Idea #2756
Updated by Tom Clegg over 10 years ago
This page needs to be good at * Monitoring progress while the pipeline is running * Troubleshooting a failed pipeline * Reviewing/investigating details of a finished pipeline Features: * If jobs are running, show log messages in real time using websockets * If jobs are queued, show queue status * If a job was created before pipeline itself, note "re-used existing job queued/finished at time X" * Show more job details, probably rearrange columns ** component name should be plain, not styled with label, not a link to the job ** component specification should be in its own column (the stuff independent of whether a job has been chosen to satisfy this component): script, script_version, script_parameters, etc. ** if a job has been chosen, job detail should be in its own column (only the parts that aren't equal to the component description by definition): details (script, parameters, git commit ** output should be its own column. Displayed as uuid (abbreviated?) link, then commit, [first few] output filenames/sizes ** log should be its own column. Displayed as uuid (abbreviated?) link, then direct link to the one .txt file it always contains filenames/sizes) * Show job log messages, e.g., ** Hover/pop-up to show latest+live logs for a single job that is running ** Persistent "fake terminal window" above/below(?) with merged latest+live log messages from all jobs in this pipeline that happen to be running Infrastructure: * API server should have access to the last few lines of every recent/current job log (via redis) as well as live logs (also via redis) and old/finished jobs (via log collection stored in Keep). Might need some more API to make them accessible via websocket (redis) or workbench (keep)?