fix(e2e): detect gateway workload for health port-forward#2400
fix(e2e): detect gateway workload for health port-forward#2400krishicks wants to merge 1 commit into
Conversation
Signed-off-by: Kris Hicks <khicks@nvidia.com>
|
Label |
elezar
left a comment
There was a problem hiding this comment.
This is a focused and sound fix. The e2e harness now discovers whether the chart created a StatefulSet or Deployment before opening the health port-forward, preserving StatefulSet-backed runs while supporting Deployment-backed installations, including HA.
The optional HA job confirms that this part works: it reaches deployment/openshell, establishes the health port-forward, and passes the readyz database check. Its later supervisor session not connected failures are the known cross-replica routing limitation addressed by #1868, so they should not block this PR.
#2256 contained the same patch and has been closed as superseded by this PR. No blocking concerns from me.
|
Label |
|
@krishicks I have removed the |
Ah, OK. Well, at least it fails now for a better reason than before! |
Summary
Fixes the e2e Kubernetes test, which is presently failing with:
This is because OpenShell is deployed as a Deployment in HA deployments, not a StatefulSet.
Related Issue
Changes
Testing
mise run pre-commitpassesChecklist