cb-24bc/ - -### Technical solution - -Reuse the processes and modal view from another pipeline (e.g. data processing pipeline). The output of a data visualization is stored as a visualization/pipeline_name branch. - -In the visualization overview this complete branch name can be used to filter out only visualizations (similar how it is done with data instances). - -",1 -23089147,2019-07-23 15:37:26.559,Add Loading for the Data Repo File Tree,"### Problem to solve - -The loading of the different content in the data repo overview takes to long. The current version, the loading spinner, is to intrusive and blocks everything. - -We need asynchonous loading in this order: -1. Loading of the folder tree and files -2. Loading of ""last commit and last date"" -3. Loading of the weights of files and quantity of files in a folder - - -### Technical Implementation - - - -## Additional Notes -While the data for ""last commit"", and ""last date"" is beeing loaded it can just be left blank and appear as soon as the data is available. As seen here: - -![image](/uploads/28265ec2d2b118faeb83ba54b8ad4901/image.png) - -To show the user that something is going on, we replace the MLReef logo in the navbar with this GIF (so does it GitLab as well): - -![MLReef_w](/uploads/2330ef8acba3e6fc8e45c71488528140/MLReef_w.gif) - -![image](/uploads/4a20611b379f1f9b2332cd4537bd79f4/image.png) - -Additional notes: - -Files API has a limit of 100 files, therefore to calculate a folder size cannot be done via API requests. For future tickets think about weight and count files in the backend and persist that information, so that there will be always a kind of report to ease the access and calculations. - -To get the last commit information, it is possible to try in the future with its own gitlab instance the next API: -https://gitlab.com/mlreef/frontend/refs/develop/logs_tree/.vscode?format=json - -For the moment, we cannot fully access this API but our own instance can be a solution.",2 -23004132,2019-07-19 15:28:03.858,Data Repo Overview Menus and Functions,"the repository overview has a number of different drop down menus and functions. -We should include the components of the variants as described in the XD files. - -Link to original data repo XD files: https://xd.adobe.com/spec/6af7ea58-6fe4-4a18-5db1-9c5e7dd598f2-1cc6/grid - -In variant B: No deeper functionallity here. -In variant C: The search function will not work, but showing the branches should (with working links). Also the file view for images needs to work! -In variant D: Dropdown should work, but no deeper functionality needed for the moment. Clone no need to work.",5 -35322019,2020-06-02 14:50:29.807,Backend: Update REST docs for groups endpoints,"### Problem to Solve - -As a frontend developer, I want to see the rest docs ;) - -### Technical Solution - -* [x] update rest-api.adoc with missing group endpoints, especially for @diegovinie -* [x] inform @diegovinie about rest docs and usable (finished, or beta) endpoints",1 -35050871,2020-05-27 08:42:33.857,"Ensure Model Output, Artifact Upload, and Frontend Artifact Download","### Problem to Solve -When a model is trained successfully it outputs a binary file. This file **should** be placed at the location `output-path`. The binary should be uploaded as a Gitlab Artifact by properly configuring the `mlreef.yml`. - -Later, a User shall be able to download the Artifact from the Frontend. - -The Scope of this ticket is to make the download work end2end - - -### Technical Solution -- [x] fix -> https://gitlab.com/mlreef/mlreef/-/merge_requests/550#note_371628096 - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -=====================",13 -35002550,2020-05-26 07:24:20.155,Update to Node 14,"Our current Node Version Node 10, even tough an LTS version is nearing its end of life: - -![image](/uploads/64259d503e24fabe7d43388461973c2b/image.png) - -In order to stay up to date with security updates and other library updates we should update to the next LTS version of node … which is Node 14. - -For this at least the following packages have to be updated: -* SCSS to 4.14+ [more info …](https://github.com/sass/node-sass#node-sass) -* Dockerfile: `web/Dockerfile` -* Script: `web/bin/node`",10 -34987707,2020-05-25 18:39:39.361,revise design on registration screens,"### Observed behaviour - -The design seems ""out of date"" for the registration screens and need revision. - -* [ ] The dropdown menu after registration needs more entries (currently only data scientist and student): -* Data scientist -* Developer -* ML engineer -* Researcher -* Student -* Team lead - -* [ ] Toastr message on sucessfull registration: Your registration was successful. Welcome to MLReef! - -### Expected behaviour - - - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -See XD file for design: https://xd.adobe.com/view/c4069ef1-b6d4-44f5-528c-4e0da5800e5a-b237/ - -relates to:",5 -34947394,2020-05-24 19:06:38.429,Error on registration,"### Observed behaviour - -The following error is given when trying to register: - -![image](/uploads/f8d994fa03bd6922a47318a2d032009f/image.png) - -### Expected behaviour - - - -### Steps to reproduce - -Open up master branch (through mlreef.com), go sign in, go create an account and then fill all fields in. - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution - -@diegovinie lets discuss when and how we want to follow up on this bug and coordinate to open up MLReef for the general public. - - - -Additional Notes -===================== - - -relates to:",1 -34862488,2020-05-21 21:15:58.866,Remove duplicate store of project information,"### Problem to Solve - -In the redux state, there is some information replicated for selectedProject and selectedProjectUUID. - -### Technical Solution -Refactor it so that all views have the same source of information.",4 -34835954,2020-05-21 08:12:20.686,Ensure Reserved Names are Denied,"### Problem to Solve -We need to save some namespaces in projects/users/groups for operational tasks, such as (need to write this in the docs as well!): - -Not all project & group names are allowed because they would conflict with existing routes used by GitLab. - -Scope of this ticket includes: -* [ ] Unit Tests for all disallowed words (TDD) -* [x] Documentation in MLReef's REST Docs -* [ ] Documentation in MLReef's User Docs -* [ ] Collaboration with the Frontend team for adaptation of the FE pages Create Project, Create User, and Create Group for using the new endpoint - - -### Technical Solution - -Suggestion: In order to centralise valid names one (or multiple) backend endpoints should be created which allow the frontend to query if a certain string is a valid name. - -The response shall contain -* TRUE/FALSE -* The `slug` of the queried name -* ErrorMsg: If the name has already been taken by another user, group, project, etc...) - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== -**Reserved project and group names** - -For a list of words that are not allowed to be used as group or project names, see the [`path_regex.rb` file](https://gitlab.com/gitlab-org/gitlab/blob/master/lib/gitlab/path_regex.rb) under the `TOP_LEVEL_ROUTES`, `PROJECT_WILDCARD_ROUTES` and `GROUP_ROUTES` lists: - -* `TOP_LEVEL_ROUTES`: are names that are reserved as usernames or top level groups -* `PROJECT_WILDCARD_ROUTES`: are names that are reserved for child groups or projects. -* `GROUP_ROUTES`: are names that are reserved for all groups or projects. - - -**Reserved project names** - -It is currently not possible to create a project with the following names: - -* `\-` -* `badges` -* `blame` -* `blob` -* `builds` -* `commits` -* `create` -* `create_dir` -* `edit` -* `environments/folders` -* `files` -* `find_file` -* `gitlab-lfs/objects` -* `info/lfs/objects` -* `new` -* `preview` -* `raw` -* `refs` -* `tree` -* `update` -* `wikis` - -**Reserved group names** - -Currently the following names are reserved as top level groups: - -* `\-` -* `.well-known` -* `404.html` -* `422.html` -* `500.html` -* `502.html` -* `503.html` -* `abuse_reports` -* `admin` -* `api` -* `apple-touch-icon-precomposed.png` -* `apple-touch-icon.png` -* `assets` -* `autocomplete` -* `dashboard` -* `deploy.html` -* `explore` -* `favicon.ico` -* `favicon.png` -* `files` -* `groups` -* `health_check` -* `help` -* `import` -* `invites` -* `jwt` -* `login` -* `oauth` -* `profile` -* `projects` -* `public` -* `robots.txt` -* `s` -* `search` -* `sent_notifications` -* `slash-command-logo.png` -* `snippets` -* `unsubscribes` -* `uploads` -* `users` -* `v2` - -These group names are unavailable as subgroup names: - -* `\-`",3 -34818067,2020-05-20 18:32:46.648,Create model training example (tutorial),"### Problem to Solve - -We want to create several tutorials as ML projects in MLReef to show our users how they can use MLReef and also learn. - -There is already one example of a tutoria: http://mlreef.com/mlreef/tutorial-cats-and-dogs/blob/master/README.md - -Create: -- HTTPS: `https://gitlab.com/gitlab-org/gitlab.git` -. - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -=====================",5 -34794338,2020-05-20 09:18:00.511,User Avatar is not displayed in project view,"### Observed behaviour -User avatar of the latest commit section in the project view is not rendered. -![image](/uploads/7eef92beddad05addb22f1fc689820fa/image.png) - - -### Expected behaviour -Avatar is rendered in project view - -### Steps to reproduce -1. Login. -2. Select the SLC(Sign Language Classifier) project.",6 -34756580,2020-05-19 16:09:56.837,Create market place view for visitors,"### Problem to Solve -Adapted view of the market place for a visitor (not authorized). - -### Technical Solution - -Reference to: https://gitlab.com/mlreef/frontend/-/issues/539 for the last screen (loading screen). - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== -Link to XD: https://xd.adobe.com/view/5eb0f210-a2c7-435c-4e25-7e43dc0887c4-2881/",8 -34706757,2020-05-18 20:57:13.316,Issues in the experiment list,"### Observed behaviour - -![Issues_to_review](/uploads/68bc9aac3130fca8d3263d39b8bcda51/Issues_to_review.png) - -### Expected behaviour - -All the information about experiments should be rendered",2 -34699148,2020-05-18 17:58:53.122,Selecting file and then selecting file again overrides the initial file,"### Observed behaviour - -When using the uploading file function in the data / code repository overview, the selected files to upload get overwritten as soon as I open the explorer popup when uploading a new file. - -### Expected behaviour - -When selecting another file to upload (by exiting the explorer view and entering the explorer view again and selecting another file for upload) the second file should persist with the other file. - -This ticket may relate to these: -* https://gitlab.com/mlreef/frontend/-/issues/534 -* https://gitlab.com/mlreef/frontend/-/merge_requests/486 - - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",2 -34699035,2020-05-18 17:54:11.421,"Pre-uploading of files in ""files upload""","### Problem to Solve - -When I select files for upload, as soon as they appear in the list, they should automatically be uploaded. When doing so the blue bar on top of each file should fill up as the files are uploaded. - -At the moment, the files are just listed with the blue bar at 100% from the very beginning. - - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== - -![image](/uploads/27a83be8e9169fc50c14c42cb03eb742/image.png)",4 -34666466,2020-05-18 05:35:48.059,Backend: Improve Experiment's lifecycle,"### Problem to Solve - -Right now the Experiments have an UUID and a nullable PipelineInfo. -The frontend wants to use a ""number"", this exists currently within the PipelineInfo. - -IF an Experiment gets created *and* started, an Experiment has a PipelineInfo with that number (locally unique per DataProject), but if it is just created, no such PipelineInfo and therefore number exist. - -This could be strange in some situations, so it is still / again technical debt: - -### Technical Solution -* Add another number, which is scoped in MLReef, does not depend on gitlab, and exists at the exactly same time, the experiment gets created -* add a constraint: experiment_number, data_project_id -* (and also for PipelineCOnfigs): pipeline_number, data_project_id -* add semaphor/synchronized, concurrency control to avoid dunplicate local ids",3 -34574367,2020-05-14 20:05:27.884,Merging does not respond,"### Observed behaviour - -When trying to merge an MR in the MR detail view, the merge doesnt respond. - -### Expected behaviour - - - -### Steps to reproduce - -* In develop, in any existing or new project -* create a branch -* add a file -* Create a MR and try to merge the changes. -* I used one cat image as test. - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -![image](/uploads/d8419fd7bf4e9dc5a02e024020831c4b/image.png)",8 -34556321,2020-05-14 13:00:14.820,Backend: Reset Password Process,"### Problem to Solve -In order to allow users to reset their passwords, we have to implement a reset-password process in the backend. - -* Upon triggering the process, the User shall receive an email which contains a password reset link which contains some form of a one-time password reset token. - * the token can only be used once - * the token shall expire after 24 hours - * the token shall expire immediately if a new token is issued -* After following this link they can enter their new password in the frontend. -* The frontend then sends the password, together with the token to the backend. -* The backend then: - * validates the token - * saves the new password - * expires the token - * performs a login for the user - * returns the Session token to the frontend, so the frontend can ""also login"" the user - - -Part of this ticket is also documenting and communicating this process sufficiently with the frontend developer who implements #500 - -### Technical Solution -The current design requires two backend endpoints. -1. For the frontend to trigger the process -2. For the frontend to set the new password - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",8 -34546112,2020-05-14 10:18:03.343,Backend: Support Data Pipeline Creation,"### Problem to Solve -In order to take full control over the execution and management of Data Pipelines, the backend has to offer an API endpoint for pipeline creation. - -This ticket is to ensure that all necessary functionality, can be finished. - - -### Technical Solution -* see what is not working and hotfix stuff -* #459 : If name generation is removed there, we have to include it in this issue! -* attention: this issues introduces technical debt, but this is not fixable now. See #561 - -### Actual Features -* [x] : Add name generation of #459 - -Additional Notes -===================== - -* https://mlreef.gitlab.io/backend/develop/#pipelines -* https://mlreef.gitlab.io/backend/develop/#pipelineinstances",2 -34382982,2020-05-11 07:45:09.657,Website / Blog post,"### Problem to Solve - -This is the standard design and layout of our blog posts on our website. - -Blog image replaces the header section and has a paralax effect. The title and author have a dropshadow for better readibility. - - -### Technical Solution - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/19fe631f-d2dd-4b6e-5a1f-4f18ad8d790b-6fcf/",1 -34322644,2020-05-08 16:24:24.447,Error creating data project - doesn't recognize initialize_with_readme,"### Observed behaviour - - - -### Expected behaviour -Create a project - - -### Steps to reproduce -Create a project with readme - - -### Terminology -https://mlreef.gitlab.io/backend/develop/#_post_data_projects - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== -![image](/uploads/e7a49e59170b58ec5c67260b0290e166/image.png) - - -relates to:",1 -34320746,2020-05-08 15:15:12.525,User_account / settings / profile,"### Problem to Solve -As a User, I want to be able to manage my profile in MLReef. This menu is accessed in the user avatar dropdown under the settings. - - -### Technical Solution -Most, if not all, of the functionality covered in this ticket, is covered by the Gitlab API. - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/811be39f-34d5-47ea-6bcc-95409cf5f74e-f72a/ - -![Settings_Profile](/uploads/2183264d484e50e92d43b876da3bcabc/Settings_Profile.png)",14 -34268469,2020-05-07 14:24:16.051,Backend & EPF: Add static base environment,"### Problem to Solve - -For the publishing process we need to store base environments in the backend, to create some new docker containers on top of them. - -Right now we think we need to store at least: -* docker_image: String ""image:version"" -* unique_slug: String ""PYTHON_2-6-GPU"" -* name: String ""Python 2.6 Cuda GPU"" -* description: String ""This package is nice an nice python 2.6 installed"" (can be used for description, requirements.txt, packages, and maybe python_version) -* requirements: String(Long) ""represents file content of requirements.txt"" -* machine_type: String(enum) CPU, GPU -* python_version : String ""2.6"" - - -### Technical Solution - - -* [ ] add those fields and invent a Enum, this should be sufficient for now",2 -34268039,2020-05-07 14:16:45.089,EPF: Define Useful DataTypes for Application,"### Problem to Solve - -Currently we have just a pile of quick-and-dirty created DataTypes as for a first draft - -We need the some DataTypes, which can be used by EPF, Backend and Frontend. - -Now we planned them as single words, but of course we could design them as slug+name, -e.g. ""image"" and ""Image"" is quite the same for single words. -""satellite-data"" and ""Satellite Data"" may differ, but is easy to calculate the relation between slug <-> name - -### Technical Solution - -Data Types suggested - -The data types can be grouped in these classes: -- Tabular: csv files, numerical, categorical, usually each row represents a sample, each column the features. -- Time series: csv files in which each row has time information , the rows are ordered usually by time. In this case (medical signals, economic indicators, prices of goods, trajectory of a car...) -- Images: the images(png,bmp, jpg..) usually come in folders with the class name, or a csv files with images in string form. -- Voice: Recordings mp3, wav, usually come in folders with the class name. -- Video: Recording wmv, mp4, mkv, usually come in folders with the class name. -- Hierarchical (optional it can be considered a subcategory of tabular): Json files that represent data and its hierarchy, usually can be considered as tabular data. for example, positional data, landmarks for face recognition, tags of objects inside an image.. etc. - -Output data types -- model : model saved as a binary file. -- All the input datatypes apply here as well.",1 -34267450,2020-05-07 14:05:49.073,Backend: Add publication fields to DataProcessor/MarketplaceEntry,"### Problem to solve - - -### Technical Solution - -Add to DataProcessor or MarketplaceEntry: - - -* [x] ""published_at"": Timestamp? -* [x] ""entrypoint"":String (with file ending implicitly) -* [x] ""base_environment"" either as slug which refers to the enum class -* [x] ""accepted_at"" and ""accepted_by"" as time and String/ or Subject_id - -Additional Notes -===================== - -* Support a base environment text id (""slug"") in the backend for DataProcessor -* Add entrypoint/command string path to Dataprocessor -* A dataProcessor Dto has a ""is_published"" flag(available via MarketplaceEntry Link) -* backend: accepted_at, or accepted_by as fields for legal reasons AND version of publishing terms",4 -34266981,2020-05-07 13:56:14.756,Backend: Prepare DataProcessor publication,"### Problem to Solve - -The backend shall offer the users/frontend and endpoint to publish/release a DataProcessor - -* some quality measuresments can be required: - * name and description set - * input and output dataType defined -* a MarketplaceEntry is created - -Think how the annotations and paramters will be translated and saved into backend, possible another endpoint triggered by EPF - -### Technical Solution - -* [ ] endpoint with minimum input: global_slug, predefined Tags - -We decided that estimation is just for the minimal part, as we dont now anything about EPF - - -Unsure: -* [ ] creates and starts the magic gitlab release pipeline - - -",5 -34254676,2020-05-07 09:31:40.684,Reset password,"### Problem to Solve - -As a User I want to be able to reset my password, if i forgot it. - -### Technical Solution -The technical solution will depend on the backend implementation - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== -Link to XD: https://xd.adobe.com/view/3d6242a2-2056-4eb2-5494-ca251026019c-4b53/ - -![Reset_pass](/uploads/72d2d235e898e77621069e366f071a39/Reset_pass.png) - -![Reset_pass_1](/uploads/06ecdf7176faea5f9052a4bcf3446a6a/Reset_pass_1.png) - -![Reset_pass_2](/uploads/7e1618ccbc96bd3c95b2773a356f9ede/Reset_pass_2.png) - - -blocked by: #527",11 -34253407,2020-05-07 09:00:26.786,"Create Impressum, Terms and Release pages on website","### Problem to Solve - -Create the views for -* [x] Impressum -* [x] Terms of use -* [x] Release terms - -based on the current texts of mlreef.com and the XD design template - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== -Link to XD: https://xd.adobe.com/view/7324878f-065f-4544-43b4-35bb877717e6-0515/",1 -34234413,2020-05-06 19:54:54.821,Backend: Experiment's input files must be saved and ONE File must be stored in mlreef.yml,"### Problem to Solve -The files selected for an experiment are not yet being stored in the backend and hence is resulting in breaking up of certain UI components. -https://gitlab.com/mlreef/frontend/uploads/883171798cc826c18914d704f3beddeb/image.png - -The information about the files on which an experiment has to be run is not stored within the config file `.mlreef.yml` as well. - - -### Technical Solution - -Frontend can send a list of files. -Backend can store a list of files. - -But generating the mlreef.yml makes only sense with ONE path (which is usually a folder). - -So, when generating a .mlreef.yml we take just ONE path (the first of the saved requests - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -=====================",3 -34170061,2020-05-05 19:02:48.965,Populate Data Visualization in Backend,"### Problem to Solve -Currently, there are no Visualizations available in the backend. - -![visual](/uploads/015bcc6e9715ae0f877c38d15b8bd1bc/visual.PNG) - - -### Technical Solution -* [x] waiting: ensure the are python files which can run as visualisation dataProcessor -* [x] @erika.torres please help backend to define that dataVisualisation -* [x] real effort: create that in dataPopulator - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -=====================",2 -34119446,2020-05-04 16:46:44.461,Adapt navbar to not registered useres (Visitors),"### Problem to Solve - -When a User is not signed in but is exploring content on MLReef (such as an open repository), the navbar should be the ""logged out"" navbar. - - - - -### Technical Solution -> **Optional:** Add technical implementation details and the results of the ticket's discussion here. - -Navbar should be fixed (not scrolling). - - -### Terminology - - - -Additional Notes -===================== - -Link to XD - **first slide!**: https://xd.adobe.com/view/c6c402fa-d7f8-4b46-4c6f-dd6257402fb4-b0f5/ - -![Data_Rep_Overview_A_loading](/uploads/8db91a06ec142ae2f9f6d1d6a53ebb66/Data_Rep_Overview_A_loading.png)",6 -34115263,2020-05-04 15:10:53.878,Rename functions in Frontend,"### Problem to Solve - -We need to rename several functions in the frontend: - -* [ ] Data Pipeline - DataOps -* [ ] Data Instance - Datasets -* [ ] Parameters advanced - optional -* [ ] Market place ""own"" - ""my ML projects""... -* [ ] Data types in project creation (@Erika takes care of this list) -* [ ] publishing - releasing -* [ ] Publication - releases -* [ ] Experiment pipeline, Algo 1 - Model 1 - - -### Technical Solution -> **Optional:** Add technical implementation details and the results of the ticket's discussion here. - - -### Terminology - - - -Additional Notes -=====================",1 -34048272,2020-05-02 08:50:22.585,Group detailed view,"### Problem to Solve - -As a User I want to see what projects a group owns. This is done via the groups detail view, which can be accessed by clicking on a group. - - -### Technical Solution -> **Optional:** Add technical implementation details and the results of the ticket's discussion here. - - -### Terminology - - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/4cec0111-f89b-4825-6363-58d95e361fef-3add/ - -![Group_overview_projects___1](/uploads/bfdf9587b1a70403b46eea63c2e7004a/Group_overview_projects___1.png) - -![Group_overview_projects](/uploads/0ff25676010e79c77a4f576e51a7d369/Group_overview_projects.png)",4 -34008326,2020-04-30 17:06:47.803,Branch dropdown shows wrong branch - shows always master,"### Observed behaviour - -When switching to a different branch, the branch dropdown shows always the ""master"" branch. - -### Expected behaviour - -It should show the selected branch, as this is the only way to know in which branch you are. - - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -The branch is probably hardcoded or it's not updated after selecting one in the dropdown. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -![image](/uploads/c2c33db3f48baed3a55a70355e9251d8/image.png)",2 -33943318,2020-04-29 10:05:45.271,Write import data API requirements,"### Problem to Solve - -For ESA BIC we need to write and demonstrate that we did! the requirements for our data import API. We currently plan to integrate with -www.sinergise.com and radiant.earth (mlhub). EODC has agreed to integrate as well, but I am not sure on how interested they really are. Further platforms and data archives will follow. - -### Technical Solution - -Lets write a new requirements.md in the company repo. - - -### Terminology - - - -Additional Notes -===================== - -API documentation: -* http://docs.mlhub.earth/#radiant-mlhub-api -* https://github.com/radiantearth/stac-api-spec - - - -Radiant Earth is STACK (Spacial Temporal … ) API compliant definition (REST API) does not include file size right now. - -JSON with metadata including links to the actual assets. - -API follows that stack and on top of that they build a proxy for additional business logic - they can add any kind of additional functionality that we need on top of that. -STACK enables filtering and searching on a metadata level. - -Sentinel Hub also deploys a STACK- compliant API - -Authentication is needed: Simple Self Service Signup with an API key. - -How are users going to use the data and derive value from it? - -Licensing is generally open and for public use - -### light datasets -In regards to User-Experience, it would be a possibility would be to limit the users to smaller datasets.",5 -33920277,2020-04-28 19:42:29.905,Create code repository views,"### Problem to Solve - -Code repositories are very similar to data-repositories. These are the main differences: - -* Code Repos do NOT have data pipelines (visualizations, dataops, experiments) -* Code Repos have a publishing process (described in frontend#447) -* Code Repos do not have GIT-LFS as a standard enabled - -All other features currently present in our data repository can be used in code repositories. - -There are currently three types of code repositories which are also handled seperately in the backend: -1. Models -2. DataOps -3. Data Visualizations - -The only visual difference between them is the color of the global marker and the naming of functions and buttons (e.g. ""publish model"" instead of ""publish DataOps"". - -### Technical Solution -> **Optional:** Add technical implementation details and the results of the ticket's discussion here. - - -### Terminology - - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/0743573d-9d17-4819-6dd4-fb5bc8fd63cc-7172/ - -![Data_Rep_Overview_A___1](/uploads/380881fbfe4465c35214369b1e2a78f3/Data_Rep_Overview_A___1.png) - -![Data_Rep_Overview_A___2](/uploads/5be3a24fe4f577d9fa5879b36d68c231/Data_Rep_Overview_A___2.png) - -![Data_Rep_Overview_A___3](/uploads/eb80cff7bdf5cd12ac81f9b9b9247866/Data_Rep_Overview_A___3.png) - -![Data_Rep_Overview_A](/uploads/d37a4c7b4c63390a7ef807da17b85599/Data_Rep_Overview_A.png)",8 -33887640,2020-04-28 07:29:25.715,ML Project / Insights / Storage,"### Problem to Solve - -As a User I want to see how much storage my ML project ocupies. - -For now, only the total storage will be displayed. Later, we can think about breaking it down into smaller pieces, such as LFS part, artifacts, etc. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/98e48030-fed8-42e8-5b6a-3dfeb3ee0906-eed5/ - -![Insights_Storage___1](/uploads/79d58863f101e4ddf828bec2cd2077ad/Insights_Storage___1.png) - -![Insights_Storage](/uploads/fa143121def17d14b20538a881360d1e/Insights_Storage.png)",6 -33847842,2020-04-27 11:38:51.410,Move Data Pipeline Creation to the Backend,"### Problem to Solve -In order to correctly hand over management of Data- and Visualization Pipelines to the backend the Frontend has to use the Backend's API endpoints instead of directly interacting with Gitlab. - -This includes -* Creation, and starting of Data- and Visualization Pipeline -* Cancelling Data- and Visualization Pipelines -* Displaying the status of pipelines - -In connection with this, the current API code and all related entries in the `dataTypes.js` shall be removed. - -### Technical Solution -There exists a backend endpoint for data pipelines which will -create a new branch in the Git Repository -configure the .mlreef.yml for the User’s pipeline -starts the Pipeline in Gitlab. - -* [x] remove `adjectives` and `nouns` arrays from `dataTypes.js` -* [x] Add at least one unit test -* [x] If useful: Add documentation to the readme or the code - -Maybe we should links to the existing endpoints -* https://mlreef.gitlab.io/backend/develop/#pipelines -* https://mlreef.gitlab.io/backend/develop/#pipelineinstances - - -### Terminology - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",6 -33841555,2020-04-27 09:25:37.040,Use backend's Data Processors API Endpoint,"### Problem to Solve -Currently, the Frontend is rendering Data Processors (options for Experiment, Data Operations, and Visualization pipelines) based on the mock data in `dataTypes.js`. - - -### Technical Solution -In order to add more and also user-generated Data processors, the Frontend shall use the backend's API endpoint - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme or the code - - -Additional Notes -===================== -/cc @flohin",3 -33837940,2020-04-27 08:14:31.734,Add endpoint for cancelling Experiments,"### Problem to Solve -In order for the Frontend to cancel experiments, the Backend must provide an appropriate management Endpoint - - -### Technical Solution -**This endpoint shall be an Idempotent endpoint** - -The request is proxied through the backend because, - -* BE receives cancel request -* passes it on to Gitlab - * if the job is already not running, update the Backend database - * if cancel was successful => save the new status in the Backend's database -* respond to frontend - - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme or the code - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",4 -33836327,2020-04-27 07:34:57.657,Backend: Add outputPath to generation of .mlreef.yml,"### Problem to Solve -In order for the EPF to be able to correctly store a data operation's output in a certain folder, the Backend has to create the `.mlreef.yml` correctly - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - -* [ ] Add at least one unit test - - - -Additional Notes -===================== -duplicates: #507 - -see: #319",2 -33740024,2020-04-24 09:59:27.539,Remove mock data from Frontend,"### Problem to Solve -It is time to get rid of the mock data in the frontend and actually use the existing functionalities of the backend. - -Parts of `dataTypes.js` which will be refactored by in other Tasks - -### Technical Solution -Get rid of the mocked data `dataTypes.js` - -Move actual configurations to `apiConfig.js` and request endpoints for still missing data from the backend team. - -* experiments -> #265 -* dataVisuzalizatons Array -> #458 -* dataPipelines Array -> #458 -* adjectives and nouns arrays -> #459 -* [ ] enumeration pipeline statuses: can maybe stay? -* [x] `filesForExperimentsDetails`? -* [ ] `/* -------------------------- Project classifications -------------------------- */` (probably can be deleted) -* [ ] `privacyLevelsArr`?",2 -33686907,2020-04-23 11:53:09.965,Feature: Backend handles gitlab ci and pipeline starting,"### Problem to Solve - -Necessary for experiment and dataPipelines - -see #265 - -Will adapt the model: -* store GitlabPipeline info in backend (new req) -* rename metrics timestamps to gitlab pipeline info timestamps -* store local id of experiment (1 for first) as ""order"" or ""number"" -* store gitlab_pipeline_id into experiment -* fix migrations again -* fix tests - -### Technical Solution - -* Write gitlab ci tests -* add more gitlab client requests for starting pipelines and handle ci files",4 -33652980,2020-04-22 20:51:32.020,Find how to upload image files to Giltab,"### Problem to Solve -The GitLab API currently does not allow to upload images. The cause of the bug is not obvious since the documentation is really poor about this functionality, it just hints that multiple formats are accepted. - -Find the way or format in which Gitlab can process image files so that we can change avatars or upload files to repositories. - - -### Technical Solution - -### Terminology -Gitlab API, Image format - - -Additional Notes -===================== -https://docs.gitlab.com/ee/api/groups.html#new-group (see the avatar field in new group payload).",4 -33605173,2020-04-22 05:29:56.972,Make Deploy script restartable,"### Problem to Solve -Sometimes, the deploy script fails for temporary reasons. E.g: the docker registry timed out. If this happens after _Step 2_ (Gitlab port configuration). The script cannot just be restarted. - - -### Technical Solution -There are different solutions possible. Either ""surgically"" remove configurations which can lead to the script failing on a second run (e.g. Gitlab's `external_url`) - -or - -Add additional `if` or `goto` statements which cover for a Gitlab that has already been configured - - -Additional Notes -===================== -It happened in this [pipeline](https://gitlab.com/mlreef/frontend/-/jobs/521324252) - -In this the [first job](https://gitlab.com/mlreef/frontend/-/jobs/521311679), during the last `docker-compose pull` the registry timed out and the script failed - -In the [second job](https://gitlab.com/mlreef/frontend/-/jobs/521324252) the script was waiting for Gitlab at port 10081. BUT Gitlab had already been reconfigured to the final port 10080",5 -33588544,2020-04-21 18:56:41.523,Add instructions for contribute with documentation,"The following discussion from !407 should be addressed: - -- [ ] @cpmlreef commented on a [discussion](https://gitlab.com/mlreef/frontend/-/merge_requests/407#note_325036571): (+2 comments) - - > Looks really good! One little thing i noticed, the clickability of the left menu could be improved a bit. I think the best way would be to change the clickable area to the entire highlighted square. - > - > Now, folders can only be clicked in the ""down arrow""; docs can only be clicked on the letters.",1 -33560243,2020-04-21 08:45:54.096,Add visual hint for sequential nature of data processing pipelines,"### Problem to Solve - -We need to show the user that data processing pipelines are sequential, meaning that one operations follows the other. Currently, all three pipelines look the same, dispite the fact that data visualizations and experiment pipelines are not sequential but parrallel (the operations are built not on each other and are independent). - -This issue gives the user a hint that the operations in data processing pipelines are sequential. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/566ad5dc-edcf-49e3-7230-c4d0e1fa4078-205b/ - -![image](/uploads/b769a5e1f63ba034f2d991aab9824fb8/image.png)",7 -33519874,2020-04-20 14:29:33.628,Enable production build on frontend,"### Problem to Solve -In order to make the frontend deployment simpler and more reliable, we want to build the frontend in production mode. - -Think about serving the frontend via NGINX - -### Technical Solution -Possibly use our NGINX as the webserver serving the frontend application - -* [ ] Use actual react production build `npm run build` instead of _develop server_ `npm start` -* [x] Remove the SKIP_PREFLIGHT_CHECKS parameter to -* [ ] Add at least one integration test - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -* https://www.youtube.com/watch?v=wkJniy5FIbk -* https://codeburst.io/how-to-setup-nginx-for-react-a504f38f95ed",9 -33519061,2020-04-20 14:16:34.834,Research payment gateway,"### Problem to Solve -As a user I want to be able to choose between most convinient payment options (VISA, Paypal, etc?) - -From business perspective, search the most covered, lowest payment commission. - -Option A: https://stripe.com/ -* Low payment fees (no montly, 1.4% Europe & 2.9% non-Europe + 0.25€ per transaction), many payment types including local (e.g. sofort), invoicing, recurring payments, dashboard, fraud detection -* [x] Check on integration: https://stripe.com/docs/api - -Option B: https://www.authorize.net/ -* Higher payment fees (25$ per month, 1.4% Europe & 2.9% non Europe + 0.30$ per transaction), many payment types not including locals, invoicing, recurring payment, dashboard, fraud detection. -* [x] Check on integration: https://developer.authorize.net/ - -Both, Stripe and Authorize.net have -* a Java SDK -* a JavsScript SDK -* a quite extensive Developer documentation -* support for recurring billing and trial periods - -**Stripe** additionally has much more built-in support for subscriptions: -* a manager GUI for creating subscription payments https://stripe.com/en-at/billing -* full-stack (frontend and backend) examples of how to implement the payment process -* full-stack (frontend and backend) examples of how to implement a subscription process - -",1 -33519002,2020-04-20 14:15:43.702,"Replace ""SamerSBN"" with official versions","### Problem to Solve -In order to make deployment, development, and maintenance of all services easier, we want to ""SammerSBN's"" docker images with official versions everywhere (frontend/docker-compose, backend/docker-compose, backend/CI tests) - -### Technical Solution - - -### Terminology - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",4 -33505609,2020-04-20 09:12:42.140,Run pipeline with docker locally (export as .rar),"### Problem to Solve -As a User I want to be able to run my pipeline locally. This can be done by creating the pipeline in MLReef and downloading the composed docker as .rar. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",16 -33374382,2020-04-16 13:36:12.466,Re-sort routes to allow guest to access the marketplace,"### Problem to Solve -As a Guest, I want to be able to navigate through the market place. - - -### Technical Solution -* [x] Move the market place to `/` (currently login is in `/` -* [ ] Adapt `/` route to allow non authenticated users. -* [ ] Adapt navbar behaviour to non authenticated users, this includes sign in buttons. -* [x] Estudy re-write every route, due most of them start with `/my-projects` - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -* guest -> non authenticated user. - - -Additional Notes -===================== - -### Gitlab routes: -```elixir -https://gitlab.com/dashboard -https://gitlab.com/dashboard/projects -https://gitlab.com/dashboard/projects/starred -https://gitlab.com/explore -https://gitlab.com/dashboard/groups -https://gitlab.com/:username -https://gitlab.com/profile -https://gitlab.com/help -https://gitlab.com/:namespace/:project_slug/-/tree/:branch -https://gitlab.com/:namespace/:project_slug/-/blob/:branch/path/to/file/:file_name -https://gitlab.com/:namespace/:project_slug/-/blame/:branch/path/to/file/:file_name -https://gitlab.com/:namespace/:project_slug/-/commits/:branch/path/to/file/:file_name -https://gitlab.com/:namespace/:project_slug/-/edit/:branch/path/to/file/:file_name -https://gitlab.com/:namespace/:project_slug/-/project_members -https://gitlab.com/:namespace/:project_slug/pipelines -https://gitlab.com/:namespace/:project_slug/pipelines/:pipeline_id -https://gitlab.com/:namespace/:project_slug/-/jobs -https://gitlab.com/:namespace/:project_slug/-/jobs/:job_id -https://gitlab.com/:namespace/:project_slug/-/merge_requests -https://gitlab.com/:namespace/:project_slug/-/merge_requests/:merge_id -https://gitlab.com/:namespace/:project_slug/-/merge_requests/:merge_id/diffs -https://gitlab.com/:namespace/:project_slug/-/merge_requests/:merge_id/commits -https://gitlab.com/:namespace/:project_slug/-/merge_requests/:merge_id/diffs?commit_id=:commit_id -https://gitlab.com/:namespace/:project_slug/-/merge_requests/:merge_id/pipelines -https://gitlab.com/:namespace/:project_slug/edit -https://gitlab.com/:namespace/:project_slug/-/project_members -https://gitlab.com/:namespace/:project_slug/-/settings/integrations -https://gitlab.com/:namespace/:project_slug/-/settings/repository -```",6 -33269325,2020-04-14 14:29:59.630,Backend: POC for sending mails,"### Problem to Solve - -EMails have to be sent to the users, for example when registering, forgot password, many notifications, but also newsletters. - -Following constraints should be met: -* sending mails with html -* support templates, likes names and emails -* also support fully dynamically html generated emails (""those are top 5 data operations this week""), so therefore we have to be flexibile! -* should not be too expensive -* may be used for newsletters in future (in the beginning we could use external services such as Mailchimp) -* if we have newsletters, we have to support unsubscribe -* MLReef based info-letters (e.g. merge request, model published, etc) required -* Info-letters need to be manageble in MLReef and through ""unsuscribe"" in the eMail - -### Technical Solution - -* [ ] Research Apis and also sendmail -* [ ] implement for proof of concept: activation with email, firstname, lastname -* [ ] activation token in email -* [ ] custom react route to show welcome page and push to backend-endpoint -* [ ] backend-endpoint activates and maybe login the user",13 -33261385,2020-04-14 11:17:21.242,Cleanup after switch to Gitlab CAS,"### Problem to Solve -After having switched to Gitlab our main CAS we need also to change the registration process in the backend and get rid of unused entities in the backend's database. - -Additionally, to be compliant with the frontend's design the registering endpoint needs to be expanded to contain the following fields - -![image](/uploads/fbd0c6bd3f4c63d9992bb8d009fe220d/image.png) - -### Technical Solution -* [ ] remove now obsolete database entities -* [ ] streamline the registration process -* [ ] add missing fields and store the values in Gitlab -* [ ] adapt the frontend to correctly call our API with all fields - -Additional Notes -=================== - -relates to: -* https://gitlab.com/mlreef/backend/-/merge_requests/51 -* #352",2 -33258518,2020-04-14 10:15:59.705,Simplify Gitlab configuration,"The following discussion from !400 should be addressed: - -- [x] @flohin started a [discussion](https://gitlab.com/mlreef/frontend/-/merge_requests/400#note_323222306): - - > I really think we should omit the HOST and PORT split, as it has no advantages over URL - > - > for example, our GITLAB_ROOT_URL could be ""https://ec2-waws.something.eu/proxy/gitlab"", there are no "":"" included, but maybe that is necessary for the future ;) - - - -`frontend/src/bin/deploy.sh` -``` -echo ""# Automatically added by the deploment pipeline .gitlab-ci-deploy.yml"" >$DOCKER_ENV -{ - echo ""# The REACT_APP_API_GATEWAY is used by the frontend to direkt API calls"" - echo ""# The GITLAB_HOST is for gitlab to serve links correctly (see docker-compose.yml)"" - echo ""REACT_APP_API_GATEWAY=http://$INSTANCE"" - echo ""# The gitlab server always serves port 80 locally. By setting the GITLAB_PORT variable,"" - echo ""# we let gitlab know, that the container's port 80 is mapped differently from the outside."" - echo ""GITLAB_PORT=$GITLAB_PORT"" - echo ""# The GITLAB_ROOT_URL is used by the backend to connect to gitlab"" - echo ""# The hostname 'gitlab' is created by the local docker network"" - echo ""GITLAB_ROOT_URL=http://gitlab:$GITLAB_PORT"" - echo """" - echo ""# The GITLAB_ADMIN_TOKEN is shared between Gitlab and the Backend"" - echo ""GITLAB_ADMIN_TOKEN=$GITLAB_ADMIN_TOKEN"" - echo """" - echo ""# These secrets are used by Gitlab to encrypt passwords and tokens"" - echo ""# Changing them will invalidate the GITLAB_ADMIN_TOKEN as well as all other tokens"" - echo GITLAB_SECRETS_SECRET_KEY_BASE=$GITLAB_SECRETS_SECRET_KEY_BASE - echo GITLAB_SECRETS_OTP_KEY_BASE=$GITLAB_SECRETS_OTP_KEY_BASE - echo GITLAB_SECRETS_DB_KEY_BASE=$GITLAB_SECRETS_DB_KEY_BASE -} -```",2 -33254220,2020-04-14 08:43:37.168,Edit MR view,"### Problem to Solve -The edit merge request view is very similar to the ""Create New Merge Request"" but has some slight differences. - -* [ ] Title of the view is different -* [ ] Title and description are prefilled -* [ ] Save changes button changed -* [ ] Add some markdown support - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -Link to XD: https://xd.adobe.com/view/d7fb2e27-4841-4fe0-4951-6a6a4c32460e-a1ca/ - -![Edit_MergeRequest](/uploads/9f72d236c8eefe43f5fa950ff9b8a122/Edit_MergeRequest.png)",8 -33247517,2020-04-14 07:18:01.887,CAS to be implemented to all controllers,"### Problem to Solve -New Security from issue 397-CAS-proof-of-concept is implemented in Groups, CodeProjects and DataProjects controllers. We need to cover all other controllers in the project - -### Technical Solution -1. Add @PreAuthorize(...) annotation to all methods of controllers that should be closed from unauthorized access -2. Add @PostAuthorize(...) possibility for post process of returning object (it is applicable when we request objects by name or any other non-id attributes) -Example: @PostAuthorize(""hasAccessToProject(returnObject, 'MAINTAINER')"") - -* [ ] Add @PreAuthorize annotations to methods of controllers -* [ ] Add @PostAuthorize annotations to methods of controllers which search objects by name -* [ ] Add one test for success access and one test for unauthorized/insufficient rights request for each request -* [ ] If useful: Add documentation to the readme - -### Further information - -* https://docs.spring.io/spring-security/site/docs/4.0.x/reference/htmlsingle/#test-mockmvc",1 -33154584,2020-04-10 22:16:55.034,"Follow-up from ""Widen cards space and improve responsiveness""","The following discussion from !402 should be addressed: - -- [ ] @diegovinie started a [discussion](https://gitlab.com/mlreef/frontend/-/merge_requests/402#note_321852518): (+1 comment) - -> ""We should discuss the breakpoints because is important to be the same throughout the project"" - -Set globally some breakpoints that fit our necessities.",4 -33129225,2020-04-10 07:43:58.615,Review Annotations/ Decorators + Parser for data operators,"### Problem to Solve - -The current annotations /decorators + parser should be reevaluated for further simplicity. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -Vie example of annotations / decorators on random crop functions in VergeML: https://github.com/mme/vergeml/blob/master/vergeml/operations/crop.py - -View parser on VergeML: https://github.com/mme/vergeml/blob/master/vergeml/operation.py",6 -33093687,2020-04-09 10:40:18.150,Get rid of User's permanent Gitlab API token,"### Problem to Solve -Currently, the backend creates and saves a (permanent) Gitlab API token. - -In the future, we want to get rid of those permanent tokens and use either the backend's Gitlab Admin Token for almost all requests. - -If necessary the User's OAUTH-token can also be used - - -### Technical Solution -* [ ] Remove the permanent API tokens from the Data Model -* [ ] Add tests -* [ ] Add doc - - -### Terminology -* **Admin Token:** The Gitlab API Admin token that is created during the installation of MLReef. Lifetime: indefinite -* **permanent token:** The Gitlab API token created by the backend during the User's registration. Lifetime: indefinite -* **OAUTH-token:** The temporary token Gitlab creates for us during a User's login. Lifetime: 2 hours - - -Additional Notes -=================== -During Login the MLReef backend currently responds with the _permanent token_ **AND** _OAUTH token_. - -Technically, the OUATH token is issued by Gitlab after the Backend requests it. - - -**Example response** -``` -HTTP/1.1 200 OK -Content-Type: application/json;charset=UTF-8 -Content-Length: 250 - -{ - ""id"" : ""cb61374e-9a0c-4b7b-a73d-19018d592722"", - ""username"" : ""vdyhttcwyy"", - ""email"" : ""vdyhttcwyy@example.com"", - ""gitlab_id"" : 1, - ""token"" : ""doesn**********"", - ""access_token"" : ""accesstoken12345"", - ""refresh_token"" : ""refreshtoken1234567"" -} -```",2 -33091272,2020-04-09 09:51:42.968,Group_overview_projects,"### Problem to Solve -As a user, when i click on a group, i will be directed to the group_overview_projects view. - -In general, this group_overview view with all its subviews is where the user can manage its group. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/4cec0111-f89b-4825-6363-58d95e361fef-3add/ - -![image](/uploads/9fab3bd00b19849072fcff1bb212137a/image.png) - -![image](/uploads/3cf45541d09d36c3e40945249d21265f/image.png)",16 -33089812,2020-04-09 09:19:24.568,Generic Projects Endpoint,"### Problem to Solve -In order to be able to provide human-readable URLs in the Frontend (e.g. `mlreef.com/namespace/projectslug/`), the frontend must be able to get information about projects just by their `namespace` and `slug`. - - - - -### Technical Solution -Add a link to the actual project entity like (`/api/v1/data-projects/0000-00000000-000000`) so that the client can easily get additional data if needed. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -[MLReef's Glossar](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook) - -* **namespace:** The name slug of the Subject who is the owner of a project; Either the project owner's username slug or the group name slug) -* **project-slug**: Slug of the project's name -* **slug:** a simplified version of a String containing only lowercase characters and with everything except 0-9 and a-z replaced with -. Use in URLs and domain names. - -",4 -33062149,2020-04-08 17:40:14.778,Test preloading of pages,"### Problem to Solve -We want to have a preloading mechanism when entering new views. Instead of just jumping in and showing posslibly empty views (as the info still needs to get fetched by the API), we use a preloading similar to how GitLab does it. - -You click, the top left icon from gitlab animates, the nav-icon on the browser tab shows a loading, and after a while it jumps to a ""nearly"" fully loaded page. - -### Technical Solution -We need to find and agree one classic loading spinner (in the design its a placeholder!). - - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -Link to XD for Data Repo Overview: https://xd.adobe.com/view/c6c402fa-d7f8-4b46-4c6f-dd6257402fb4-b0f5/ - -![Data_Rep_Overview_A_loading](/uploads/944eb2c9b90208b750a9ac3183846659/Data_Rep_Overview_A_loading.png) -",4 -33004692,2020-04-07 16:57:39.670,Error in experiment view when no files in the project,"### Observed behavior - -![image](/uploads/813d6328bf4461afaaf95b395da3df90/image.png) - -### Expected behavior - -FE handles this error without breaking",2 -33004261,2020-04-07 16:44:57.037,Wrong bread crumb when in experiment overview,"### Observed behaviour -The breadcrum in the experiment overview is mlreef > test project > Data > Experiments - -and the ""data"" part is clickable. - -### Expected behaviour - -It should say: mlreef > test project > Experiments - -and the ""test project"" should be clickable! - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== -![image](/uploads/cb0bbc69bb5043948bb587b932fc84e7/image.png)",1 -33003761,2020-04-07 16:29:30.432,"Create MR button in data repo overview stays always ""inactive""","### Observed behaviour - -Even though commits happened compared to a master branch, the create merge request button from the data repo overview is inactive and not clickable (nothing happens) - - -### Expected behaviour -When there tare commits / changes one should be able to create a merge request. - - -### Steps to reproduce -Develop, some project, created new branch from master, commited some changes, switched to new branch on repository overview. - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -![image](/uploads/62c3e5ca7017c7a872e81119569a6512/image.png)",3 -32961062,2020-04-06 21:45:35.734,CORS error when attempting to create a group,"### Observed behavior - -An error has been thrown when calling the GitLab endpoint `/groups`. -If the same machine URL is used for the rest of the API calls everything works fine, so this error is probably caused by a problem with the Nginx gateway configuration. - -![image](/uploads/f6f730c7a6cee64a715c101c2cd8ae9c/image.png) - -### Expected behavior - -API URL does not throw a CORS error. - -### Steps to reproduce - -Click on the ""Create group"" button - -### Terminology -CORS",1 -32952480,2020-04-06 16:50:32.839,Add global project class marker to repository view,"### Problem to Solve -As a user I want to have the colored marker when navigating to the different project classes repositories (ML Project, Model, Data Operation, Data Visualization). - -The project marker is found - -a. below the navbar - -b. in the selected main tabs - -The colors change when navigating in the repository classes. - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -Link to XD: https://xd.adobe.com/view/c6c402fa-d7f8-4b46-4c6f-dd6257402fb4-b0f5/ - -![image](/uploads/ca7909ae1d361b8fba154f99c5c064bf/image.png)",4 -32833370,2020-04-03 13:06:44.493,Groups Navbar,"### Problem to Solve -As a User, I want to be able to explore groups (similar to how projects are accessed) via the navbar. - -In order to not depend on the backend, recent / often visited group can be saved on the client (Cookie,...) - -### Technical Solution - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -Link to XD: https://xd.adobe.com/view/829b266c-e18b-4d1c-6ad9-3691c79c38a3-f4aa/ - -![Groups_Navbar](/uploads/1966677c06e6f93193265db5929e5750/Groups_Navbar.png)",1 -32643905,2020-03-30 17:15:57.291,Fe Domain-driven design,"### Problem to Solve -We currently have no domain model in the frontend, hence there is a little bit of mess regarding the flow of information through the application. - - -### Technical Solution -Explore the use of DDD(Domain-driven design) in the FE so that we have a consistent information model in the projet. - - -### Terminology -(DDD) Domain-driven design - - -Additional Notes -===================== -1. https://khalilstemmler.com/articles/typescript-domain-driven-design/ddd-frontend/ -2. https://khalilstemmler.com/wiki/stable-dependency-principle/",4 -32526750,2020-03-27 08:48:09.589,Insight / Job log,"### Problem to Solve - -Through the job tab in the insight menu I want to be able to access the log from specific jobs. This view can be access by clicking on: - -![image](/uploads/98a4ca539aa7c49cf5e823992a7356fe/image.png) - -Relates to: frontend#254 - - -### Technical Solution - -The core elements of the view are almost exactly the same as from the experiment details / training log vie (add: ""class"" and change ""experiment pipeline"" to ""pipeline"" - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/324184e3-e127-48d8-5e24-568401dbcfe1-4470/ - -![Job_logs](/uploads/fe823a71ab6a5f941b96831a12a6974c/Job_logs.png)",2 -32499714,2020-03-26 17:27:59.146,User profile v1,"### Problem to Solve -As a user in Alpha I want to be able to see other user profile pages. As a first version, we only show a very limited view. - -This ticket relates to the more complete view: frontend#385 - -The information regarding every ""ACTIVE"" user in gitlab is required. The backend can provide with a user endpoint which enables the frontend to use the API and receive user information. -cc: @flohin - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/df12316e-1fd1-42a2-7977-0f93cd346306-a923/ - -![User_Profile_V1](/uploads/b123586f8e2a93489adc73b9da30e733/User_Profile_V1.png)",2 -32399011,2020-03-24 16:53:37.001,Backend: Create Marketplace Search&Filter endpoint,"### Problem to Solve - -The frontend needs to search in Marketplace, see the Spec: [3.1.3. Search the Marketplace](https://gitlab.com/mlreef/company/-/blob/master/Docs/Product/Requirements/req-3-marketplace.md#313-search-the-marketplace) - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [x] Add endpoint for paginated filtering -* [x] add textual search for title and description -* [x] add or-filter for dataType -* [x] add and-filter for TAG -* [ ] add star-filter - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -* https://dzone.com/articles/postgres-full-text-search -* https://www.digitalocean.com/community/tutorials/how-to-use-full-text-search-in-postgresql-on-ubuntu-16-04 -* https://www.postgresql.org/docs/9.5/textsearch-intro.html",6 -32398778,2020-03-24 16:47:23.541,Backend: Create Marketplace Entities and basic CRUD,"### Problem to Solve - -The backend shall provide the frontend the necessary Information for Marketplace Exploration. - -The frontend must be able manipulate and retrieve the marketplace Entries, which are connected to Projects and Processors - -needs kind of: -- title -- description -- multiple Tag -- owner -- stars (aggregated, stored) -- multiple DataTypes - -needs new entities: -- starring -- tag -- tag relation -- has datatypes relation - - -* [x] Add entity Marketplace Entry -* [x] Add sub-entity or relation for Tags -* [x] Add relation with DataTypes -* [x] link to DataProject, CodeProject, and all DataProcessors -* [x] Add endpoint for GET -* [ ] Add endpoint for POST -* [ ] Add endpoint for PUT -* [x] Add endpoint for GET all",4 -32396819,2020-03-24 16:27:05.624,Epic: Marketplace and Search (minimal),"### Problem to Solve - -See Requirements here: https://gitlab.com/mlreef/company/-/blob/master/Docs/Product/Requirements/req-3-marketplace.md - -### Technical Solution - -* [x] Backend: Create Marketplace Entries -* [x] Backend: Update Marketplace Entries -* [x] Backend: Get all Marketplace Entries (paginated) -* [x] Backend: Find Entries via filter (paginated) -* [x] Frontend: Add Datatypes to Data/Code Projects -* [x] Frontend: Use Backend's Search endpoint for final productive view",0 -32395051,2020-03-24 15:52:20.975,Adapt project creation view to include data types,"### Problem to Solve -We need to adapt the project creation view to include data types and free tags to the projects. Please be aware, that each project class (ML project, model, data operation and data visualization) is a bit unique to each other. - -Attached are the different designs for each project class! - - -### Technical Solution - -* [ ] Adapt for ML Projects -* [ ] Adapt for Models -* [ ] Adapt for Data Operations -* [ ] Adapt for Data Visualizations - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -**Note**: The texts in the different views change! - -Link to **ML Projects** XD: https://xd.adobe.com/view/c1bef4a8-42a9-40b0-6669-00305c4adbd7-b78f/ - -Link to **Models** XD: https://xd.adobe.com/view/d8195b71-15c4-427c-4cac-f566c6f569f3-ae28/ - -Link to **Data Operations** XD: https://xd.adobe.com/view/71d53cb2-9bf6-49ad-4824-c11aec44f46e-07e4/ - -Link to **Data Visualization** XD: https://xd.adobe.com/view/d747bb09-471f-4084-7c68-37da0c5c6089-a176/",2 -32348196,2020-03-23 18:26:00.519,Create complete URLs for Data pipelines View,"### Problem to Solve -As a User, I want to be able to navigate to dynamic views through a generated URL. For example, I want to be able to access an experiment detail with `mlreef.com/…/-/experiments/:experiment-id` - -We need this for the following views: - -* [x] Experiment details: mlreef.com/…/-/experiments/:experiment-id -* [x] Executed data pipelines (from datasets): mlreef.com/…/-/data-pipelines/:pipeline-id -* [x] Insights/Jobs: mlreef.com/…/-/insights/jobs/:job-id - - -Info: The experiment ID is a counter that counts how many experiments within this project were done. @flohin we need this, please! - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -Additional Notes -=====================",4 -32333233,2020-03-23 12:53:30.333,Implement a CAS - Proof of Concept,"### Problem to Solve -We need to provide group membership and security authentication. - - -### Technical Solution -Use Gitlab as Central Authentication Service - -**Permissions:** How to solve role- **AND** resource-based permissions with spring - -eg: German is Owner of repo *Cats* but has only Guest privileges of repo *Dogs* - -Simple logic can be implemented in Annotations with `SPEL` complicated logic can be implemented in e.g. Spring's RequestFilterChain or SecurityFilter. Another solution is to use aspect-oriented programming -**Data Access / Caching:** Token handling can be handled by caching the JWT tokens / the whole user Blob in Redis - -**OAuth Tokens:** will be sent in the response body - -* [ ] How does the EPF authenticate against the backend? - During the creation of Pipelines, the EPF receives a token from the backend. This token is scoped as narrow as possible for a single project or Gitlab pipeline id (or even experiment / model / …) -* [ ] Will environment variables persist between pipeline retries - [gitlab docu](https://docs.gitlab.com/ee/api/pipelines.html#retry-jobs-in-a-pipeline) -* [ ] Should the Frontend create groups directly in Gitlab? - - -### Terminology -> Define terms that are important in the context of this task - -Additional Notes -===================== -Additional info can be found in the [Epic](#389) - -/cc @si-ge-st @flohin",4 -32261630,2020-03-21 12:55:42.719,Several bugs in commit detail views,"### Observed behaviour - -In the commits detail view (inital commit), there are several bugs / missing infos - -* [x] ... authored 4 days ago (the timer seems mocked) -* [x] Commit message has mocked text -* [x] ... showing xyz files in xzy additions... the numbers in the initial commit is 0, despite having many new files - -In another commit (from a pipeline by entering the branch view and selecting the commit code): - -* [ ] the counters of xyz files changed shows changes, but these are not shown further down - -### Expected behaviour - - - -### Steps to reproduce - -Open develop, open cats_dogs project, open commits view and select ""initial commit"" for the first bugs. -Same procedure but select ""branches"" and there select the commit ID of any branch for the second bug. - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -Link to XD design: https://xd.adobe.com/view/489d424c-6bd9-41c5-43cd-0ad0f220adf4-3eac/ - -First bug list (from commits view): -![image](/uploads/cb12f63580addcaaaa4858ceb5e32fa4/image.png) - -Second bug list (from branches view): -![image](/uploads/902887ce8c215ca2cb5d568261d2ea9c/image.png)",2 -32261426,2020-03-21 12:38:23.584,Deleting operators in pipelines view mixes up what it deletes,"### Observed behaviour - -In a pipeline (doesnt matter which), when I duplicate an operator with the duplication function and then, afterwards, delete the first operations (pressing the x button), the second (duplicated) operator gets deleted, not the first (where i clicked the delete button). - -### Expected behaviour - - - -### Steps to reproduce - -Develop environment, open up data pipeline, drag an operator to the list and write something in a para field to keep track of it, duplicate the operator, delete the first operator. The second will be deleted, not the first. - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -![image](/uploads/428ef262cf648dbfb4f17467fbca144a/image.png)",6 -32245478,2020-03-20 18:23:17.631,Some views were broken after style were merged,"This is a list of errors caused by the last change in the styles, so everyone is encouraged to contribute with this list. - -Every error (or suggestion) should be numbered. - -### Error reported status: - -1. done -2. done -3. done -4. done -5. pending -6. done -7. done -8. done -9. done -10. pending -11. done -12. pending -13. done -14. done -15. done -16. done -17. done -## The list of errors: - -1. ![image](/uploads/37f17d980f356b7a89e0abab5934871c/image.png) - -2. Navbar should spread wide -![image](/uploads/ed193b3d25d2ea7787eb2da1c742af35/image.png) - -3. Experiment card buttons and space -![image](/uploads/8d0ca8d3cc69929a4afe481494688017/image.png)",2 -32196095,2020-03-19 17:23:43.217,Almost all automatic tests are broken,"### Observed behavior - -The automatic tests got broken after installing the new scss support libraries -![image](/uploads/f6ada36bc25e965b3e039ed211277958/image.png) - -### Expected behavior - -All tests should run correctly without errors - -### Steps to reproduce - -Execute `npm test` - -### Possible Solution -> File extensions have to be included in the configuration of the tests.",2 -32191238,2020-03-19 15:13:35.327,File view has some dummy content,"### Observed behaviour - -In the file view there is still some dummy content that we need to change to dynamic: - -- Commit date is set at 4 days instead of real time passed -- Contributors is set to three without the avatars of the contributors - -### Expected behaviour - -Expeted to be dynamic. - -### Steps to reproduce - -1. Open develop environment -2. Open a project with some files in the repositorory -3. Open a file withing the repository to access the file view - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== -How it should look like from XD file: https://xd.adobe.com/spec/808428ab-eb22-4924-5713-82a0e4109865-f6b5/ - -How it looks like: -![image](/uploads/469a8bb81d00f86006b97534fff04bf4/image.png)",5 -32138941,2020-03-18 14:54:36.472,ML Project / settings / members,"### Problem to Solve -As a user I want to be able to change who is member in a project. This is done through the settings / members view. - -Only the ""maintainer"" role can view this page. There are two different types of member invitations, individuals and groups. The main difference is, that when inviting an entire group, all group members get access with a defined maximum permission level. - -**Permissions:** -How to solve role- **AND** resource-based permissions with spring - -eg: German is Owner of repo _Cats_ but has only Guest of repo _Dogs_ - -Simple logic can be implemented in Annotations with `SPEL` complicated logic can be implemented in e.g. Spring's RequestFilterChain or SecurityFilter. Another solution is to use aspect-oriented programming - - -### Technical Solution Backend - - -**Data Access / Caching:** -Token handling can be handled by caching the JWT tokens / the whole user Blob in Redis - - -### Terminology - -* Searchable members are all MLReef users. They can be within the group or external users. -* Inevitable groups can only be sub-groups within the project group. External groups will NOT appear in the search and can not be invited. - -Additional Notes -===================== -Link to XD: https://xd.adobe.com/view/06099f51-b2c6-4532-4020-800b4f7e5efa-7038/ - -![Settings_Overview___2](/uploads/7ba6c330b030e93a9f9aa6830a8f1c26/Settings_Overview___2.png) - -![Settings_Overview___1](/uploads/8110f6f2290589ec59882cac342805ad/Settings_Overview___1.png) - -![Settings_Overview___3](/uploads/0f5493ff3bf1a39f2a0b66553b22e214/Settings_Overview___3.png) - -![Settings_Overview](/uploads/c2be2dd626fa4eac24de28514350eb3c/Settings_Overview.png)",12 -32136636,2020-03-18 13:58:47.690,Frontend: Use MLReef-Backend for Experiments,"### Problem to Solve - -The frontend has to use the Backend endpoints for Experiment manipulation - -There are the following here: https://mlreef.gitlab.io/backend/feature/265-refactor-experiments/#_get_data_projects - -``` -Experiments - /data-projects/:id/experiments -GET …​/experiments -GET …​/experiments/:id -POST …​/experiments # create new one, update is not supported -GET …​/experiments/:id/metrics -PUT …​/experiments/:id/metrics # just useful when called by EPF! -GET …​/experiments/:id/mlreef-file -POST …​/experiments/:id/start # starts the execution of this Experiment -``` - -### Technical Solution - -* [x] Create a DataProject (https://mlreef.gitlab.io/backend/feature/265-refactor-experiments/#_post_data_projects) -* [x] Use that DataProject to get the ID and the gitlab Project -* [x] Create Experiment via that REST call -* [x] Get the MLReef file (for debug purposes? ) -* [x] Start the experiment -* [ ] Get the Metrics and update that chart in GUI -* [x] Get the log output via gitlab rest - -@flohin is here to help! - -@SaathvikT -See also frontend#265 and feel free to ""order"" more endpoints or more information you need!",5 -31683084,2020-03-06 17:04:23.335,Create a list containing all the standirized components,"### Problem to Solve -We currently create new components(inputs, buttons, radio buttons, etc) all the time when developing a new feature. - - -### Technical Solution -List all the components that we are currently using for basic FE needs so that the application looks uniform. - -Additional notes: -The list should be added in one of the manuals available.",2 -31552473,2020-03-04 15:19:28.665,Find 2 - 3 data visualizations per data type,"### Problem to Solve -For alpha 2 we need to have some data visualizations ready to use for our alpha users (availbale in the visualization pipeline). We need to have at least 2-3 data visualization per data type: - -**Images** -* tsn-e (already available) -* [PCA](https://scikit-learn.org/stable/auto_examples/decomposition/plot_pca_iris.html) - - -**Text** -* [Wordcloud](https://github.com/amueller/word_cloud) - -**Tabular** -* [Histograms, Scatter Plots](https://towardsdatascience.com/introduction-to-data-visualization-in-python-89a54c97fbed) - - -### Technical Solution - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",4 -31552374,2020-03-04 15:16:28.433,Find 10 data processing operations for each data type,"### Problem to Solve (The code is available in EPF, this story cannot be finalized) -For alpha 2 we want to have a set available for data processing for the alpha users to use. These operations need to rely on the top three data types, therefore we need to find (list) at least 10 data operations per data type and list them here (ideally with source code). - -**Images** (Done) -* Augment (already available contains 7 transformations in one script) -* Random Crop : Randomly crops the image to a fixed size. -* Lee filter -* Color Jitter: Randomly change the brightness, contrast and saturation of an image. -* Random affine: transformation of the image keeping center invariant. -* Random Erasing: Randomly selects a rectangle region in an image and erases its pixels. -* Add noise: Add different types of noise to images. -* Image Rotation: Rotates the image. -* Thumbnail: Creates thumbnail from image. - -(Done) -**Text (the concept is different here we cannot fabricate synthetic data with this transformations)** -* Bag of Words: Transforms the words to vector.(Done) -* TF-IDF: Calculates Term Frequency and Inverse Document Frequency of the words and documents. -* Stemming and lemmatization: The goal of both is to reduce inflectional forms and - sometimes derivationally related forms of a word to a common base form. -* Stopwords filtering: Remove the words that are present in all texts but don't provide meaning. Like articles, connectors, adverbs etc.(Done) -* Special Characters filtering: Some texts may contain special characters and that is not a desirable for training. (Done) -* Converting all letters to lower or upper case (Done) -* converting numbers into words or removing numbers (Done) -* removing punctuations, accent marks and other diacritics (Done) -* removing white spaces (Done) -* expanding abbreviations (not applicable) -* removing stop words, sparse terms, and particular words (Done) -* text canonicalization - -**Tabular** -* Normalization: Normalize the data with mean and standard deviation. -* Filtering: Remove some data using some criteria, for example, ""remove all rows with NULL values"". -* Data Augmentation: Create random points with normal distribution parameters. -* Numerical to Categorical: Transform numbers to categories using ranges. -* Categorical to Numerical: Transform categories to a number. -* Data completion: Use the most frequent value of a feature to complete null values in other rows. -* Add Noise: Add white noise to a series of numbers (column), commonly used to avoid overfitting. - - - - - - -### Technical Solution -Code samples for image transformation -https://github.com/pytorch/vision/blob/master/torchvision/transforms/transforms.py -https://pytorch.org/docs/stable/torchvision/transforms.html - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",8 -31552177,2020-03-04 15:11:15.800,Publish models and dataops in productions,"### Problem to Solve -We want to offer a good amount of script samples to increase engagement from users in the page. - - -### Technical Solution - -Publish models and datasets that are interesting for the users, in order to attract more attention to the site and also show the features of MLReef. -- [x] Datasets and models for image classification: Resnet, mobilenet, inception, vgg, etc. -- [x] Chatbot and text dataops -- [ ] GAN's, photo2cartoon (this one requires GPU) -- [ ] Music tagging -- [ ] Sentiment Analysis with BERT - -* Age prediction -- too heavy to be used now but the scripts are tested - -* Demand Forecasting --require fbprophet still not installed due to image size concerns - -* Sentiment Analysis -- requires language models still not installed due to image size concerns - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -=====================",8 -31345538,2020-02-27 15:17:55.339,The select data div dissapers after selecting data in experiments,"### Observed behaviour - -In the experiments pipeline, after I select the data files through the modal, the data block dissapers from the data pipeline: - -![image](/uploads/3b9827cfa716ac1de5b96e254b3a2e2f/image.png) - -**Note**: This is only in the experiment pipeline. For the other pipelines (data visualization, data processing) this works perfectly fine. - -### Expected behaviour - -It should show this: - -![image](/uploads/cd9518c958401192a3978c532eb3b2b0/image.png) - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",1 -31344825,2020-02-27 14:56:57.426,Avoid complete new page load with spinner for data repo overview,"### Observed behaviour - -When opening the data tab coming from e.g. the experiments tab, the page goes whit with the spinner on the top left screen. This really feels odd. - -### Expected behaviour - -The same handling as coming from the data tab to open the experiment tab - a smooth transition is expected. - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",2 -31344727,2020-02-27 14:53:59.138,User hint appears two times for data visualizations,"### Observed behaviour - -The user hint appears twice when opening data visualizations: - -![image](/uploads/73ac8a163fd6d6a65a111b1ed350b5bd/image.png) - - -### Expected behaviour - -It should open onle once. - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",1 -31297103,2020-02-26 15:57:00.515,ML project / settings / general / advanced - Delete Project,"### Problem to Solve - -As a user I want to be able to adapt the settings of my repository. - -**Note**: The ""settings"" tab in any directory is **only** visible for members of that repo. - -**Note**: These settings view is for the **ML project** repositories. - -### Technical Solution - -Note: The remove project opens the already implemented modal of deleting a project which currently is placed in the projects overview. We need to move the modal from the project overview to the advanced settings. - -![image](/uploads/6cb44a9ed95a46f067b68bba0e867827/image.png) - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Documentation: https://mlreef.gitlab.io/backend/develop/#_delete_code_projectsid - -**Relevant designs in XD file** - -* Slide 3 - -Link to XD: https://xd.adobe.com/view/65e88d76-2c29-4503-47d1-3148049ed484-13d9/ - -![Settings_General_Advanced](/uploads/76fefa60d0a8ff931b1b2eacf3a5f049/Settings_General_Advanced.png)",6 -31213593,2020-02-24 18:32:24.973,Research on git LFS,"We need to know how GIT LFS works, in order to see if this is the ideal solution for MLReef. - -Here are some questions that need to be answered: - -* [x] Can git LFS be used for data deduplication? -> Maybe -* [x] Are there major limitations? (sizes, performance, usage, etc) -> Not as far as I can see -* [x] What does the user (client) need to setup in order to work? -> Major git clients support it out of the box -* [x] What do we (server / gitlab / infrastructure) need to setup? -> Gitlab supports LFS - -Additional Notes -====================== -* Atlassian Keynote regarding LFS: https://www.youtube.com/watch?v=006RUPVIP-c -* Git LFS Client with AWS S3 backend: https://github.com/jasonwhite/rudolfs",1 -31212615,2020-02-24 17:53:08.820,ML Project / Settings / General / Naming,"### Problem to Solve - -As a User, I want to be able to change the _name_ and the _description_ settings of my repository. - -**Note**: The ""settings"" tab in any directory is **only** visible for members of that repo. - - -**Note**: These settings view is for the **ML project** repositories. - -### Technical Solution -* Project Name is saved in Backend & Gitlab -> Update via Backend -* Description is saved in Backend only -> Update via Backend -* Avatar -> Not in the scope of current milestone - -* Tags & Slug -> Not in the scope of the current milestone. - -https://mlreef.gitlab.io/backend/develop/#_put_data_projectsid -``` -PUT /data-projects/:id -Example request -$ curl 'http://localhost:8080/api/v1/data-projects/3b3e4811-fae3-46cd-8e07-84e3f94b3655' -i -X PUT \ - -H 'Content-Type: application/json' \ - -H 'Accept: application/json' \ - -d '{ - ""name"" : ""New Test project"", - ""description"" : ""description"" -}' -Response body -{ - ""id"" : ""3b3e4811-fae3-46cd-8e07-84e3f94b3655"", - ""slug"" : ""slug-1"", - ""url"" : ""www.url.com"", - ""owner_id"" : ""aaaa0000-0001-0000-0000-cccccccccccc"", - ""gitlab_group"" : ""mlreef"", - ""gitlab_project"" : ""New Test project"", - ""gitlab_id"" : 1, - ""experiments"" : [ ] -} -``` - - -Additional Notes -===================== - -**Relevant designs in XD file** - -* Slide 1 -* Slide 2 - -Link to XD: https://xd.adobe.com/view/65e88d76-2c29-4503-47d1-3148049ed484-13d9/ - -![Settings_Overview](/uploads/d66e912bc44a73e6895d2c8db1808ae8/Settings_Overview.png) - -![Settings_General_Naming](/uploads/98a1c8845211aa62763435e476aaa9a3/Settings_General_Naming.png)",8 -31200244,2020-02-24 12:00:38.479,Create experiment UUID,"### Problem to Solve - -Currently, in the experiment detail view, the experiment ID is the experiment name. We need to change this to be a true UUID, which is also stored in our backend DB (@flohin). - - - -### Technical Solution - -As we are currently working with branches (an experiment creates a branch where it stores the output files), we could use the UUID created by the Gitlab for branches - if Gitlab does that? - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== -> Give a more technical explanation of how you want the app to function. Include screenshots and links to the relevant documentation",1 -31200004,2020-02-24 11:52:28.139,Refine CSS on experiments overview,"### Observed behaviour - -The current experiment overview does not look like it should. - -![image](/uploads/f05a07cf69609c2989df7f9adee3e624/image.png) - -### Expected behaviour - -It should look like this, link to XD: https://xd.adobe.com/spec/7d7e3c44-50a2-4980-5c95-c30af600762d-0b4f/ - -### Steps to reproduce - -Current state of experiment overview. - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",2 -31199059,2020-02-24 11:22:00.826,Experiment values not showing up,"### Observed behaviour - -A successfull experiment does **not** show the output values as shown in the attached image. - -### Expected behaviour - - - -### Steps to reproduce - -Open dev environment, cats dogs project, experiment tab and expand the experiment ""super-sirene"" - -### Terminology -> Define terms that are important in the context of this task - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -![image](/uploads/1eaabaf0b62ce501130db6b995c95cac/image.png) - -relates to:",4 -31137785,2020-02-21 18:32:06.064,Forking in progress view,"### Problem to Solve - -We need an forking project view, as the forking progress might take some time. There are some ""specialities"" for this view: - -1. There is a gif ;) -2. The top area needs to be hidden away (the grey area contracts and only shows what is included in the screen), because: -3. We need to automatically refresh the page to see if the project has been completely forked. Gitlab refreshes every 2 seconds or so. - - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/b4d538d5-666e-40d6-614c-6a995806ee4d-c2e4/ - -![forking](/uploads/0a02eeaa043309e0ad74ab2da3e50513/forking.png) - -![Forking](/uploads/0243c38f03da263c37a8c2d28fdda8a1/Forking.png)",2 -31136869,2020-02-21 17:49:15.974,Empty experiments view,"### Problem to Solve - -We needed a nicer and more inviting view for empty experiments. This is it! - -Note: No filter buttons are present. The create a new experiment button now is centern and on the bottom. - -### Technical Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/e233e6ca-7ec7-4842-58ff-f724872b4612-9402/ - -![Experiments_empty](/uploads/d6da3b9c958c9dffa93bb3b41a0d92dc/Experiments_empty.png)",1 -31121247,2020-02-21 12:36:16.458,Cancel running experiment,"### Problem to Solve - -As a user I want to cancel a running experiment. - -Note: Cancelling a running experiment is only possible by owners of a project. The button ""cancel"" is not visible for not owner. - - -### Design -Use the liniar indeterminate progress bar from material ui as progress bar: https://material-ui.com/components/progress/ - -### Technical Solution -THe pipeline experiment can be canlled through Gitlab, but then backend is out-of-date. So after receiving the successful response then the backend has to be updated. - - - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/036b80f1-9769-4112-7315-a6ed5c55f165-755a/ - -![Abort_experiment](/uploads/f3f63086501e534cc32402ea056be56d/Abort_experiment.png)",4 -31075148,2020-02-20 12:26:26.011,Registration,"### Problem to Solve - -As a user I want to be able to register ;) - - -### Technical Solution - -In the dropdown, these are the options (data scientist, ml engineer, data analyst, team lead, product manager) - -* [ ] Add at least one unit test -* [ ] If useful: Add documentation to the readme - - -### Terminology -> Define terms that are important in the context of this task - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/c4069ef1-b6d4-44f5-528c-4e0da5800e5a-b237/ - -![Register_B](/uploads/9f43564552b9488923fb049e7015adda/Register_B.png) - -![Register_A](/uploads/3b39cbb39b685708734530145ccbf75a/Register_A.png) - -![Register_C](/uploads/266dae716e3f969d2a2801266803009c/Register_C.png) - -![Register_D___1](/uploads/f588ecff89a6a852b5793a7b33e0ed87/Register_D___1.png) - -![Register_D___2](/uploads/87f0e22ed19b7eebb7a6e25ae0c3064f/Register_D___2.png) - -![Register_D](/uploads/e5cf958fcc8aeba9af904c5170c7f4bc/Register_D.png)",4 -80948548,2021-03-15 16:22:21.050,Some resources like icons are not loading correctly,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behavior - -Some icons do not load correctly due to a protocol issue, the browser is apparently requesting them by HTTP in the prod environment. -![image](/uploads/b5102f350fe7d8af86a3588e9c3b1447/image.png) - -### Expected behavior -The FE knows the protocol in which the app is running, hence the resources are always fetched in this one and no error is written in the log. - -### Steps to reproduce -1. Login -2. Click on a project, check the dev options. - -### Solution -Add some logic/rule in the Nginx that reroute this insecure traffic to the right target.",5 -79119805,2021-02-15 15:08:47.662,Refine delete branch modal / bug,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour -When trying to delete several branches in one go, the second branch I want to delete is hindered by an inactive button in the modal view. - -Also, the modal needs UI designing (background blur and size of the modal related to the text). - -### Expected behaviour - -If I want to delete several branches in one go, this should be possible. - -### Steps to reproduce - - -### Workaround - - -### Solution - - - - -Additional Notes -===================== - -![image](/uploads/960695be22fd337c25ccd0bf4146e4b1/image.png)",2 -78909785,2021-02-11 13:16:08.842,E2Etest failing for one use case,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour - -During the k8s migration, multiple failures were noted in e2etest job. All have been fixed one by one **except the last 1 error**. It does not look like infra related, please debug it. - -![image](/uploads/fbf7468799878f4c48f0b28b631e21bc/image.png) -![image](/uploads/a62509d0c9218ca56afdd2e84f1acdfd/image.png) - - -### Expected behaviour - -It should run without any failure - -### Steps to reproduce - -Any pipeline execution - -### Workaround - - -### Solution - - - - -Additional Notes -=====================",2 -76559815,2020-12-29 15:42:11.652,Implement the UI for Import data overview,"### Problem to solve -The implementation of the ""Import data overview"" needs the UI first, so that in the future these wireframes can be wired up with the BE and the rest of services, as well as animated with some logic. - - -### Proposal for Technical Solution -Implement the view, link to access it, the cards and other HTML elements without any interaction with other services. - -### Documentation -Link to XD design: https://xd.adobe.com/view/71b56162-b41c-47f1-8747-f2a0858be11f-24e6/ - -### Links / references -https://gitlab.com/mlreef/mlreef/-/issues/601",3 -75967445,2020-12-11 16:07:09.069,Strage margins in the project creation view,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -As a user, I want to review the view for project creation. Currently, in this view, the margins are very narrow and use all the width of the screen. - -### Proposal for Technical Solution -add width or margins according to with the rest of the application. - - -### Documentation -![image](/uploads/eb1c5672ccb84ee13b3d6e1eb1c71ea1/image.png)",1 -75056910,2020-11-26 15:38:03.149,Problems fetching projects in the Marketplace,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behavior - -After enabling the code-project-creation buttons a new bug was caused, so the infinite scroll has to be activated for the tabs different of `ML Projects` because with the current implementation the user won't be able to see all his code projects. - -### Expected behavior - -Users can use the infinite scroll for the models, data processors, and visualization tabs. - -### Steps to reproduce - -1. Open the market place -2. Go to the code project tabs, see how many projects are fetched.",5 -74964200,2020-11-25 10:55:28.072,Publishing button needs permission wrapper,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour - -In prod environment, the publishing button is clickable on not-owned code projects. Although it does not do anything, other than redirect the user to an empty screen, this button should show the forking modal. - - -### Expected behaviour - -For any other user except the owner or maintainer of a code project, the publishing button should not be visible. This way we avoid any misunderstanding. - -### Steps to reproduce - -In prod navigate to any code project that is not yours. - -### Workaround - - -### Solution - - - - -Additional Notes -===================== - -![image](/uploads/8fbd1e17d63f3fd395769df07b7b8df7/image.png)",1 -74563951,2020-11-18 13:16:26.824,Adapt the FE to the newly created republish endpoint,"### Problem to solve -As a user I want to have the possibility to publish a project again. This capability is currently not working because an exception is thrown by the server. - -To solve this issue, the new endpoint republish has to be called when the project has more than one publication. - - -### Proposal for Technical Solution -The only way to know whether a project has been published or not, is by getting the pipelines and assert that array length is greater than 0. - - -### Permissions and Security -The permissions have to be the same as for publishing",5 -74534855,2020-11-18 01:35:25.837,Adjustments in the pipeline creation view,"### Problem to solve -In the pipelines creation view, we have some minor issues that should be solved: - -- [x] Review execute button, it's enabled only when the user clicks outside the processor card. -- [x] Add popups for cases when the execute button is disabled and the user clicks on the button. Those popups should contain instructions about the missing steps. -- [x] Remove the advance options chevron button when there are not advanced options/params. -- [x] Adapt the cards (the list on the right) to have the global marker, desing: https://xd.adobe.com/view/566ad5dc-edcf-49e3-7230-c4d0e1fa4078-205b/ -- [x] Update the global marker for the cards (border with color) in the market place",5 -74522908,2020-11-17 18:30:53.583,Enable create project button for different sorts of projects,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -In the market place, we currently have the button for creating projects disabled except for the data projects. - -Enable it as we now have the endpoints to create different sort of projects - -### Proposal for Technical Solution -Pass a parameter in the URL so the ""create project"" view knows whether the project is data or code project - - -Additional Notes -===================== -![image](/uploads/baa8e627e773f03c66025b9cfe3b6984/image.png) - -![image](/uploads/c5ecfc73d9061e43b3dde82a37420236/image.png) - -Make sure the endpoint for creating projects in the backend can distinguish among the code project subtypes",3 -74100794,2020-11-09 16:53:03.178,404 file not found while trying to view a dataset file,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour - -In staging environment, i successfully created several dataops pipelines. When accessing the datasets withing the dataset detail view, i am getting this error: - -![image](/uploads/1b978195f61cb38c7e77e50465cb2c56/image.png) - -### Expected behaviour - -It should show the file that i clicked. - -### Steps to reproduce - -with mlreef user go here: http://staging.mlreef.com/mlreef/catsndogs/-/blob/branch/data-pipeline%2Fcute-shrimp-9112020-1/path/output%2Fcats%2Fcat.0_re.jpg - -### Workaround - - -### Solution - - - - -Additional Notes -=====================",5 -74080487,2020-11-09 11:40:53.028,Refine SEO information on mlreef,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -When searching in google for mlreef (incognito mode), the following results are shown. Some information are missleading / not sufficient: - -- [ ] There is no meta information on mlreef.com (not about.mlreef.com). This is the first entry shown in the list and should be made more attractive. ALTERNATIVELY: Is there a way of pushing about.mlreef.com as the primary site to show in the list? -- [ ] The location of business of MLReef (as shown in the image) is from science park graz (we are incubdated by them). -- [ ] Review the possibility to include subpages in the google entry (as in image 2, 1st circle). -- [ ] What is the right info box and how can we create such a thing?! (as in image 2, 2nd circle). - - -### Proposal for Technical Solution - - - -### Permissions and Security - - -### Documentation - - -### Availability & Testing - - - -Additional Notes -===================== -### Is this a cross-stage feature? - - -### Links / references - -![image](/uploads/ae316236865c038caa6f26636466c88a/image.png) - -![image](/uploads/97f524a79b083141f09c277717267d7c/image.png)",5 -73536998,2020-10-28 18:44:32.417,bug frontend: Glitch in value list in data operation,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour - -Using the data operation with value list option we observe this glitch, we cannot see the values from the list -![image__2_](/uploads/94e752470c5840d2124e9676b088436c/image__2_.png) -Another issue is that you cannot change the value from the list after choosing. - -### Expected behaviour - -The list value should be seen clearly. - -### Steps to reproduce - -### Workaround - - -### Solution - - - - -Additional Notes -=====================",3 -73237518,2020-10-23 09:34:38.947,Bugs in data visualization overview and details view,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -**Data visualization overview** - -- [x] The stile differs from datasets & visualization overview. Visualizations has all tabs (in progress, active, expired) while datasets has only the one with an entry. How we are doing it with datasets is much better imo. -- [x] The button view pipeline in visualization has no function. It should have the same style as in the data visualization detail view. -- [x] The button in view pipeline from dataset overview redirects to the data ops pipeline, but it doesnt reconstruct the pipeline. This works when you enter the dataset detail view. Could you adapt this to work on all cases? -- [x] Remove the ""Use"" and ""Expires in"" fields from data visualizations. -- [x] The active icon in visualization is wrong (exclamation mark) compared to checks (see dataset). -- [x] Remove the pre-fix ""data-visualization/"" of the data visualization name. -- [x] The username is not clickable. -- [x] The delete function is not working (no modal to delete the data visualization, as in datasets). -- [x] We should rename the ""expired"" tab to ""failed"" and ""canceled"". We should rename the ""active"" to ""success"". In both data visualization and dataset view. -- [x] I noticed that the active data visualization in your repo ""well-megalodon"" has status failed but is in the ""active"" section. This should be then in the new ""failed"" section. - -**Data visualization detail view** - -- [x] The status should have color (e.g. failed in red, pending in orange, success in green, canceled in red). -- [x] @andres46 implemented in the experiment view (executive summary) that I could click the ""files selected from path"" (src) whick links to the commit hash view of that path and the branch name, linking to the commit hash view of that branch. He knows more about this but we should include this function in both visualizations and dataset view. Not sure if there is a separate ticket for this, though? -- [x] The data visualization uses (e.g. t-sne) should be clickable and link to the code repo (in a new tab). This for both data viz and dataset. -- [x] The abort button is always active, despite the datavisualization being finished. It should be replaced by an X (delete). This deletes the entry and the corresponding branch. - - -### Proposal for Technical Solution -The tasks mentioned above should be completed and make the pipelines views bug-free. - -### Availability & Testing -- [ ] Few unit tests for the detail and overview components",10 -73028256,2020-10-20 13:22:02.909,"WIP: Follow-up from ""Add routing to all breadcrumb components""","The following discussion from !891 should be addressed: - -- [ ] @diegovinie started a [discussion](https://gitlab.com/mlreef/mlreef/-/merge_requests/891#note_431000726): (+4 comments) - - > Things that are required, are required for good reasons. If those reasons don't change then this is not good.",8 -72775915,2020-10-15 13:37:39.170,Review branch dropdown: listed branches,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -We have several places in MLReef where we show the branch dropdown button with the list of branches in one repo. This ticket aims at 1. restructure the automated branch name for branches created in a pipeline and 2. show sorted branches list (without pre-fixes). -Extra: 3. If possible on the ""pipeline-branches"" show the status as in ticket: #725 - -Here the overview of where we have branch dropdowns and what to show in them (this changes!) - -- [x] Repo overview: Show all branches (as in the design) -- [x] Commit view: Show all branches -- [x] Repo history: Show all branches -- [x] New merge request: Show all branches -- [ ] Dataselector modal: Show normal branches and datasets, hide data visualizations and experiments - - -### Proposal for Technical Solution - - - -### Permissions and Security - - -### Documentation - -Link to XD design (slide 5): https://xd.adobe.com/view/c6c402fa-d7f8-4b46-4c6f-dd6257402fb4-b0f5/ - -![image](/uploads/ec45ed7f1db05dab9beb693e0ff73afc/image.png) - -### Availability & Testing - - - -Additional Notes -===================== -### Is this a cross-stage feature? - - -### Links / references -![image](/uploads/ba49f9c0ee60b211d28a60c5b1b1a808/image.png)",7 -72773117,2020-10-15 12:46:05.335,The jobs list seems to have a delay showing information,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour - -The insights tab shows information about the pipelines and their corresponding status. The information present seems to have a delay of sometimes a few minutes. - -- [ ] Pipeline started and no entry of the pipeline is shown in the jobs list -- [ ] Status (e.g. running, pending, failed) has ""old"" information, although pipeline already has a different value -- [ ] Pipeline status does sometimes not match from the job overview and job detail view - - -### Expected behaviour -Information should be up-to-date. - - -### Steps to reproduce - -Start any pipeline and review the status on the jobs list. - -### Workaround - - -### Solution - - - - -Additional Notes -=====================",5 -72724123,2020-10-14 16:08:20.033,Accessing dataset / visualizations files gives an undefined error,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour - -When trying to access a datafile / path in a dataset, a root error happens and therefore, the files are not viewable. - -- [x] Also check on data visualizations! - - -### Expected behaviour - -The user should be directed to the folder / file that he wants to see in the repo view with the dataset branch selected. - -### Steps to reproduce -In master, select a dataset and explore the view to find this error. - -### Workaround - - -### Solution - - - - -Additional Notes -===================== -![image](/uploads/73610168b58df14df83fa2cd9b64dd9f/image.png)",2 -72673569,2020-10-13 21:05:50.664,"Follow-up from ""Fix deleting files and add tests"" minor fixes in file deletion modal","The following discussion from !877 should be addressed: - -- [ ] @cpmlreef started a [discussion](https://gitlab.com/mlreef/mlreef/-/merge_requests/877#note_429094482): (+1 comment) - - > Hey @andres46 this works perfect now! Really great :) - > - > Just very minor improvements: - > - > 1. Integrate headers such as in the design so the users know what to do: - > See design: https://xd.adobe.com/view/b093afeb-118a-48f2-5b24-8684dbf5014c-c6a2/ - > ![image](/uploads/455e453a76d06f80ab355523eba0d36e/image.png) - > - > 2. When creating a new branch, clicking outside the dropdown does not close the dropdown: - > - > ![image](/uploads/010147e9643951f41a180584942e055b/image.png) - > - > 3. When creating a new branch for deleting and after pressing the delete button, the user should be redirected to this branch (now it returns to master) - > - > Question: are there tests for this? ;)",3 -72661656,2020-10-13 16:14:49.681,"Follow-up from ""Fix deleting files and add tests""","The following discussion from !877 should be addressed: - -- [ ] @SaathvikT started a [discussion](https://gitlab.com/mlreef/mlreef/-/merge_requests/877#note_428008208): (+2 comments) - - > @andres46 Can't this be a helper global function? I am sure we will need the delete file feature in other places as well, so maybe we should create this as a function in `helpers.js`.",3 -72630500,2020-10-13 09:47:52.279,DRY end2end tests,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve - -As a developer, I want it to be really easy to create additional end-to-end tests. For almost all tests it is necessary to be logged in and create at least one project. - -Therefore, it makes sense to have something like a TestDirector which covers the following use cases. - -- [ ] Creates user and logs in user if needed -- [ ] Creates a project which can then be used in the tests. - -Additionally (for extra credit :smile: ) it might make sense we -- [ ] let the TestDirector cache User and/or Project info so that we are creating as few new users as possible. - - -### Proposal for Technical Solution - - - - -Additional Notes -===================== -/cc @andres46",10 -72628036,2020-10-13 09:09:31.844,Assert publishing of Docker Images,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve - -Currently, we have [one end2end test which publishes a code ](https://gitlab.com/mlreef/mlreef/-/blob/master/web/src/__end2end-tests__/CodePublishing.test.jsx), **BUT** it never asserts that a Docker Image was actually put into the docker registry. - -The target of this ticket is to add ""an assertion"" to the test which queries Gitlab's API for information about the Docker registry and asserts that the newly created Image was uploaded. - -The process in the backend works as follows: -* backend receives publish request at `api/v1/projects/{id}/publish` - * creates Dockerfile in repository - * create `.mlreef.yml` in the repository -* Gitlab starts a pipeline based on the `.mlreef.yml` - - [ ] Test needs to assert that a pipeline was started - - [ ] Test needs to wait until the pipeline is finished -* The pipeline builds the docker image and pushes it to the built-in docker registry - - [ ] Test needs to assert that a new image is present - - -### Proposal for Technical Solution - - -Add the steps at the bottom of https://gitlab.com/mlreef/mlreef/-/blob/master/web/src/__end2end-tests__/CodePublishing.test.jsx - - -### Documentation -The test is itself documentation - - - -Additional Notes -===================== -For testing please refer to the ""Known Issues"" section in `web/bin/ci-run-end2end-tests` added in Merge Request !881 - -We already have some experience with code that uses the API to delete images from our docker registry on gitlab.com -``` -# 1117031 is the id of the ""gateway's"" sub-repository id -curl --request DELETE --header ""PRIVATE-TOKEN:$GITLAB_COM_API_TOKEN"" - https://gitlab.com/api/v4/projects/$CI_PROJECT_ID/registry/repositories/1117031/tags/$CI_COMMIT_REF_SLUG -```",10 -72458070,2020-10-09 14:50:51.517,Refactor weird code in the new merge request overview,"The following discussion from !868 should be addressed: - -- [ ] @SaathvikT started a [discussion](https://gitlab.com/mlreef/mlreef/-/merge_requests/868#note_426977170): (+1 comment) - - > @andres46 A new ticket maybe? - -### Documentation - -Open the merge-request-overview file. Look for the next annotation: - -`{/* TODO: refactor the next code, no need to copy 4 times the same */}`",3 -72199202,2020-10-05 15:14:05.736,Move end2end test code,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -Context: To make builds faster and more efficient, Gitlab allows adding conditionals to build scripts. So e.g. the frontend build and FE unit tests are only executed if frontend code has been changed. - -e.g: -``` -only: - changes: - - web/**/* -``` - -Currently, the end2end tests are located under `web/src/__end2end-tests__` and unfortunately, this means, whenever an end2end test is edited or added we have to wait for the frontend build (and deploy) to be finished. - -So the topic of this ticket is to move the end2end sources out of `web/src/` so that we can use a Gitlab ci build filter on the `web/src` folder. - - - -### Proposal for Technical Solution -Proposal: Refactor end2end test source to e.g. `web/test/__end2end-tests__` and `web/tests/__unit-tests__` - - - -Additional Notes -===================== -### Links / references -Frontend build takes about 18 Minutes - -![build](/uploads/4af416ba0e8d57cc77d8524600e8528f/build.png)",8 -72176176,2020-10-05 09:24:45.590,Change folder upload to file(s) upload,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour - -The current function of uploading folders has several issues and need to be adapted to files upload: - -- [x] Drag and dropping a file into the drop field results in an error (it does simply not work). -- [x] When opening the explorer view (by clicking on the drop field), the explorer want to find folders. We need to change this so the explorer searches for files to upload. Selecting multiple files (not only limted to one) should be possible. -- [x] There seems to be a cap in size to upload of 10MBs in total (sum of all files). This is a limit that is not realistic, as there are no datasets that are below 10MBs. We should have a limit for this of severl GBs (if we need a limit). -- [x] The root where the user navigated before in the tree view of the repository should be kept for uploading files. So if he moved to cats/train and presses the + button to upload files, the newly uploaded files should be placed there. -- [x] This root information should also be shown in the upload view (if design is required, please tell). -- [ ] The drag-drop function removes previous files, review that - - - -### Expected behaviour - - - -### Steps to reproduce -In staging, try the upload folder function. - -### Workaround - - -### Solution - - - - -Additional Notes -=====================",13 -72070683,2020-10-02 10:19:53.950,Only a limited number of projects are shown as a visitor,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour - -As a visitor I can only see a very limited amount of public ML projects when accessing mlreef.com/explore. - -Furthermore, when logged in, my own projects now do not appear under ""my projects"" but only under the explore tab. - - -### Expected behaviour - -I should see all public projects. - -### Steps to reproduce - -Go to https://mlreef.com/explore - -### Workaround - - -### Solution - - - - -Additional Notes -===================== - -What one can see: - -![image](/uploads/c5ef8339b1916c72e84482d06e4799fa/image.png) - -What one should see: - -![image](/uploads/199b7a2825d873a16384d31e46040945/image.png)",11 -72034720,2020-10-01 18:14:25.821,Use constants for permission levels,"The following discussion from !843 should be addressed: - -- [ ] @andres46 started a [discussion](https://gitlab.com/mlreef/mlreef/-/merge_requests/843#note_422219857): (+1 comment) - - > Hey @diegovinie - > - > what about using constants for the access levels? - > - > like developer = 30 - > visitor = 10 - > - > Just to have an idea of what these numbers represent.",3 -71873920,2020-09-29 13:35:07.609,Adjustments and minor bugs to review in the pipelines process,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behavior - -- [x] In the ticket https://gitlab.com/mlreef/mlreef/-/issues/683 some changes were requested for the ""Experiments Overview"". -These changes have to be replicated wherever the data sets and pipelines information is presented. -- [x] In the dataset details, the breadcrumb does not work. -- [x] Make the abort button work for the data set details view. The button is shown sometimes when it makes no sense because the pipeline was already canceled or finished by itsel - the abort button is shown in the detail view all the time. It should be replaced by an ""X"" if the job is somehow complete and the entry can be deleted. -- [x] Data visualizations overview and data visualization details view have to look like data sets and data sets details view -- [ ] Counter of data visualizations in the repo overview and number of data visualization entries in data visualizations overview do not match -- [x] hide datavisualization prefix ""data-visualization/"" - so show only the name -- [ ] Change the format of the time stamp for all pipelines :) currently -15102020-1 to **- 15/10 14:30:01** -- [x] Can we include the weight of the branch in the ""use: "" section? Does gitlab provide this info? If not, lets delete this sections. -- [x] The view pipeline button has no function - -![image](/uploads/5482fb0485c4592cbe051d3c7733dfdb/image.png) - -![image](/uploads/c10561ee1dba1e00e882ee6261e53a45/image.png)",5 -71868287,2020-09-29 12:00:54.041,Create pipeline output entry as soon as pipeline started,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behavior -As a user I want to have inmediate feedback as soon as I created any pipeline and I executed it (by pressing the ""proceed to results"" button in the modal view""). Today, and most of the times, a toaster appears with e.g. ""experiment created"" but no entry is found in the experiment overview. This is the same for all other pipelines. - -Then, after a while a new toaster appears with pipeline was created and then, the entry is available. - -This time difference results in confusion and bad UX. - - -### Expected behaviour - -The entry should be created immediately. The pipeline can start afterwards if the machines are not available yet. - -### Steps to reproduce - -In staging, execute any pipeline and jump directly to the results. No entry is there until, if it goes fast, a few seconds but sometimes it takes minutes! - -### Solution -In order for the backend to create a pipeline in Gitlab, some steps have to be performed -* [PipelineServiceL209](https://gitlab.com/mlreef/mlreef/-/blob/master/backend/mlreef-rest/src/main/kotlin/com/mlreef/rest/feature/pipeline/PipelineService.kt#L209) and following - -One step is to look for or create the EPFBot user. -> Searching is done by select all users from Gitlab and iterating through them. --> `AuthService.CreateOrFindGitlabUser()` --> AuthServiceL263. - -`AuthService.CreateOrFindGitlabUser()` is the first place to be changed to a more scalable approach. - -We believe, optimising this function will solve the performance problem. - -First approach to try out -> https://docs.gitlab.com/ee/api/search.html#scope-users - - - -* [ ] Add at least one test - - - -Additional Notes -=====================",6 -71868006,2020-09-29 11:55:43.196,Projectview throws exceptions when requesting project info,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour - -When accessing data in a finalized dataset, the following error appears and no data is shown. - -### Expected behaviour - -The user should be able to navigate the data tree and discover the files in the file view. - -### Steps to reproduce -Use this link and navigate into a folder or file in the data view at the bottom of the dataset detail view: -http://staging.mlreef.com/MLReef_1/flower-classifier/-/datasets/06b72641-3c32-4b24-abc1-0991cabd1d0c - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== -![image](/uploads/07d916063422ad64bded08ff66de6196/image.png) - -* http://ec2-18-156-26-49.eu-central-1.compute.amazonaws.com:10080/MLReef_1/flower-classifier/commits/data-pipeline/profound-shark-29092020-1 -* http://staging.mlreef.com/my-projects/fcb1f674-0351-4ee8-b08a-a3ffa096ed9e/data-pipeline%2Fprofound-shark-29092020-1/commits",3 -71867457,2020-09-29 11:45:48.095,Deleting files does not work properly,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour -As a user I want to delete files in my repository. This function seems broken as it does not do anything. Furthermore, the style is still old (no greyed out and blurried background effect). - - -### Expected behaviour -The file should be deleted and the user redirected to the repo view. - - -### Steps to reproduce - -In staging, explore an owned project and then navigate into a file. Try to delete the file. - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -=====================",5 -71818645,2020-09-28 17:16:30.745,Adapt frontend for authenticated visitors,"1. [x] Hide execute button - -![image](/uploads/ac1524cd43aa79bbb4f6d962e294e9d3/image.png) - -2. [x] Hide *New Experiment* and *delete* buttons - -![image](/uploads/61c38acbae6e7f9fa331a3b86c0b38da/image.png) - - -3. [x] Hide **+** button - -![image](/uploads/b4ec7829c7c11dbba81773309a7fb229/image.png) - -4. [x] Hide *abort/delete* button - -![image](/uploads/7170f4d4096a377eb089c7f04f6ec2a9/image.png) - - -5. [x] Hide *delete* button - -![image](/uploads/d814dfd90cc70efa80b03ed148e56843/image.png) - -6. [x] Hide *delete* button - -![image](/uploads/e14ce2daa72697d778f9911ff5c6ca13/image.png) - -7. [ ] Hide *delete* button - -![image](/uploads/5bc7046a106813592e8cb592fc908466/image.png) - -8. [x] Fix branch button -9. [x] Hide *new branch* and *delete* buttons - -![image](/uploads/cb23792e12073ade24346fde2eb08312/image.png) - -10. [x] Hide *New Merge Request* button (merge requests view) -11. [x] Hide action buttons (edit, close) in merge request view -12. [x] Hide merge button in merge request view",8 -71803557,2020-09-28 13:56:00.628,File size changes not shown in merge request view,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour -All changed files have a NaNKb size added... - - -### Expected behaviour - - - -### Steps to reproduce - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== -![image](/uploads/e9b9d8bfb61158042eaef2b2f6a29a42/image.png)",2 -71553829,2020-09-23 08:27:34.113,Adapt pipeline views to include global marker and input icons,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -As a user I want to be reminded of the connex between pipeline types (e.g. data visualization) and the data operators that are available in this pipeline type. - -- [ ] Therefore, the corresponding global marker makes sense in these views as well. -- [x] In addition, the information about input data type from the data operators makes complete sense. -- [x] In addition, include the star count from the corresponding code repos in the card list (as in the desing) - -### Intended users - - -### User experience goal - - -### Proposal for Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -### Permissions and Security - - -### Documentation - - -### Availability & Testing - - -### What does success look like, and how can we measure that? - - - -Additional Notes -===================== -### What is the type of buyer? - - -### Is this a cross-stage feature? - - -### Links / references - -![image](/uploads/3ea27aa25d6ef656c121e75a47af07ea/image.png) - -Link to adapted design (example for experiment pipeline, but the global marker needs to be set in all three with their corresponding color): https://xd.adobe.com/view/8d73060c-174a-43e6-6a96-78861d008514-fb42/",4 -71530611,2020-09-22 19:53:15.591,Fix REST docs: /projects/:id/users/:userId,"We have some duplicated endpoints, which offer the same features, but with a different syntax: - -POST /projects/:id/users/:userId -POST /projects/:id/users?userId.. -POST /projects/:id/users with objects - -There is no need for that, we should just decide for an endpoint, design a pattern, and work consistently in that area. - -Decide for an architectural style and just establish it. - -## Todo - -* Fix Rest tests and create snippets -* or refactor endpoints and delete obsolete rest docs",2 -71530483,2020-09-22 19:47:56.609,Fix REST docs: projects-retrieve-public,"## Todo - -* Fix Rest tests and create snippets -* or refactor endpoints and delete obsolete rest docs - -GET /projects/public: operation::projects-retrieve-public[snippets='curl-request']",1 -71506063,2020-09-22 13:21:05.596,"Add meta description to project view, sitemap, robot.txt and rel=canonical","# -# Please select the most appropriate ticket template from the dropdown above -# -![image](/uploads/f7adaad66643a8bd13f06ff1a8d1f908/image.png)",8 -71322608,2020-09-17 16:49:14.477,A small graphical bug in settings / members,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Observed behaviour - -In a project / settings / members. When I click on a project member the dropdown to change his role (no dropdown is present at the moment), a frame with scroll opens, which should not happen... - -### Expected behaviour - - - -### Steps to reproduce - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== -![image](/uploads/3b6f35c851312dbc07b07c0e0a10a51b/image.png)",2 -71255987,2020-09-16 15:05:03.264,Visual errors when accessing code repos from a pipeline,"### Observed behaviour - -I am in a pipeline (does not matter which) and click on a card to jump to the code repository: - -- [x] the green marker (ml projects) is applied to the code repository (instead of the code repo specific color) -- [x] the tab ""data"" is there instead of ""code"" -- [x] Had an issue with the data in the repo as well. In the code repository, the data from the data repo (e.g. training-data) was still shown. - -After refreshing the page the errors are gone and the code repo is shown properly. - -### Expected behavior -The global marker from the code repository should be applied. - - -### Steps to reproduce - -In the prod environment, open a pipeline, expand on any card and click ""view repository"" - -### Terminology -> Define terms that are important in the context of this task - - -### Solution -For the solution, we can show the spinner while the project is loading or showing a small spinner for the project view comp, discuss this with @cpmlreef before the ticket's implementation. - - - -Additional Notes -===================== -This is a data ops repo, colors should be Violette: -![image](/uploads/8b9b30aaabd16c96780f2c25c0e62203/image.png) - -relates to:",5 -71252767,2020-09-16 14:16:27.712,Job Log throws Error when job is still pending,"### Observed behaviour -Sometimes it can happen that Jobs in Gitlab are pending for a long time. If for example there are no runners currently available. - -In this case, the job log will throw an error (see screenshot) - - -### Expected behaviour -Since an empty log is expected for a pending job, some sort of icon or text which lets the user know that everything is ok - - -### Steps to reproduce -* Create a new Experiment pipeline -* Go to the job log fast enough so that you see a pending job - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - - - - -Additional Notes -===================== -![image](/uploads/2759eabd66db7a81c16a6722693f5af4/image.png)",2 -71129248,2020-09-14 17:48:08.524,Implement modal for creating folders (directory),"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -As a user, I want to have a modal like Gitlab's one, in which the user can create a folder into a specific route. The route is selected, by navigating in the repository overview in the specific folder and then selecting the + sign and the menu option ""New directory"". - -### Intended users -All types of users with writing access. - -### User experience goal - -Ease the content creation in the repository. - -### What does success look like, and how can we measure that? - -### Design - -Link to XD: https://xd.adobe.com/view/38fd22ad-6ab0-443b-afd9-89ba7cebcf1a-6dae/ - -![Create_new_folder](/uploads/a21100d273c1bae935a27cd0319c101d/Create_new_folder.png)",6 -71120762,2020-09-14 15:11:30.290,Backend: Refactor FileLocation Enum Value PATH to FILE and FOLDER," -instead of ""Path"" we need -* PATH (deprecated) -* PATH_FILE -* PATH_FOLDER - - -relates: #683",2 -70999742,2020-09-10 21:14:01.382,Implement first stage of the publishing wizard,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve - -Implement the user flow of the publishing wizard, the wireframes are ready and HTML elements will be done soon. - -Use these elements to create the UI from the ""Select entry point view"" to the interlude. This will be the first stage, it encompasses to provide all the files and data necessary to publish the code project and persist it in the database, call the API's and show the interlude view. - -In the next link, the prototype can be found. For this issue *only* the slides 1-4 are required. - -### Proposal for Technical Solution -These are the technical requirements for this stage: - -* Select entry point: - -1. Publish one branch at the time which will be the master branch (protected branch from gitlab) as a starting phase. The commit hash is NECESSARY as well. -1. The entry point is the file or script to run in python. -1. The purpose of the publication process is to publish a data operator. - -note: Requirement.txt should be placed in the root, and we try to find it there. Should not be picked by the user, validated in this stage. - -* Select base environment: - -1. Can only select one environment, only some that are very useful will be as options for the moment. -1. We need from backend the list of environments available, and an endpoint to send the environment selected(to be called after the publish button is pressed) - -* Publish model: - -1. Clarify the model type, whether more options should be shown or not. -1. An endpoint to list the model-types -1. An endpoint to list ml categories -1. The model type and ML categories selected should be a field in the backend and included in the publish endpoint -1. Boolean to persist whether the user accepted the terms of publication, manage versions(probably a timestamp). - - -### Links / references - -Link to XD design: https://xd.adobe.com/view/2154ca1c-fbf7-4ba8-bac4-d5edc077d0ac-7407/",15 -70969486,2020-09-10 09:31:57.752,CodeProjectPublishing API and Backend," -### Problem to solve - -The frontend developers will develop the publishing wizard with the mockups/wireframes. - -During that development, we might find out the logical point when the frontend wants to send data to the backend (start the wizard) and what information the frontend sends and expects from the backend. - -Due to missing planning, we dont know that yet in the backend. Therefore we want to try the upstream design approach: - -**-> Frontend tells Backend how the endpoint should be like, and backend follows that proposal.** - -### Proposal for Technical Solution - -we might need several endpoints: - -* starting the publishing (backend checks project and request for validity) - > the first iteration is very basic still. When An existing ""unpublished"" Data Repo is published, the following things will happen - - [x] A `Dockerfile` is added to the project's Git Repo - - Does the Dockerfile need to be customized? - - [x] A `.mlreef.yml` is added to the Git Repo - - does the `.mlreef.yml` has to be customized? - - push the docker image as `:latest` version to the docker registry - - [x] Gitlab triggers a pipeline based on the `.mlreef.yml` for every commit on the ""master"" branch - - [x] this pipeline creates a new version of the Docker image - - [x] pipeline is tested in end2end tests? --> based on !806 - - [x] the new docker image version is published to Gitabl's docker registry - - [x] parse the python file -> #780 - - [x] persist the DataProcessorVersion from the Python file to the database on the first build - - [x] update/replace the DataProcessorVersion on future builds (remember every commit to master triggers a new build -* [x] fetch info or state of tasks/pipelines? -> go directly to GitLab -* [x] commit/cancel publishing? -> frontend can stop and cancel pipelines -* [x] stop publishing -> frontend offers to delete the `.mlreef.yml` -* [x] Publishing Porcess needs to store ./#{#MAIN_SCRIPT}"" -->CORRECTLY <-- in database -* [x] Create an E2E TEST which tests a custom data operation -> [template e2e test](https://gitlab.com/-/ide/project/mlreef/mlreef/tree/master/-/web/src/__end2end-tests__/DataPipelines.test.jsx/); [target file](https://gitlab.com/-/ide/project/mlreef/mlreef/tree/master/-/web/src/__end2end-tests__/CodePublishing.test.jsx/) -* [x] Delete CMD from [Dockerfile template](https://gitlab.com/-/ide/project/mlreef/mlreef/tree/master/-/backend/mlreef-rest/src/main/resources/code-publishing-dockerfile-template/) - -If everything goes as planned, end2end testing can be implemented in NodeJs with help from the FE team :) (see !806)",2 -70918289,2020-09-09 09:24:46.621,Creating a group does not work,"### Observed behaviour - -In dev (and all other environments) creating a group does not work - probably because an image was selected? - -### Expected behaviour - -The group should be built. - -### Steps to reproduce - -Create a group in dev with all fields filled up (in this case also as a public group) - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - -![image](/uploads/045df4166e2a9b59a410e05239f252ca/image.png) -relates to:",4 -70848654,2020-09-08 01:20:44.691,Show custom title for project view,"The title for a project should be: - -`MLReef | {project's name}` - -![image](/uploads/957b8d518f68c8efb0d7a8ecf2967035/image.png) - -## Notes: -Consider to apply *react-helmet* or similar - -related to: #737",3 -70684743,2020-09-03 12:54:50.357,Backend: Cleanup project endpoints and fix visibilities,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve - - -**IMPORTANT: endpoint meanings:** - -* own: I created -* my : my own and shared with me (member of project or group) -* public: everything public, can include own or my if they are public -* ""/"" all: my+public = every project I can use or attend -* starred: every project under ""all"" which has my star - -Endpoints were mixed up, - -- add systemTests for OWN, SHARED, ALL and PUBLIC Projects -- add fixes to controllers -- add some more JPA and service methods -- add new unpaged endpoint",2 -70642520,2020-09-02 16:00:52.555,Create UI component for Publishing Process,"This are the list of required UI components for publish process' design. - -Weight between bracets. - -+ page 1 -- [x] Tab Steps component [2] -- [x] Statement component [1] -- [x] Folder view component (from similar) [3] -+ page 2 -- [x] Brick grid component (from similar) [2] -- [ ] Card component [1] -- [x] Filters component (from similar) [3] -+ page 3 -- [x] Vertical Steps component [2] -- [x] Colored radio-button [2] -- [ ] Checkbox input (from similar) [1] -- [x] Vertical Tabs adaptation (few changes) [0] -+ page 5 -- [x] Pipeline [5] -+ page 6 -- [ ] Job display from (similar) [1] -+ page 7, 8, 9 -- [x] Parameters display (from similar) [1] - -Related to: - -## Resources: -https://xd.adobe.com/view/4dc536f4-bea0-453c-8826-406ae2f8312d-1896/screen/16412dee-042e-49e3-b502-d8719ab89f78/ - -## Notes: -- UI components don't perform API request nor Redux actions, therefore this issue doesn't encompass the whole publishing process rather the look and feel.",24 -70625094,2020-09-02 10:09:36.004,Include lables for projects view (marketplace),"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -As a user I want to see if: - -a. A project is public of private - -b. My role in this project (e.g. maintainer, owner, developer...) - -c. If a code repository has been published. - - -### Intended users - - -### User experience goal - - -### Proposal for Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -### Permissions and Security - - -### Documentation -The design can be taken from the groups-projects view: https://xd.adobe.com/view/4cec0111-f89b-4825-6363-58d95e361fef-3add/ - - -### Availability & Testing - - -### What does success look like, and how can we measure that? - - - -Additional Notes -===================== -### What is the type of buyer? - - -### Is this a cross-stage feature? - - -### Links / references",5 -70624241,2020-09-02 09:49:08.470,Upload file - 413 error message needs adaptation,"### Observed behaviour - -When uploading a too big file via the file uploader, the error message from Gitlab is: - -Failed to load resource: the server responded with a status of 413 (Request Entity Too Large) - -The frontend (toaster) shows: ""File could not be uploaded"". - -### Expected behaviour - -The frontend (toaster) error should show: ""File(s) could not be uploaded. File limit of XX Mb exceeded."" - -- What is the current file limit?? - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",4 -70367065,2020-08-27 10:08:52.027,Error in Data instance Details view,"### Observed behaviour -![Screenshot__68_](/uploads/6bbe327fdfbfcb15a2e810b4739c2255/Screenshot__68_.png) - - -### Expected behaviour -The user can access data instance detail view without any bugs. - - -### Steps to reproduce -1. Go to the data instance details view -2. Copy the url. -3. Open a new tab -4. Paste the link in the new tab and the UI breaks. - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",8 -70331986,2020-08-26 16:53:41.891,Show status of datasets in dataselector branch dropdown,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -As a user I want to see the status of a dataset that might be still running in the data selector, branch drobdown. - -For example, I create a new dataops pipeline where the branch and the dataset are inmediately created. I should not be able to select this dataset for any pipeline, as the pipeline is not finished. The best approach would be to show the data set in the data selector dropdown, but it is not selectable and it has a view of the pipeline status. - - -### Intended users - - -### User experience goal - - -### Proposal for Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -### Permissions and Security - - -### Documentation -- Red means: Pipeline failed or was cancelled. -- Orange means: Pipelines is pending or running. -- Green means: Pipeline means success and dataset is ready to be used. This is the only one that is selectable. - - -### Availability & Testing - - -### What does success look like, and how can we measure that? - - - -Additional Notes -===================== -### What is the type of buyer? - - -### Is this a cross-stage feature? - - -### Links / references -XD Link: https://xd.adobe.com/view/f059d209-7ad7-4789-7f96-20b99b464f11-8668/ - -![image](/uploads/adb41a5197352e41ca370248e1133b0b/image.png)",7 -70331823,2020-08-26 16:46:17.966,"In the jobs view, the filter button ""finished"" does not show finished jobs","### Observed behaviour - -In the jobs view (insights tab), the filter button ""finished"" does not show any jobs that are finished. - -Finished are jobs that are either -- success -- failed -- cancelled - -### Expected behaviour - - - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",3 -70331274,2020-08-26 16:29:14.161,Add systemTest for Publishing Docker Images to Gitlab/MLReef infrastructure,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -The most basic part of publishing user-generated Data Processors within MLReef is building and pushing -Docker Images to MLReef's Gitlab based Docker Registry. - -It might be necessary to activate or configure the Docker Registry -> https://docs.gitlab.com/ee/administration/packages/container_registry.html#enable-the-container-registry - -There exists already a [test project on gitlab.com](https://gitlab.com/mlreef/devops/develop-data-processor) which can be cloned and then pushed to the relevant branch environment or other infrastructure. - -### Todo - -* call the API codeProject -* checkout codeProject -* add some mlreef.yml manually to the project -* push it -* test it? - * does a pipeline run? - * does it succeed? - * API call to check docker registry if something is there? - -### User experience goal -MLReef Developers can manually define a Dockerfile and mlreef.yml file to build and push docker images to the MLReef Docker registry. - - -### Proposal for Technical Solution -The smallest and simplest possible Dockerfile as a reference to `alpine:latest` without further modifications -```Dockerfile -FROM alpine:latest -``` - -The corresponding `.mlreef.yml` would look something like this: -```yaml -build: - image: docker:19.03.5 - services: - - docker:dind - script: | - echo ""$CI_REGISTRY_PASSWORD"" | docker login ""$CI_REGISTRY"" --username=""$CI_REGISTRY_USER"" --password-stdin - docker build --tag ""$CI_REGISTRY_IMAGE:$CI_COMMIT_REF_SLUG"" -f Dockerfile . - docker push ""$CI_REGISTRY_IMAGE:$CI_COMMIT_REF_SLUG"" -``` - -For more references of how to build docker images with Gitlab's CI, you can look here -* https://gitlab.com/mlreef/mlreef/-/blob/develop/web/.gitlab-ci.yml -* https://gitlab.com/systemkern/s5/-/blob/master/.gitlab-ci.yml - - -### Permissions and Security - - -### Documentation -The implementation of this ticket will yield much clarity about how a future MLReef Dockerfile template has to look like in order to be useful and usable - - -### Availability & Testing -This is a core feature and testing for this has to be completely automated. While writing the testing code, keep in mind that we will support on-premise installations. - - -### What does success look like, and how can we measure that? -MLReef Developers can build and push docker images to the MLReef Docker registry. - - - -Additional Notes -===================== -Keep in mind that, for on-premise installations, we will need to be able to provide our base images preinstalled with the installation.",2 -69974680,2020-08-18 18:27:23.203,Call the backend star project function,"### Problem to Solve -From the frontend, call the backend endpoint to star a project when the star button is pressed. -![image](/uploads/303a73d9e9dc7254becf4ae4f487b7cb/image.png) - -### Technical Solution -Change url to address the call to the another API - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -=====================",3 -69924900,2020-08-17 17:32:44.367,Backend: REST is docs incomplete,"### Observed behaviour - -https://mlreef.gitlab.io/mlreef/rest-api/feature/fix-system-start/index.html -Half of rest-docs is missing: - -``` -GET /projects -Example request -Snippet curl-request not found for operation::genericprojects-retrieve-all - -Response body -Snippet response-body not found for operation::genericprojects-retrieve-all - -Response fields -Snippet response-fields not found for operation::genericprojects-retrieve-all - -GET /projects/public -Example request -Snippet curl-request not found for operation::genericprojects-retrieve-public - -Response body -Snippet response-body not found for operation::genericprojects-retrieve-public - -Response fields -Snippet response-fields not found for operation::genericprojects-retrieve-public -``` - - -### Solution - -* [ ] Fix or remove those lines, some of those ""projects"", ""generic projects"", etc, are duplicates maybe",2 -69909834,2020-08-17 14:20:14.574,Fatal: Backend is not starting in docker,"### Observed behaviour - - - -### Expected behaviour - - - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",6 -69906611,2020-08-17 13:21:32.022,The joblog views (e.g. experiments / insights) need to refresh automatically,"### Problem to Solve -The job log view should automatically refresh - at least as long as it is in ""running"" or ""pending"" stage. This makes it easier for the user to follow the progress. - - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -=====================",5 -69896496,2020-08-17 08:54:57.687,Re-run a pipeline the same way it was initially executed,"### Problem to Solve -As a user I want to simply re-run a past experiment, dataops pipeline or data visualization. This is done by clicking on a ""view pipeline"" button in the resulting experiment / dataset / datavisualization which brings me back to the pipeline view, with **all** parameters set as they initially was. This includes: - -- Data path from data selector selected and pointing to the state of the data as it was (commit history) -- All data operators are opened with the hyperparameter values set as they were -- The data operators' script needs to be the same as it originally was. This might be more tricky, as we somehow would need to store the version of the script used. In relation to our publishing process and the docker images, this might be a challenge... - - -### Technical Solution -Q: How do we solve the script versioning in relation to the publishing process? - -**Detailed implementation: ** - -The commit id should be received in the view in order to rebuild the same project files structure when the data operation was executed. This involves only the file selector, the Gitlab API request should send the commit id as the ref, and then the user will see everything as it was then. - -![image](/uploads/b28361b4c3166878249b933fc345452c/image.png) - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -=====================",6 -69748546,2020-08-12 06:36:29.841,Link data versions in pipelines,"### Problem to Solve -As a user I want to be able to easily track what version of files I used for any of our pipelines. For example, in the experiment executive summary view this is done by clicking on the data links (see additional notes). - -### Technical Solution -Use GIT (commit history) to track the data used: - -MLReef Pipeline -> Gitlab Pipeline -> Commit Hash - -For example: -""master"" should link to ""185b6788dd948bc9c93be5ee01081dd3640eec30 on master"" - -IDK if our frontend can handle this already but in Gitlab it would bring you to the state of the repo at this exact commit. - -https://gitlab.com/mlreef/mlreef/-/tree/185b6788dd948bc9c93be5ee01081dd3640eec30 - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== -![image](/uploads/107adbc1af17fd8483a7ac0d070996e8/image.png)",4 -69577082,2020-08-08 17:44:43.089,Add sorting options to the market place view,"### Problem to Solve -In the marketplace view (visitor) show to tabs: - -* ""all"" -> sort by most recent commit -* ""popular"" -> sort by most stars (4) (FE ticket pending) - -![image__4_](/uploads/c10c14ce578792ec85d01c123ee5152e/image__4_.png) - -### Technical Solution - -* This information for example ""most recent commit"" is not currently available in the projects response payload. So for this first case a good solution would be to show all projects without sorting. - -* The popular sorting option is actually doable. - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== - -Link to XD: https://xd.adobe.com/view/5d9dcfd2-3bfb-4ee0-5069-8c9e2f0a5b7b-2bac/",4 -69576584,2020-08-08 17:18:06.618,Backend: Add user update endpoint and role information,"### Problem to Solve -As a user I want to modify some information for the user: - -- [x] 1. Store in backend that the user has accepted terms during registration(**true** or **false**). -- [x] 2. Store in backend and send to mailchimp that user want newsletters(**true** or **false**). -- [x] 3. Store in backend user type after registration, these are the next: -- [x] 4. Delete pre-Fix of users ""mlreef-users-""username"" in Gitlab - -![image](/uploads/7e729ebb6f939331db48b87dda0a9604/image.png)",3 -69416044,2020-08-04 15:30:54.882,"Add ""Starring"" feature for Projects","### Problem to Solve -As a user I want to rate projects by giving them stars (similar how github / gitlab) does it. By doing so, i can then filter / order projects by stars in the marketplace view or in the pipeline views (e.g. experiment pipeline). - -Note: in addition to addings stars the user can also remove a previously given star to a project. - - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -=====================",5 -69396825,2020-08-04 07:28:19.589,Backend: Protect data from inconsistency by creating constraints in DB,"### Problem to Solve - -As we have many methods like ""findByGitlabId"" or ""findBySlug"", we have soft business requirements that some entity fields are unique or ""kind of"" unique. There was an error, when gitlabId was used twice and everything broke down due to our security mechanisms. - -Therefore we could do two things: -* replace those soft ""unique""s with lists -* harden and use constraints - -We have very weak constraints relations in Postgres. Sometimes we got inconsistency state in db that requires manual fix by developers. - - -### Technical Solution -1) Create constraints in database -2) Check the code and possible add business logic that rejects inconsistent data before it has gone to database -3) if some ""soft constraints"" are not hard enough to be ""unique"", those methods must change to support lists! - - -### Terminology -Constraint - programming entity in databases that links tables among each other and does not allow to create records that do not satisfy this linkage",4 -69385088,2020-08-03 21:12:15.378,Backend: The Forbidden error,"### Observed behaviour - -Since a couple of weeks I witnessed that some endpoints were throwing a 403 error. - -At the beginning the only one was the experiments creation endpoint - -![image](/uploads/3a3a23c1a47a5b2e6b79875bc7cf4efb/image.png) - -but in the last days I noticed that more of them are throwing this error. - -![image](/uploads/30394912eaddc32ad74b1b8f4d6a8031/image.png) - -I even tested with Postman to discard the *FE* and got the same: - -![image](/uploads/308a28d63572aaccc67528bc62b7764a/image.png) - -My conclusion is that a global security problem was introduced in the code in the last weeks. So this ticket seems completely necessary. - -### Expected behaviour - -Endpoints work for authenticated fronend users - -### Steps to reproduce - -1. Login -2. Click on Models - -or - -Elaborate a Postman request",5 -69349132,2020-08-03 06:18:43.581,403 Error for Experiment finish,"### Observed behaviour -When finishing an Experiment the EPF wants to call the `api/experiments/:id/finish` endpoint and receives an `HTTP 403` error. - -This can be seen in [this end2end test](https://gitlab.com/mlreef/mlreef/-/jobs/667229461), which created [this project](http://SMOKE-TEST-iaa:password@ec2-18-194-183-93.eu-central-1.compute.amazonaws.com:10080/SMOKE-TEST-iaa/test-project) and [this pipeline](http://ec2-18-194-183-93.eu-central-1.compute.amazonaws.com:10080/SMOKE-TEST-iaa/test-project/-/jobs/16) - - -``` -Running after script... - $ epf-finish - EPF_PIPELINE_SECRET is d8288fba7a8a2f222ed0c9795c306707 - EPF_PIPELINE_URL is ec2-18-194-183-93.eu-central-1.compute.amazonaws.com/api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6 - Finalize EPF execution and send finish information to backend - ec2-18-194-183-93.eu-central-1.compute.amazonaws.com/api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6/finish - % Total % Received % Xferd Average Speed Time Time Time Current - Dload Upload Total Spent Left Speed - 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0* Trying 172.31.15.168... - * TCP_NODELAY set - * Connected to ec2-18-194-183-93.eu-central-1.compute.amazonaws.com (172.31.15.168) port 80 (#0) - > PUT /api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6/finish HTTP/1.1 - > Host: ec2-18-194-183-93.eu-central-1.compute.amazonaws.com - > User-Agent: curl/7.58.0 - > Accept: */* - > EPF-BOT-TOKEN: d8288fba7a8a2f222ed0c9795c306707 - > Content-Type: application/json - > - < HTTP/1.1 403 - < Server: nginx/1.19.1 - < Date: Sun, 02 Aug 2020 12:14:53 GMT - < Content-Length: 0 - < Connection: keep-alive - < Access-Control-Allow-Credentials: true - < Access-Control-Allow-Methods: * - < Access-Control-Max-Age: 3600 - < Access-Control-Allow-Headers: * - < - 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0 - * Connection #0 to host ec2-18-194-183-93.eu-central-1.compute.amazonaws.com left intact - HTTP/1.1 403 - Server: nginx/1.19.1 - Date: Sun, 02 Aug 2020 12:14:53 GMT - Content-Length: 0 - Connection: keep-alive - Access-Control-Allow-Credentials: true - Access-Control-Allow-Methods: * - Access-Control-Max-Age: 3600 - Access-Control-All -``` - - -### Expected behaviour -Well ... no error ;-) - - -### Steps to reproduce -* Create and push new branch and for the pipeline to finish -* Go to the end2end job and look at the log: - * Look for the text ""Project URL:"" usually around 1201 and follow the link -* Look at the project pipelines - - -### Solution -> Add technical implementation details and the results of the ticket's discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== -complete job log: - -``` - Running with gitlab-runner 13.2.2 (a998cacd) - on Multi runner dispatcher on ec2-18-194-183-93.eu-central-1.compute.amazonaws.com -FyArkXY -Preparing the ""docker+machine"" executor -06:02 - Using Docker executor with image registry.gitlab.com/mlreef/mlreef/epf:feature-555-ensure-local-testing ... - Pulling docker image gitlab/gitlab-runner-helper:x86_64-a998cacd ... - Using docker image sha256:af98a1efed2728f2ded0b1b4141a75d1fbc4a2598b8ff655044f2019574374cb for gitlab/gitlab-runner-helper:x86_64-a998cacd ... - Pulling docker image registry.gitlab.com/mlreef/mlreef/epf:feature-555-ensure-local-testing ... - Using docker image sha256:2438f99238ed607f28f14de0c5cdc4a9c68dac46bf497aa98d4a928eaae4e0b6 for registry.gitlab.com/mlreef/mlreef/epf:feature-555-ensure-local-testing ... -Preparing environment -01:50 - Running on runner--fyarkxy-project-30-concurrent-0 via runner-fyarkxy-mlreef-aiops-1596369941-2fdef18a... -Getting source from Git repository -00:01 - Fetching changes with git depth set to 50... - Initialized empty Git repository in /builds/SMOKE-TEST-iaa/test-project/.git/ - Created fresh repository. - Checking out 186f39f7 as experiment/happy-jellyfish_2962020... - Skipping Git submodules setup -Executing ""step_script"" stage of the job script -01:18 - $ cat .mlreef.yml - ################################################################################ - # This is the MLReef configuration file. # - # It contains the current configuration of your model-experiment (training) # - # as well as the temporary configurations executed data pipelines # - ################################################################################ - # This is the docker image your model training will be executed in - image: registry.gitlab.com/mlreef/mlreef/epf:feature-555-ensure-local-testing - variables: - # Change pip's cache directory to be inside the project directory since we can only cache local items. - PIP_CACHE_DIR: ""$CI_PROJECT_DIR/.cache/pip"" - EPF_GITLAB_HOST: ""ec2-18-194-183-93.eu-central-1.compute.amazonaws.com"" - # The EPF Pipeline URL is the complete URL to the EPF endpoint for this particular pipeline - # mlreef.com/api/v1/epf/experiments/:experiment_id - EPF_PIPELINE_URL: ""ec2-18-194-183-93.eu-central-1.compute.amazonaws.com/api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6"" - EPF_PIPELINE_SECRET: ""d8288fba7a8a2f222ed0c9795c306707"" - INPUT_FILE_LIST: ""PATH:data/"" - TARGET_BRANCH: ""experiment/happy-jellyfish_2962020"" - # The before_script handles everything git related and sets up the automatic committing - before_script: - # Debug output mlreef.yml - - cat .mlreef.yml - - echo ""####################"" - #- it just aint working! - - git remote set-url origin http://${GIT_PUSH_USER}:password@""$EPF_GITLAB_HOST"":10080/SMOKE-TEST-iaa/test-project.git - - git config --global user.email ""SMOKE-TEST-iaa@example.com"" - - git config --global user.name ""SMOKE-TEST-iaa"" - # TODO this can maybe be deleted - #- git checkout master - # TODO this can maybe be deleted - #- export GITLAB_API_TOKEN=""${GIT_PUSH_TOKEN}"" - - echo ""Target branch is $TARGET_BRANCH"" - - background-push & - after_script: - - epf-finish - data-pipeline: - script: - - git checkout -b $TARGET_BRANCH - - echo ${CI_JOB_ID} >> data_pipeline.info - - python /epf/model/resnet50.py --input-path data --output-path . --epochs 2 --height 36 --width 36 - - git add . - - git status - - git commit -m ""Add pipeline results [skip ci]"" - - git push --set-upstream origin $TARGET_BRANCH - - git push$ echo ""####################"" - #################### - $ git remote set-url origin http://${GIT_PUSH_USER}:password@""$EPF_GITLAB_HOST"":10080/SMOKE-TEST-iaa/test-project.git - $ git config --global user.email ""SMOKE-TEST-iaa@example.com"" - $ git config --global user.name ""SMOKE-TEST-iaa"" - $ echo ""Target branch is $TARGET_BRANCH"" - Target branch is experiment/happy-jellyfish_2962020 - $ background-push & - $ git checkout -b $TARGET_BRANCH - EPF_PIPELINE_SECRET is d8288fba7a8a2f222ed0c9795c306707 - EPF_PIPELINE_URL is ec2-18-194-183-93.eu-central-1.compute.amazonaws.com/api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6 - ------------------------------------------------------------ - Background Commit Iteration 0 - experiment.json does not exist yet, waiting for experiment to start.. - Switched to a new branch 'experiment/happy-jellyfish_2962020' - $ echo ${CI_JOB_ID} >> data_pipeline.info - $ python /epf/model/resnet50.py --input-path data --output-path . --epochs 2 --height 36 --width 36 - 2020-08-02 12:13:38.142932: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libnvinfer.so.6 - 2020-08-02 12:13:38.144635: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libnvinfer_plugin.so.6 - 2020-08-02 12:13:39.229355: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1 - 2020-08-02 12:13:39.253162: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero - 2020-08-02 12:13:39.254006: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1555] Found device 0 with properties: - pciBusID: 0000:00:1e.0 name: Tesla K80 computeCapability: 3.7 - coreClock: 0.8235GHz coreCount: 13 deviceMemorySize: 11.17GiB deviceMemoryBandwidth: 223.96GiB/s - 2020-08-02 12:13:39.254059: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1 - 2020-08-02 12:13:39.254123: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10 - 2020-08-02 12:13:39.256069: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10 - 2020-08-02 12:13:39.256441: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10 - 2020-08-02 12:13:39.258616: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10 - 2020-08-02 12:13:39.259850: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10 - 2020-08-02 12:13:39.259906: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7 - 2020-08-02 12:13:39.260042: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero - 2020-08-02 12:13:39.260927: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero - 2020-08-02 12:13:39.261704: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1697] Adding visible gpu devices: 0 - 2020-08-02 12:13:39.262074: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA - 2020-08-02 12:13:39.284876: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2300090000 Hz - 2020-08-02 12:13:39.285249: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x58f9e90 initialized for platform Host (this does not guarantee that XLA will be used). Devices: - 2020-08-02 12:13:39.285279: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): Host, Default Version - 2020-08-02 12:13:39.900102: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero - 2020-08-02 12:13:39.901013: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x596fd70 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices: - 2020-08-02 12:13:39.901042: I tensorflow/compiler/xla/service/service.cc:176] StreamExecutor device (0): Tesla K80, Compute Capability 3.7 - 2020-08-02 12:13:39.901295: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero - 2020-08-02 12:13:39.902054: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1555] Found device 0 with properties: - pciBusID: 0000:00:1e.0 name: Tesla K80 computeCapability: 3.7 - coreClock: 0.8235GHz coreCount: 13 deviceMemorySize: 11.17GiB deviceMemoryBandwidth: 223.96GiB/s - 2020-08-02 12:13:39.902117: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1 - 2020-08-02 12:13:39.902151: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10 - 2020-08-02 12:13:39.902179: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10 - 2020-08-02 12:13:39.902199: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10 - 2020-08-02 12:13:39.902213: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10 - 2020-08-02 12:13:39.902234: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10 - 2020-08-02 12:13:39.902256: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7 - 2020-08-02 12:13:39.902353: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero - 2020-08-02 12:13:39.903148: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero - 2020-08-02 12:13:39.903881: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1697] Adding visible gpu devices: 0 - 2020-08-02 12:13:39.903931: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.1 - 2020-08-02 12:13:40.141917: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1096] Device interconnect StreamExecutor with strength 1 edge matrix: - 2020-08-02 12:13:40.141965: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1102] 0 - 2020-08-02 12:13:40.141976: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] 0: N - 2020-08-02 12:13:40.142224: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero - 2020-08-02 12:13:40.143052: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:981] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero - 2020-08-02 12:13:40.143822: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1241] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 10769 MB memory) -> physical GPU (device: 0, name: Tesla K80, pci bus id: 0000:00:1e.0, compute capability: 3.7) - ------------------------------------------------------------ - Background Commit Iteration 1 - experiment.json does not exist yet, waiting for experiment to start.. - ------------------------------------------------------------ - Background Commit Iteration 2 - experiment.json does not exist yet, waiting for experiment to start.. - ------------------------------------------------------------ - Background Commit Iteration 3 - iteration: 3 - Background push to URL: ec2-18-194-183-93.eu-central-1.compute.amazonaws.com/api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6/update - % Total % Received % Xferd Average Speed Time Time Time Current - Dload Upload Total Spent Left Speed - 100 2 0 0 100 2 0 58 --:--:-- --:--:-- --:--:-- 58 - HTTP/1.1 403 - Server: nginx/1.19.1 - Date: Sun, 02 Aug 2020 12:14:06 GMT - Content-Length: 0 - Connection: keep-alive - Access-Control-Allow-Credentials: true - Access-Control-Allow-Methods: * - Access-Control-Max-Age: 3600 - Access-Control-Allow-Headers: * - {}------------------------------------------------------------ - Background Commit Iteration 4 - iteration: 4 - Background push to URL: ec2-18-194-183-93.eu-central-1.compute.amazonaws.com/api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6/update - % Total % Received % Xferd Average Speed Time Time Time Current - Dload Upload Total Spent Left Speed - 100 2 0 0 100 2 0 142 --:--:-- --:--:-- --:--:-- 142 - HTTP/1.1 403 - Server: nginx/1.19.1 - Date: Sun, 02 Aug 2020 12:14:21 GMT - Content-Length: 0 - Connection: keep-alive - Access-Control-Allow-Credentials: true - Access-Control-Allow-Methods: * - Access-Control-Max-Age: 3600 - Access-Control-Allow-Headers: * - {}2020-08-02 12:14:23.365843: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10 - 2020-08-02 12:14:23.581907: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7 - Found 150 images belonging to 2 classes. - Found 50 images belonging to 2 classes. - {'category_1': 0, 'category_2': 1} - 2 - Epoch 1/2 - 1/4 [======>.......................] - ETA: 1:02 - loss: 1.3077 - accuracy: 0.5938Error encountered: float() argument must be a string or a number, not 'NoneType' - 2/4 [==============>...............] - ETA: 21s - loss: 1.7901 - accuracy: 0.5156 Error encountered: float() argument must be a string or a number, not 'NoneType' - 3/4 [=====================>........] - ETA: 7s - loss: 1.6938 - accuracy: 0.5521 Error encountered: float() argument must be a string or a number, not 'NoneType' - Error encountered: float() argument must be a string or a number, not 'NoneType' - 4/4 [==============================] - 23s 6s/step - loss: 2.3449 - accuracy: 0.5391 - val_loss: 130.8565 - val_accuracy: 0.5312 - Epoch 2/2 - 1/4 [======>.......................] - ETA: 3s - loss: 2.3619 - accuracy: 0.4545Error encountered: float() argument must be a string or a number, not 'NoneType' - 2/4 [==============>...............] - ETA: 1s - loss: 1.7885 - accuracy: 0.5227Error encountered: float() argument must be a string or a number, not 'NoneType' - 3/4 [=====================>........] - ETA: 0s - loss: 1.8629 - accuracy: 0.5132Error encountered: float() argument must be a string or a number, not 'NoneType' - Error encountered: float() argument must be a string or a number, not 'NoneType' - 4/4 [==============================] - 2s 576ms/step - loss: 1.7177 - accuracy: 0.5000 - val_loss: 15.0612 - val_accuracy: 0.5556 - Using TensorFlow backend. - /usr/local/lib/python3.6/dist-packages/keras_applications/resnet50.py:265: UserWarning: The output shape of `ResNet50(include_top=False)` has been changed since Keras 2.2.0. - warnings.warn('The output shape of `ResNet50(include_top=False)` ' - /usr/local/lib/python3.6/dist-packages/keras/callbacks/callbacks.py:1042: RuntimeWarning: Reduce LR on plateau conditioned on metric `val_acc` which is not available. Available metrics are: val_loss,val_accuracy,loss,accuracy,lr - (self.monitor, ','.join(list(logs.keys()))), RuntimeWarning - $ git add . - ------------------------------------------------------------ - Background Commit Iteration 5 - iteration: 5 - Background push to URL: ec2-18-194-183-93.eu-central-1.compute.amazonaws.com/api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6/update - % Total % Received % Xferd Average Speed Time Time Time Current - Dload Upload Total Spent Left Speed - 100 193 0 0 100 193 0 12866 --:--:-- --:--:-- --:--:-- 12866 - HTTP/1.1 403 - Server: nginx/1.19.1 - Date: Sun, 02 Aug 2020 12:14:36 GMT - Content-Length: 0 - Connection: keep-alive - Access-Control-Allow-Credentials: true - Access-Control-Allow-Methods: * - Access-Control-Max-Age: 3600 - Access-Control-Allow-Headers: * - {""0"": {""acc"": null, ""val_acc"": null, ""loss"": 2.3448652923107147, ""val_loss"": 130.85653686523438}, ""1"": {""acc"": null, ""val_acc"": null, ""loss"": 1.7075497375594244, ""val_loss"": 15.06124496459961}}$ git status - On branch experiment/happy-jellyfish_2962020 - Changes to be committed: - (use ""git reset HEAD ..."" to unstage) - new file: data_pipeline.info - new file: experiment.json - new file: model_Resnet50_2020-08-02-12:14:30_epochs_2.h5 - new file: training.log - $ git commit -m ""Add pipeline results [skip ci]"" - [experiment/happy-jellyfish_2962020 1ec61d7] Add pipeline results [skip ci] - 4 files changed, 5 insertions(+) - create mode 100644 data_pipeline.info - create mode 100644 experiment.json - create mode 100644 model_Resnet50_2020-08-02-12:14:30_epochs_2.h5 - create mode 100644 training.log - $ git push --set-upstream origin $TARGET_BRANCH - ------------------------------------------------------------ - Background Commit Iteration 6 - iteration: 6 - Background push to URL: ec2-18-194-183-93.eu-central-1.compute.amazonaws.com/api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6/update - % Total % Received % Xferd Average Speed Time Time Time Current - Dload Upload Total Spent Left Speed - 100 193 0 0 100 193 0 12866 --:--:-- --:--:-- --:--:-- 12866 - HTTP/1.1 403 - Server: nginx/1.19.1 - Date: Sun, 02 Aug 2020 12:14:52 GMT - Content-Length: 0 - Connection: keep-alive - Access-Control-Allow-Credentials: true - Access-Control-Allow-Methods: * - Access-Control-Max-Age: 3600 - Access-Control-Allow-Headers: * - {""0"": {""acc"": null, ""val_acc"": null, ""loss"": 2.3448652923107147, ""val_loss"": 130.85653686523438}, ""1"": {""acc"": null, ""val_acc"": null, ""loss"": 1.7075497375594244, ""val_loss"": 15.06124496459961}}remote: - remote: To create a merge request for experiment/happy-jellyfish_2962020, visit: - remote: http://ec2-18-194-183-93.eu-central-1.compute.amazonaws.com:10080/SMOKE-TEST-iaa/test-project/merge_requests/new?merge_request%5Bsource_branch%5D=experiment%2Fhappy-jellyfish_2962020 - remote: - To http://ec2-18-194-183-93.eu-central-1.compute.amazonaws.com:10080/SMOKE-TEST-iaa/test-project.git - 186f39f..1ec61d7 experiment/happy-jellyfish_2962020 -> experiment/happy-jellyfish_2962020 - Branch 'experiment/happy-jellyfish_2962020' set up to track remote branch 'experiment/happy-jellyfish_2962020' from 'origin'. - $ git push - Everything up-to-date -Running after_script -00:01 - Running after script... - $ epf-finish - EPF_PIPELINE_SECRET is d8288fba7a8a2f222ed0c9795c306707 - EPF_PIPELINE_URL is ec2-18-194-183-93.eu-central-1.compute.amazonaws.com/api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6 - Finalize EPF execution and send finish information to backend - ec2-18-194-183-93.eu-central-1.compute.amazonaws.com/api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6/finish - % Total % Received % Xferd Average Speed Time Time Time Current - Dload Upload Total Spent Left Speed - 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0* Trying 172.31.15.168... - * TCP_NODELAY set - * Connected to ec2-18-194-183-93.eu-central-1.compute.amazonaws.com (172.31.15.168) port 80 (#0) - > PUT /api/v1/epf/experiments/a13154a4-b981-4509-b6ea-b5fe94bd30f6/finish HTTP/1.1 - > Host: ec2-18-194-183-93.eu-central-1.compute.amazonaws.com - > User-Agent: curl/7.58.0 - > Accept: */* - > EPF-BOT-TOKEN: d8288fba7a8a2f222ed0c9795c306707 - > Content-Type: application/json - > - < HTTP/1.1 403 - < Server: nginx/1.19.1 - < Date: Sun, 02 Aug 2020 12:14:53 GMT - < Content-Length: 0 - < Connection: keep-alive - < Access-Control-Allow-Credentials: true - < Access-Control-Allow-Methods: * - < Access-Control-Max-Age: 3600 - < Access-Control-Allow-Headers: * - < - 0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0 - * Connection #0 to host ec2-18-194-183-93.eu-central-1.compute.amazonaws.com left intact - HTTP/1.1 403 - Server: nginx/1.19.1 - Date: Sun, 02 Aug 2020 12:14:53 GMT - Content-Length: 0 - Connection: keep-alive - Access-Control-Allow-Credentials: true - Access-Control-Allow-Methods: * - Access-Control-Max-Age: 3600 - Access-Control-All -```",8 -69339749,2020-08-02 17:26:14.313,End2end Testing on all developer machines,"During the implementation of 555 it became clear that backend and to an extend EPF developers need to be able to run automated round trip tests on their developer machines. - -Currently we do not have a really good integration testing concept. Though not the main focus of this ticket, it makes sense to think about a sustainable test concept as well. - -Some of the needed tests are: -* start a model training -* upload training statistics to the Backend -* upload trained model results -* start data pipeline and verify results -* cancel experiment - - -## Additional Notes -There currently exist some bash and curl based tests which could be extended and adapted, but dont seem to run on each machine. -There exists also an existing integration test concept, which spins up a gitlab machine and has a clean, reproducible state to run tests. Integration tests can be run on each machine which could start the server, as they rely on docker and kotlin. There are already many integration tests for various reasons which tests different strategies and can be debugged, this concept seems better in terms of maintainability, system architecture and matches the rest of the technology. - -On Linux and OSX these existing curl-bash-scripts tests already work pretty ok but @si-ge-st had problems running them on Windows.",15 -69329543,2020-08-01 20:44:37.981,"FE: Set the value of the parameter ""input-path"" with the path from Data selector","### Problem to Solve -“As a user i will choose the data input using the Data Selector, this path should be the value for the hyperparameter input-path” - -Set the value of the hyperparameter called ""input-path"" with the Path from Data Selector. This hyperparameter can be overwritten by simply writing another input path over the ""standard"" value from the data selector. - -For this, we need to: -- [ ] Limit the number of selections a user can do in the data selector to 1 (either a folder or file) -- [ ] Redisign the data operators cards to include the automatic input-path from the data selector - -### Technical Solution -Probably, the input-path field needs some automatic updating. For example, if I start by dragging data operators in the pipeline and then select data from the data selector, the input-path field needs to be automatically updated. - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== - -Changed data selector for selecting only one folder or file - XD: https://xd.adobe.com/view/cd08a4b8-ecab-4508-4e6a-a00bb3967697-bb39/ - -Changed input-path input field: https://xd.adobe.com/view/566ad5dc-edcf-49e3-7230-c4d0e1fa4078-205b/",5 -69175488,2020-07-29 14:02:13.168,Budgeting Service,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve - - -### Intended users - - -### User experience goal - - -### Proposal for Technical Solution - - - -### Permissions and Security - - - -### Documentation - - -### Availability, Testing & Test Cases - - -### What does success look like, and how can we measure that? - - - -Additional Notes -===================== -### What is the type of buyer? - - -### Is this a cross-stage feature? - - -### Links / references - -/cc @rainerkern",10 -69174702,2020-07-29 13:43:53.883,Billing Backend,"[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -### Problem to solve -The backend should track the following points for user / group billing: -* Tracking of tier account of the user (free, tier1=""pro"") -* Tracking of time of upgrade of an account and expiration date (for auto-renew or auto-downgrade) -* Projects the user upgraded with ""accelerated inference"" package for boosting model show inference times -* Number of projects he has boosted (not sure yet, but possibly the ""accelerated inference"" package will allow the user to boost more than one project (e.g. 3 or so). -* Keep payment information - * payment type and information? e.g. credit card credentials? // security?? - * transaction history -* Create email notifications for accepted upgrades, automatic renew, auto-downgrade. - - - -### Intended users - - -### User experience goal - - -### Proposal for Technical Solution - - - -### Permissions and Security - - - -### Documentation - - -### Availability, Testing & Test Cases - - -### What does success look like, and how can we measure that? - - - -Additional Notes -===================== -### What is the type of buyer? - - -### Is this a cross-stage feature? - - -### Links / references",3 -69150478,2020-07-29 03:35:52.515,Review mobile responsiveness,"### Observed behaviour -The MLReef app is currently not optimized for mobile views. We should optimize this on it1 for the most relevant screens: - -- [ ] Marketplace -- [ ] Repo overview -- [ ] Pipeline creation views -- [ ] Experiment overview - - -### Expected behaviour - - - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== -![3E68AB8F-FC74-45DD-865A-99400B610B32](/uploads/15d7862a03cb70392b12945c326569ac/3E68AB8F-FC74-45DD-865A-99400B610B32.png) - -relates to:",40 -68948833,2020-07-24 13:32:23.431,Publishin process - specification,"### Problem to Solve -As discussed with @flohin we need to continue the final specification for the publishing process. I, @cpmlreef, would suggest to do this in 2 weeks time (next sprint) - - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -=====================",5 -68915450,2020-07-23 15:41:16.280,Backend: Rename some data operators,"### Observed behaviour - -As user I have witnessed that some names of projects can be misleading. - -The patten applied make them look like a slug. - -![image](/uploads/7e5586da7dcb3df819f1bb0667099c7c/image.png) - - -### Expected behaviour - -Rename these autogenerated projects and further registers so that the name is more user friendly - -### Steps to reproduce - -1. Go to http://ec2-3-127-170-253.eu-central-1.compute.amazonaws.com/#personal -2. Login -3. Read names - - -### Terminology -Project names, slug",1 -68106633,2020-07-20 14:04:42.528,Prioritize Cypress integration tests,"### Problem to Solve -As develpers, we need to create cypress integration tests for the higher priority functionality within the user interface workflow. It is plausible to perform this operation in the form of iterations/steps. In the current scope of this ticket, we can focus on e2e tests for experiments and pipelines workflow. - - -### Technical Solution -Research and understand the e2e testing process, propose a mechanism and technology we need to launch this process. So that we can automate the process of UI testing. - - -### Terminology -Cypress, Integration tests - - - -Additional Notes -=====================",16 -68070454,2020-07-20 12:25:46.932,ML : Model for Demand forecasting,"### Problem to Solve -> “As a user i can train a demand forecasting model following the instructions in the documentation provided by MLReef.” - - -### Subtasks -- [x] Find a dataset for demand forecasting that is widely used in ML platforms -- [x] Find a model that can be trained with this dataset and ""port"" to MLReef. -- [x] Add a README file for this dataset and model, it should contain an abstract and instructions to run the experiment. - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== - -https://www.kaggle.com/tandonarpit6/store-item-demand-forecasting-challenge-fbprophet - -https://github.com/LevinJ/Supply-demand-forecasting - -https://github.com/antoniopaisfernandes/Demand-Forecasting - -https://www.kaggle.com/c/demand-forecasting-kernels-only",5 -68067865,2020-07-20 12:20:15.412,"ML : Model for Sentiment analysis contains dataset , model , documentation","### Problem to Solve -> Describe the desired state of the system in the style of: -> “As a user i can train a sentiment analysis model following the instructions in the documentation provided by MLReef.” - - -### Subtasks -- [x] Find a dataset for sentiment analysis that is widely used in ML platforms -- [x] Find a model that can be trained with this dataset and ""port"" to MLReef. -- [x] Add a README file for this dataset and model, it should contain an abstract and instructions to run the experiment. - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== -https://github.com/vonsovsky/bert-sentiment",5 -67967023,2020-07-20 08:16:24.727,Migrate Backend Documentation Publish to Main Repo,"### Problem to Solve -Currently, in the main MLReef repo the backend rest docs are not published to Gitlab pages. - - -### Technical Solution - -**Option 1** - -The Gitlab pages for the main MLReef repo are linked docu.mlreef.com so the backend docu has to be rendered in a subfolder. - -This was already implemented in the Backend repo and the backend implementation can be used as template - -https://gitlab.com/mlreef/backend/-/blob/develop/.gitlab-ci.yml#L63 - -**Option 2** -Publish Restdocs on our own NGINX Gateway - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== -* `backend/bin/ci-build-backend` -* `backend/bin/ci-unit-test-backend` (Rest Docs generation) -* `backend/.gitlab-ci.yml` -* `backend/Dockerfile` -* `docs/.gitlab-ci.yml` (Pages Publishing) docs.mlreef.com",5 -61453269,2020-07-16 14:55:39.118,Backend: Invent data-populator with automagic data migration,"### Problem to Solve - -* either parse a json thingy, or have super nice kotlin -* maybe even a small kotlin dsl? -* refactor DataPopulator to support merge/update",8 -53976219,2020-07-06 14:51:20.787,Bug on pipeline creation (docker deamon),"### Observed behaviour - - - -### Expected behaviour - -We are recieving error messages on all pipelines: http://ec2-3-127-170-253.eu-central-1.compute.amazonaws.com:10080/mlreef/cats-and-dogs/-/jobs/4 - -### Steps to reproduce - -dev environment and run any pipeline. - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - -![image](/uploads/852ba4ddfefb6cb5d18888af992eee09/image.png) - -Additional Notes -===================== - - -relates to:",18 -53621859,2020-07-02 09:17:56.065,Join Backend to Monorepo,,13 -53621821,2020-07-02 09:17:11.436,Join EPF to Monorepo,,13 -53608947,2020-07-02 08:29:35.032,Sustainably fix URL workaround,"### Problem to Solve -In order to fix a cross-cutting connection problem between EPF and Backend @rainerkern introduced a workaround script which would change environment variables in this EPF MR [epf!58](https://gitlab.com/mlreef/epf/-/merge_requests/58/diffs) - - -### Technical Solution -- [ ] use the actual CI SERVER URL in the `.mlreef.yml` in the backend -- [ ] remove the whole _environment_ script or at least the then deprecated parts of it from the EPF -- [ ] activate or delete this code line (https://gitlab.com/mlreef/backend/-/merge_requests/89/diffs#b191ad777ddb8beb02d92383f761d6ea89669730_59_54) -- [ ] activate or delete this code line -(https://gitlab.com/mlreef/backend/-/merge_requests/89/diffs#b191ad777ddb8beb02d92383f761d6ea89669730_43_39) - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -=====================",20 -53049367,2020-07-01 14:01:29.352,Pipelines fail from wrong external URL,"### Observed behaviour -The pipelines fail because the EXTERNAL_URL is corrupted and replaced with a random ID - -![fail](/uploads/721835d4d10b8c7dbd8b38523f3a6e2a/fail.PNG) - -### Expected behaviour -Pipelines work successfully - - -### Steps to reproduce -1. Create a data processing pipeline (ideally with Augment operation) -2. Check the job log for that particular pipeline. - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",8 -52310017,2020-06-30 15:15:04.904,Add support for tags when creating a project,"### Problem to Solve - -As a user, I want to add tags to the project that I will create. This will help to better describe the project and filter/index them in the future. - -* FE should receive the list of tags in the Enum -* The project creation payload should contain the Tags field(array) -* The user should be able to adapt the tags later in the settings menu - -Related to: backend!88 - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. -1. Fetch tag list from mlreef backend, [doc here](https://mlreef.gitlab.io/backend/feature/598-add-projects-endpoints/#_get_exploretags). -2. Create an auto-complete input using the fetched tags. -3. Include the in the POST following the [API doc](https://mlreef.gitlab.io/backend/feature/598-add-projects-endpoints/#_post_data_projects) - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== -Following up https://gitlab.com/mlreef/backend/-/merge_requests/88#note_370855061",5 -51781540,2020-06-29 22:49:59.638,Refactor data operation views,"### Problem to Solve -> We currently have 3 views that behave very similarly; new experiment, new pipeline and new data visualization. - -### Technical Solution -> Since we can reuse more code than we currently do, create an only file which replaces just the data model depending on the view requested. This will lead to a more maintainable component instead of 3 files that share 90% of code. - - -### Terminology -Data pipelines, data visualizations, experiments",5 -50128734,2020-06-26 13:55:45.306,Get experiment status from Gitlab,"### Problem to Solve -We can currently cancel a running experiment through Giltab pipelines API, but since Gitlab and BE are not synchronized the Backend experiment keeps the status as RUNNING. -This situation causes confusion and frustrations to the user and developer because we show misleading info in the frontend. - -### Technical Solution - -*Some Possible solutions:* -- Implement an update experiment endpoint so that we can sync both systems - - builds on the existing cancel experiment endpoint - - still need for handing failed pipelines; since it is possible that they do not report their status - -- Get the experiment Status from Gitlab - - Either: the Client does this (backend does not contain this information at all) - - Or: Experiment Status is updated frequently **from** Gitlab (maybe even on every Request) - - handling failed pipelines would be easier",3 -49956617,2020-06-23 13:52:27.144,Adapt the view data pipeline button,"### Problem to Solve - -When migrating to the backend the Mlreef file was deprecated, so the ""View pipeline"" button has to be adapted. -The button should adress to the ""new pipeline view"". The data operations and data input that the user selected for that operation has to be shown. - -![image](/uploads/6e2586ae1c382a2c68b7a70eb609cfdd/image.png) - -### Technical Solution - -Get from the backend endpoint the information that was previously parsed from the mlref file.",5 -49901621,2020-06-22 11:18:10.478,Complete view for User_Account / Settings / Profile,"### Problem to Solve -As a user, I want to be able to change/update the username, avatar. - -Gitlab requires administrator access to update such details, so it has to be implemented first in the backend - -### Technical Solution -Please use backend for this ticket's functionality. - -### Additional Notes - -Link to XD: https://xd.adobe.com/view/811be39f-34d5-47ea-6bcc-95409cf5f74e-f72a/ - -relates to: #512 - -**Already implemented**: -Status: Gitlab API - -**TODO**: -Avatar, Username: Backend API - -![Settings_Profile](/uploads/2183264d484e50e92d43b876da3bcabc/Settings_Profile.png) -=====================",5 -49519946,2020-06-19 17:01:45.967,Backend: Allow groups to be added to project,"### Problem to Solve -+ As a User that owns a project I should be able to add groups to my project. - -+ Currently is only allowed to add users. - -+ The ideal situation is **to invite** groups instead of just add them. - - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. -Same treatment as users. - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - -related: #389 - - -Additional Notes -=====================",4 -49424929,2020-06-19 15:23:07.607,"Follow-up from ""Create user settings page"" - Refactor frontend routing","The following discussion from !512 should be addressed: - -- [ ] @diegovinie started a [discussion](https://gitlab.com/mlreef/mlreef/-/merge_requests/512#note_364184452): (+1 comment) - - > Hey @SaathvikT - > - > I really like this pattern for the routes, do you think we could implement it in the project routes? - > - > cc: @andres46 - -A better pattern for routes handling was found by Saathvik. It makes sense to apply it and refactor the router component in order to improve the readability of that code. - -To apply this pattern in the routes file, an array of objects would probably be better. The object needs a field to indicate that the route is private or not. -``` -{""projectName"": { ""isPrivate"": true, pattern: 'my-projects/project-slug/some-sort-of-name/param/:some-param' }} -``` - -So when we want to use it in a component, we can just replace :some-param in the code and use the link from a centrilized source.",8 -48097907,2020-06-18 15:28:04.415,Backend: Project owner's uuid is not the same owner uuid,"### Observed behaviour -Project owner's uuid is not the same owner uuid - - -### Expected behaviour -They should be the same uuid. - - -### Steps to reproduce - -![image](/uploads/c58f93291596954fb19994323c848aa1/image.png) -Here look at the **ownerId** - -![image](/uploads/fad2fe38c985e503d073017dbb873e51/image.png) -Here the user's id (in fact uuid) from the `/data-projects/:projectUUID/users` endpoint - - -### Todo - -* [ ] add a flyway migration to add some more FKs - * e.g., more subject-related FKs -* [ ] inform *all devs* when this MR is merged instantly -* *every machine with data inconsistencies will break and crash at beginning!* -* then we know, which machine has consistency problems and needs to clean its database state (dropping or fixing) - -Please apply that patch LOCALLY: - -``` --- If some strange data exists (Account is owner instead of Subject), this should fix it OR tell you that your database is broken --- IF your local database is broken, please clean it. FKs and constraints already exist. -UPDATE mlreef_project -SET owner_id = a.person_id -FROM account a -WHERE owner_id = a.id; - -UPDATE data_processor -SET author_id = a.person_id -FROM account a -WHERE author_id = a.id; - -``` - - -relates to: mlreef!503",5 -47592036,2020-06-18 07:22:39.757,Backend EPF Magic Everything ticket,"# -# Please select the most appropriate ticket template from the dropdown above -#",30 -46589542,2020-06-17 13:25:34.036,Create mlreef routes,"### Problem to Solve -Create mlreef routes as compared to gitlab in ticket #441 - -### MLReef public routes - -``` -https://mlreef.com/explore/mlprojects -https://mlreef.com/explore/visualizations -https://mlreef.com/explore/dataops -https://mlreef.com/explore/groups -https://mlreef.com/:username -https://mlreef.com/:namespace/:project_slug/ -https://mlreef.com/:namespace/:project_slug/-/tree/:branch -https://mlreef.com/:namespace/:project_slug/-/blob/:branch/path/to/file/:file_name -https://mlreef.com/:namespace/:project_slug/-/blame/:branch/path/to/file/:file_name -https://mlreef.com/:namespace/:project_slug/-/commits/:branch -https://mlreef.com/:namespace/:project_slug/-/commits/:branch/path/to/file/:file_name -https://mlreef.com/:namespace/:project_slug/-/project_members -https://mlreef.com/:namespace/:project_slug/-/jobs -https://mlreef.com/:namespace/:project_slug/-/jobs/:job_id -https://mlreef.com/:namespace/:project_slug/-/experiments -https://mlreef.com/:namespace/:project_slug/-/experiments/:pipeline_ID -https://mlreef.com/:namespace/:project_slug/-/experiments/:experiment_ID -https://mlreef.com/:namespace/:project_slug/-/experiments/:experiment_ID#training -https://mlreef.com/:namespace/:project_slug/-/experiments/:experiment_ID#files -https://mlreef.com/:namespace/:project_slug/-/datasets -https://mlreef.com/:namespace/:project_slug/-/datasets/:pipeline_ID -https://mlreef.com/:namespace/:project_slug/-/datasets/:dataset_ID -https://mlreef.com/:namespace/:project_slug/-/visualization -https://mlreef.com/:namespace/:project_slug/-/visualization/:pipeline_ID -https://mlreef.com/:namespace/:project_slug/-/visualization/:visualization_ID -https://mlreef.com/:namespace/:project_slug/-/merge_requests -https://mlreef.com/:namespace/:project_slug/-/merge_requests/:mergerequest_ID -https://mlreef.com/:namespace/:project_slug/-/merge_requests/:mergerequest_ID/diffs -https://mlreef.com/:namespace/:project_slug/-/merge_requests/:mergerequest_ID/commits -``` - - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -### Terminology - -* **dashboard**: this is our marketplace. I like dashboard better as our marketplace is not a real marketplace where you buy stuff. - - -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -===================== - -related: #441",0 -46145602,2020-06-17 06:01:14.504,Backend: Update user endpoint,"### Problem to Solve -Initially we decided that User profile operations would go to Gitlab directly, but after some research there was found that Update operation was impossible by simple user (https://docs.gitlab.com/ee/api/users.html#user-modification). -Therefore the update operation should be performed by backend using chain: frontend -> backend -> Gitlab - - -### Technical Solution -Add new endpoints for user update operation. Obtaining user information should go to Gitlab directly as it was decided before - -* [ ] check, what data we want support in backend -* [ ] change name, email in backend -* [ ] solution 1) forward avatar image blob to gitlab -* solution 2) avatar will be changed via gitlab directly (no backend entity), if that is possible (which maybe it is not, then use backend for that) - -### Terminology -Simple user - non-administrator user - -* https://docs.gitlab.com/ee/api/users.html#user-modification -User update operation - changing name, email, avatar etc of user",6 -43018523,2020-06-12 17:06:05.159,Esthetic bugs in develop branch,"### Observed behaviour - - - -### Expected behaviour -- [x] ""New Merge Request"" button in new merge request view has missing text: -![image](/uploads/1f4ec6f938780b284676f7f3162419f6/image.png) - -- [x] Strange looking edges in the project avatar (with or without images, these edges appear) -![image](/uploads/924b4f71e46ed0721e19d09aec5f8f79/image.png) - -- [x] Data visualization view has a grey line (should be deleted) and the filter buttons should be left aligned -![image](/uploads/9dbf9acd3a6a3cc4d722d8a86eb2abaa/image.png) - -- [x] Same for datasets view: -![image](/uploads/0a725abea01a22d6fd2fb85d9c17657e/image.png) - -- [x] Select data modal has undifined as first rout? and the checkbox is hidden and not aligned with the file view: -![image](/uploads/4144c460dfa49222f4b192eda12c15c5/image.png) - -- [x] The parameter name and the input field are not aligned properly (very slightly though) -![image](/uploads/965ef31376baf3960bbfe0754650e44e/image.png) - -- [x] In the experiment list, the first entry (only the text on the left side) should be moved a bit down. And there is a separation line missing between experiments: -![image](/uploads/a19491164ca956ea684ea885c43839fb/image.png) - -- [x] In the experiment detail view, the success status is red. It should be green on success, oragne on pending, green on running and red on failed or aborted. -> please I need the step to reproduce it - -**Open develop / open resnet50 project / go to experiments / open running experiment (holy-siren) / in the details tab the ""running"" is red** - -![image](/uploads/6904ceeed37514ca5d4fc31a4e8a6dc8/image.png) - - -- [x] The download and delete buttons need mlreef styling ;) -> please I need the step to reproduce it - -**Open develop / open resnet50 project / go to experiments / open first successful experiment / move to files /** - -![image](/uploads/c4849c390f83568efae8e763e95bb80c/image.png) - -- [x] In the market place, when more than 3 projects appear, the filter and the new ML project buttons are separated. They should stay on the same when more projects are added. -![image](/uploads/22a5173e45bc8a82446f2448f24bd3c2/image.png) - -- [x] In the market place, the data types icons are always the same as well as the description. Probably also related to https://gitlab.com/mlreef/mlreef/-/issues/401 - -Link to XD: https://xd.adobe.com/view/5d9dcfd2-3bfb-4ee0-5069-8c9e2f0a5b7b-2bac/ - -![image](/uploads/1bce458db027c7c4a88fe68327375dcf/image.png) - - -### Steps to reproduce - -All visual bugs are found in the current develop branch (12/06/2020) - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",13 -43017841,2020-06-12 16:42:42.420,New Group leads to user profile,"### Observed behaviour - -When clicking on the navbar drobdown to ""new group"", one gets to the (still mocked?) profile view. - -![image](/uploads/3818eb3f99544b552ac706ea29d655f7/image.png) - -@andres46 I think we already had the form for creating a new group, didnt we? - -### Expected behaviour - - - -### Steps to reproduce - - - -### Terminology -> Define terms that are important in the context of this task - - - -### Solution -> Add technical implementation details and the results of the tickets discussion here. - -* [ ] Add at least one test - - - -Additional Notes -===================== - - -relates to:",1 -41162369,2020-06-10 16:59:21.108,Extra parameters and fields required for data pipelines,"### Problem to Solve -A data processing pipeline currently provides the following information: -![pipeline](/uploads/257372836cc04f6b9b164fd3dbad3514/pipeline.PNG) - -Some additional information that is necessary from backend: -1. The corresponding gitlab pipeline information (as we have with experiments). -This will allow getting information such as **pipeline status**, **pipeline creation, updation times** and **duration** -of pipeline etc.... - -### Technical Solution -The required information is available for the frontend - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -=====================",1 -35483576,2020-06-05 13:42:57.430,Backend: Refactor DataProjects to have more fields from MarketplaceEntries,"### Problem to Solve - -* refactor Projects to have DataTypes[] (move them from Marketplace Entries) -* Tags are on Marketplace Entries - -### Technical Solution - - -### Terminology - -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md)",5 -35390200,2020-06-03 14:43:07.645,Add Tests and Docu for Delete Data Instance,The Documentation for deleting a data instance is missing,1 -35330205,2020-06-02 18:23:32.454,Improve EPF Bot lifecycle together with group setup,"### Problem to Solve - -after mlreef#526 and backend!72 it is necessary to clean up technical debt -Problem: right now we use the user as GIT_PUSH_USER and GIT_PUSH_TOKEN, which is a security leak. -We already have te mechanism to create an ""EPF-Bot"" user, but now this was created for the user's group, and is a Maintainer, which is equally insecure as using the user, and even worse, is not really connected to project. - -So actually we need to solve many problems: -* when creating a user, a user group must be created. *This group must be used as the owner* for every new project. -* For every dataproject of this group, a specific EPF-Bot can be created, for example ""mlreef-project-$UUID-$ID-bot"" for a data/code project with UUID $UUID and the connected gitlabId $ID. -* when a pipeline is started (before the gitlab pipeline call takes place), the following must be ensured: - * the project EPF-Bot exists and has a token - * that token can be stored as GIT_PUSH_TOKEN for that project (once per project is enough, no need for every CI) - * that epf-bot username can be stored as GIT_PUSH_USER for that project - * the custom EPF_URL (to update/finish the experiment/pipeline) is created for this distinct PipelineInstance, therefore it shall exist as a pipeline-scoped CI variable and must not exist as a project-scoped CI variable - -**Final Notice**: Right now we have the user as the owner of its resources. But actually we need to have the user's group to be the owner, to allow to add the epf-bot as a group member of that project (with DEVELOPER roles, should be sufficient). As a hack, we could add the epf-bot as a free floating user and just assign to that project. Maybe that is even easier, as we dont need the group- - -### Technical Solution -> Add technical implementation details and the results of the ticket's discussion here. - - -### Terminology -[MLReef's Glossary](https://gitlab.com/mlreef/www-mlreef-com/-/tree/master/handbook/glossary.md) - - - -Additional Notes -=====================",11