Repository navigation
Conversation
The production stop/start regression fails on the current main source because the first immediate start returns success over a stale pre-stop read. The signed candidate requires a complete zero-capacity observation before issuing the first restore. Saved capacities survive a pending stop, an interrupted restore, and a repeated stop. The provider and state packages and full CI harness passed locally. The EKS smoke workflow now compares the complete managed-nodegroup set and desired/minimum/maximum sizes before stop, after stop, and after start. It uses a bounded poll for both transitions. Workflow lint and strict Go lint passed. Native CI and the live EKS capacity and cleanup readbacks remain pending. |
✅MegaLinter analysis: Success✅ Linters with no issuesactionlint, bash-exec, git_diff, hadolint, jscpd, jsonlint, lychee, markdown-table-formatter, markdownlint, prettier, prettier, shellcheck, shfmt, stylelint, syft, trivy-sbom, trufflehog, v8r, v8r, yamllint See detailed reports in MegaLinter artifacts
|
Code coverage overviewLanguages: Go Go / code-coverage/go
View coverage of the 4 most impacted files.
ℹ️ Calculations represent line coverage. Results may change as calculations are updated. |
@codex review for EKS lifecycle state consistency across repeated or interrupted stop and start |
Codex Review Summary: ✅ CompletedThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
Codex Review: Didn't find any major issues. You're on a roll. Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |

Why
Starting an EKS cluster immediately after stopping it can report success while AWS is still scaling the workers down. That also discards the sizes needed to recover the cluster.
What
KSail checks that stop has taken effect before restoring workers and keeps their saved sizes until AWS reports the restored capacity. The EKS smoke test checks both transitions independently.
Part of #7721