Closed RonnanSouza closed 1 year ago
If the non leader instances become ready then Flux will stop working because non-leaders don’t produce artifacts.
Ohh, I see. But this behavior only applies to the source controller, right? I noticed that all the other controllers have the non-leaders ready. Now thinking better about this I understand that it makes total sense but I think we lack docs about it (actually about the whole HA thing on flux), mainly because where depending on the chart-specific config to make the entire flux work, can't we handle this on the source controller side?. Anyway, thanks for the clarification on it Stephan.
source-controller is the only stateful component of Flux, that's why we can't make the non leaders ready like we do with all the stateless controllers. If you want Flux to recover faster after the node where source-controller runs crashes, you should use PVC for the /data
volume, when source-controller starts on a new node, it doesn't have to pull all the upstream repos since the PV will contain the latest artifacts.
fix: set right port and path for readiness probe
Signed-off-by: Ronan ronan.souza@wildlifestudios.com
What this PR does / why we need it:
It fixes the readiness probe for the source controller in the flux2 chart. Currently, the chart is using the wrong port and path for the readiness probe causing non-leader source controller pods to keep in a not-ready state indefinitely.
Which issue this PR fixes
(optional, in
fixes #<issue number>(, fixes #<issue_number>, ...)
format, will close that issue when PR gets merged)Special notes for your reviewer:
Checklist
make reviewable