Arvados

• Assigned to the run-tests pipeline the same set of sub-jobs as developer-run-tests to avoid duplicated pipelines.
  • I figured that while individual sub-job histories will get messier in the sense that both developer and merged commits will get included, we can still track failures by just looking at their parent job's history. I think it's a good compromise to get things simpler on the managing side while still having a way of tracking failures based on the parent jobs.
• Jenkins cleanup of disabled and unused jobs
  • Arvados Pipeline Jobs
    • diagnostics-ce8i5
    • deploy-workbench2-to-tordo
    • deploy-workbench2-to-ce8i5
    • deploy-workbench2-to-9tee4
    • workbench2-run-tests
    • workbench2-build-packages
    • deploy-to-9tee4
    • deploy-to-ce8i5
    • run-tests-federation
    • build-composer-rc-packages
    • run-tests-services-nodemanager
  • Packer Images
    • compute-image-cleaner-ce8i5
    • compute-image-cleaner-su92l
  • Diagnostics
    • diagnostics-su92l
    • diagnostics-9tee4
  • Performance
    • performance-suite-4xphq
    • qr1hi-automated-performance-suite
  • Developer
    • developer-diagnostics-9tee4
    • developer-run-tests-services-nodemanager
    • developer-tests-k8s-GKE
    • developer-tests-k8s-minikube
  • CWL
    • Deploy production cwlviewer (view.commonwl.org)
    • Deploy staging cwlviewer (cwlviewer-dev.arvados.org)
    • run-cwl-test-9tee4-old-unused
    • run-cwl-test-c97qk
    • run-cwl-test-ce8i5-old-unused
  • Deploy
    • deploy-to-4xphq
    • deploy-to-c97qk

Potential follow-up tasks¶

• Deleting the unused run-tests-* and developer-run-tests-* subjobs
• Renaming developer-run-tests-remainder, developer-run-tests-doc-pysdk-api-fuse, developer-run-tests-services-workbench2-e2e and developer-run-tests-services-workbench2-components so that they don't include the word "developer" as they're being used on different main pipelines. (I haven't done this yet because it would break some links from recent/in-progress tickets, and that might be annoying so I left that for further discussion)

Lucas Di Pentima wrote in #note-7:

• Assigned to the run-tests pipeline the same set of sub-jobs as developer-run-tests to avoid duplicated pipelines.
  • I figured that while individual sub-job histories will get messier in the sense that both developer and merged commits will get included, we can still track failures by just looking at their parent job's history. I think it's a good compromise to get things simpler on the managing side while still having a way of tracking failures based on the parent jobs.

I agree, this LGTM. Thanks.