rusackas opened a new pull request, #44120:
URL: https://github.com/apache/superset/pull/44120

   ### SUMMARY
   
   The 🎪 Showtime "Sync PR to desired state" job has been sitting perpetually 
`pending`/queued on nearly every PR, while every other CI job runs fine. Root 
cause: `showtime-trigger.yml` fires on 
`labeled`/`unlabeled`/`synchronize`/`closed` — far more events per PR than 
typical CI's `synchronize`-only trigger — with no `concurrency:` group. Every 
one of those events queues a brand-new run, and nothing ever cancels a stale 
one.
   
   Checked the live queue directly: **68 of the repo's 91 currently-queued 
Actions runs (75%) were this one workflow**, while every other workflow had 1-2 
queued (normal). They were stacking per-PR — one branch had 6 simultaneously 
queued runs, all competing for the same runner pool everything else shares 
cleanly. In an actively automated repo (frequent relabeling, force-pushes, 
rebases), that backlog balloons fast and the job ends up starved.
   
   `showtime-cleanup.yml` (the sibling workflow) only does scheduled TTL 
cleanup of deployed *environments* every 6 hours — nothing currently cancels 
stale *Actions runs* for this workflow.
   
   ### THE FIX
   
   Adds a per-PR concurrency group with `cancel-in-progress: true`, so a new 
trigger for a PR cancels any still-queued/in-progress Showtime run for that 
same PR instead of stacking behind it.
   
   This is correct here, not just a queue-relief hack: the job's whole purpose 
(`showtime sync`) is reconciling to the *current* desired state, so a run still 
working toward an already-superseded state is wasted work whether it's queued 
or mid-sync — cancelling it in favor of a fresh run is the right behavior, 
matching the `cancel-in-progress` pattern already used elsewhere in this repo's 
CI (e.g. `enforce-single-migration-head.yml`).
   
   ### TESTING INSTRUCTIONS
   
   No functional change to what Showtime does, only when a run gets superseded. 
`zizmor` (GHA security audit) passes on the change. Should observe the 
queued-run count for this workflow drop sharply once merged; can verify with:
   
   ```bash
   gh api "repos/apache/superset/actions/runs?status=queued&per_page=100" \
     --jq '[.workflow_runs[] | select(.name=="🎪 Superset Showtime")] | length'
   ```
   
   ### ADDITIONAL INFORMATION
   
   - [ ] Has associated issue:
   - [ ] Required feature flags:
   - [ ] Changes UI
   - [ ] Includes DB Migration (follow approval process in 
[SIP-59](https://github.com/apache/superset/issues/13351))
     - [ ] Migration is atomic, supports rollback & is backwards-compatible
     - [ ] Confirm DB migration upgrade and downgrade tested
     - [ ] Runtime estimates and downtime expectations provided
   - [ ] Introduces new feature or API
   - [ ] Removes existing feature or API


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to