Repository navigation
GHCR node credential rollout fails when a node changes during the rollout #5016
Description
🤖 Generated by the Agentic Engineer (interactive Claude Code session)
Evidence
Found by the merge-queue failure count in #4993.
- 3 failed merge-group runs on 2026-09-20 failed in the deploy job's GHCR node credential rollout: the node identity or cordon state changed during the rollout, or the registry credential was not accepted.
- Both affected pull requests later merged with no new push.
- Evidence basis: check annotations only; no logs were read.
- Examples: 35503104796, 35505627116, 35524936547.
Problem
A node change that the candidate did not cause evicts it. The window shows one burst; whether it can recur is not known.
Expected behaviour
A node that something outside the deploy replaces or cordons during the rollout is handled as a known state, and only a real credential failure fails the deploy.
Acceptance criteria
- Read the logs of the runs above and record which condition failed in each.
- Decide whether the rollout can treat that condition as expected; fix it with a test if so, or close this issue with the reason if not.
Size: small. Part of #4979.
Activity
coderabbitai commentedon Oct 17, 2026
📝 Issue Planner
Check the box below or use the
@coderabbitai plancommand to generate an implementation plan and prompts that you can use with your favorite coding assistant.🧪 Issue enrichment is currently in open beta.
You can configure auto-planning by selecting labels in the issue_enrichment configuration.
To disable automatic issue enrichment, add the following to your
.coderabbit.yaml:💬 Have feedback or questions? Drop into our discord!