Add a endpoint based sync point after scaling up the pods in import step #14
Conversation
Signed-off-by: Sergen Yalçın <yalcinsergen97@gmail.com>
sergenyalcin
force-pushed
the
change-sync-point-to-svc-endpoints
branch
from
9d3d078
to
3e1eca3
Compare
| fi | ||
| done | ||
|
|
||
| if ${KUBECTL} get managed ; then |
There was a problem hiding this comment.
The API conversion functions (invoked by the conversion webhooks) may not kick in here, if the CRD API storage version is the highest priority version, e.g., when the available versions are v1beta1 and v1beta2 and the storage version is v1beta2.
How about parameterizing this function with the arguments to be passed to kubectl and retry the actual command for a finite count till it succeeds?
Looks like we should also be able to utilize a DeploymentRuntimeConfig to introduce a readiness probe for the webhook service. This together with the check on the endpoints can be an alternative.
There was a problem hiding this comment.
I added another script to retry the patch step.
Signed-off-by: Sergen Yalçın <yalcinsergen97@gmail.com>
sergenyalcin
force-pushed
the
change-sync-point-to-svc-endpoints
branch
from
b67322a
to
840f4e7
Compare
| - script: | | ||
| #!/bin/bash | ||
|
|
||
| function check_endpoints { |
There was a problem hiding this comment.
Isn't it possible to define a generic kubectl_retry function that will retry the passed kubectl command together with its arguments and call it for kubectl get endpoints & kubectl patch?
Signed-off-by: Sergen Yalçın <yalcinsergen97@gmail.com>
sergenyalcin
force-pushed
the
change-sync-point-to-svc-endpoints
branch
from
3b4b1b0
to
701c15b
Compare
Description of your changes
Follow-up of #13. We observed some issues on pod-based sync point for import step in provider-gcp tests. This PR adds an endpoint-based sync point after scaling up the pods in import step.
kubectl patchsteps.I have:
make reviewable testto ensure this PR is ready for review.How has this code been tested
Tested locally.