- Add view to retrieve log output of job from external server
- Add argo backend
- Allow setting custom resource requirements for s3fs
- Add liveness probe for s3fs
- Always run s3fs as root
- Update pygeoapi base to 0.18
- Jobs pagination sorting
- Jobs pagination fix
- Set autosave-every to 60
- Implement pagination
- Save all process configuration as parameters, not just input values
- Add image pull secret for image processor
- Allow defining process properties from config for ImageProcessor
- Save execution parameters for image processes
- Allow mounting secrets as env or file for image processes
- Implement failed/success notifications
- Disallow custom images if now allowed regex is specified
- Minor changes for image processor
- Setup fixes
- Add container image execution
- Upgrade underlying pygeoapi to current version
- Send 400 error with message for client errors
- Show executed notebook as part of parameters
- Add extra resource specifications
- Remove kernel autoselect based on ancient images
- Refactor output notebook handling
- Add new
extra_volumes
andextra_volume_mounts
config
- Remove result-link feature
- Remove unused job_resources view
- Use more recent pygeoapi version
- Improve error handling on missing job id
- Add
output_dirname
parameter
- Update to pygeoapi 0.15.0
- Add support for conda store kernels
- Add results_in_output_dir config option
- Version jump due to accidental chart misversioning
- Remove custom job setup script feature
- Set HOME env var to support nebari based images
- Use jovyan user as owner of result notebook to try to avoid spurious permission denied
- Add more qhub-related options
- Run papermill_slack also if papermill fails
- Run papermill_slack if env var is defined
- Setup for new hub
- Save parameters which are passed as yaml as json
- Fix build
- Allow dynamic override of node_purpose protected by allowed_node_purposes_regex
- Delete pods of cancelled jobs by force
- Write output files in subdirectory according to the current day
- Disable papermill autosave
- Disable autosave in papermill
- Write output files in subdirectory according to the current day
- Update s3fs
- Obfusicate parameters which look like passwords in UI
- This is really the same as 0.0.78 but it feels stable now
- Support arbitrary default branch name on git checkout
- Allow specifying images for each execution
- Delete pods of jobs when deleting job
- Fixed label setting
- Add job id to default job output filename
- Implement automatically mounting EDC secrets
- Disable service links
- Allow passing custom tolerations
- Allow specifying sub_path for extra pvcs
- Implement run_on_fargate as request param and allow_fargate in config
- Show more infos for accepted jobs (e.g. not runnig because resource requests too high)
- Allow job service account to be configured
- Allow configuring node_purpose
- Add log_output config
- Extend gpu image detection to support aws images
- Fix retrieving resource usage of jobs with long names
- Hotfix
- git checkout as jovyan user
- Generalize result data directories to arbitrary dirs, not just s3
- Allow mounting secrets as env vars via
env_from
- Wait for result file if queried right after job finish
- Make sidecar container termination more robust
- Add patch to allow overwriting job ids
- Allow specifying git revisions in job request parameters
- Implement resources endpoint for jobs
- Update to latest pygeoapi api (mime type for sync jobs)
- Implement
result_data_directory
parameter for jobs (this mounts a subdir of s3 in/home/jovyan/result-data
) - Add option for git checkout as init container
- Allow for secrets for jobs
- Handle kubernetes interface empty lists being null
- Set propagation policy when deleting kubernetes jobs
- Fix build issues
- Don't wait for s3 forever
- Fix crash when no extra config is specified
- Add kernel option to wps-client
- Validate job parameter
- Set async as default again
- Add wps-client.py
- Switch to new repo and package
- Improve s3 mount start detection
- Wait for s3 mount before starting job
- Change permission error workaround
- Reduced job web-UI due to changes upstream
- Use new job result logic from upstream
- Use pygeoapi/master as base since the manager branch has been merged
- Web UI is now reduced (can't start jobs anymore in browser)