fix readiness probe regression#18679
Conversation
Signed-off-by: Rama Chavali <rama.rao@salesforce.com>
| if p.lastKnownState.versionStats.CDSVersion > 0 && p.lastKnownState.versionStats.LDSVersion > 0 { | ||
| // Envoy seems to not updatig cds version (need to confirm) when partial rejection happens. | ||
| // Till that is fixed, we should treat LDS success as readiness success as LDS happens last. | ||
| if p.lastKnownState.versionStats.LDSVersion > 0 { |
There was a problem hiding this comment.
@howardjohn for partial cluster rejections (that is what is happening in tests), Envoy seems to be not updating this (need to verify though), that is why we are seeing cds as Not Received and lds as Received. For now, I changed it like this - I will verify in Envoy and follow-up
howardjohn
left a comment
There was a problem hiding this comment.
I don't think this fully solves the regression, we are seeing constant timeouts on the readiness probe..
Will approve since this fixes it a bit, but if we can't properly fix we should revert to old behavior most likely?
Sure let us see how this goes. But timeout is very surprising to me. Let us watch and if it continues I can revert even though it is not fully functionally correct. |
Signed-off-by: Rama Chavali <rama.rao@salesforce.com>
|
@howardjohn If this continues to have issues, I have prepared a PR #18685 to revert back to update status stats and cache them. - Please merge that |
* fix readiness probe regression Signed-off-by: Rama Chavali <rama.rao@salesforce.com> * fix unit test Signed-off-by: Rama Chavali <rama.rao@salesforce.com>
Fixes #18665