ci: stop running vLLM GCE test on every PR - #958
Merged
Conversation
This job provisions a real GPU-backed GCE instance per PR run, and has been flaking heavily on shared infra issues unrelated to code changes: GPU zone capacity stockouts across multiple regions, serial-port-output fetch errors, and leftover-instance naming collisions from prior failed runs. It also already carried (broken) logic to skip itself on dependency-bump PR titles, which never matched our actual titles, so it kept running and blocking merges on every dependency PR. Switch it to workflow_dispatch-only so it can still be run manually or wired into a release/nightly job later, without gating every PR on GCE capacity availability. Signed-off-by: Nick Mitchell <nickm@us.ibm.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Test in GCE VMprovisions a real GPU-backed GCE instance per PR run and has been flaking heavily today on shared infra issues unrelated to code changes: GPU zone capacity stockouts across multiple regions, serial-port-output fetch errors, and leftover-instance naming collisions from prior failed runs. It's also a required status check, so every flake blocks merging regardless of content.The workflow already carried (broken) logic to skip itself on dependency-bump PR titles (
chore(deps)/fix(deps)), but that never matched our actual PR titles, so it kept running and repeatedly blocking merges today.Switches the trigger to
workflow_dispatch-only so it can still be run manually (or wired into a release/nightly job later) without gating every PR on GCE GPU capacity availability.Note: this PR alone won't fully unblock things —
Test in GCE VMis also in the branch ruleset's required-status-checks list, so it'll need to be removed there too (same fix as the earlier staleCodeQLcheck), otherwise a disabled job just leaves the check permanently pending.Test plan
pull_requestanymore)