Check both Supervisor takeover overrides#143
Conversation
|
A repository maintainer needs to approve these workflow run(s). To approve, maintainers can either: • Submit an approval review on this pull request, OR • Submit a review comment starting with Then re-run the failed job(s) via the Checks tab above. Reviews must be on the specific commit SHA of the workflow run to be considered. |
There was a problem hiding this comment.
Pull request overview
This PR updates the legacy supervisor takeover logic so Helios only considers takeover “already configured” when both supervisor DB overrides (apiEndpointOverride and listenPortOverride) match the expected values, ensuring the API route can be restored via service-relay when needed.
Changes:
- Update the embedded takeover script to check both
apiEndpointOverrideandlistenPortOverridebefore declaring idempotent success. - Adjust takeover config documentation to reflect that
listenPortOverrideis no longer treated as the sole idempotency key.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Only treat a legacy Supervisor takeover as configured when both the API endpoint and listen-port overrides match. This lets Helios restore an API override cleared after an orphaned relay is detected. Change-Type: patch Signed-off-by: Dries Van de Velde <dries.vandevelde@c-battery.eu>
7e1014e to
639a408
Compare
|
Hi @driesvdv, thanks again for your contribution. The existing code looks only for the I see this assumption no longer holds with your supervisor PR. I've added some comments there and we can revisit this one once those are resolved. |
Summary
Problem
Helios considered takeover complete when
listenPortOverridealone matched. IfapiEndpointOverridewas removed while the listen-port override remained, Helios would not restore the API route throughservice-relay.This complements the Supervisor recovery in balena-os/balena-supervisor#2536.
Testing
git diff --check