fix: update cluster-proportional-autoscaler path at MCR #3091
Conversation
Codecov Report
@@ Coverage Diff @@
## master #3091 +/- ##
==========================================
- Coverage 70.65% 70.62% -0.03%
==========================================
Files 147 147
Lines 25407 25407
==========================================
- Hits 17951 17944 -7
- Misses 6333 6341 +8
+ Partials 1123 1122 -1
Continue to review full report at Codecov.
|
pkg/api/k8s_versions.go
Outdated
| @@ -19,7 +19,7 @@ const ( | |||
| calicoCNIImageReference string = "cni:v3.8.0" | |||
| calicoNodeImageReference string = "node:v3.8.0" | |||
| calicoPod2DaemonImageReference string = "pod2daemon-flexvol:v3.8.0" | |||
| calicoClusterProportionalAutoscalerImageReference string = "cluster-proportional-autoscaler-amd64:1.1.2-r2" | |||
| calicoClusterProportionalAutoscalerImageReference string = "oss/kubernetes/autoscaler/cluster-proportional-autoscaler:1.1.2-r2" | |||
There was a problem hiding this comment.
I think we need to turn this into the full mcr URL, and get rid of the kubernetesImageBase prefix dep. Basically we don't need to support both GCR and MCR types here.
Also, not to add scope creep, but is it worth seeing if a newer version of this works w/ the current calico spec?
There was a problem hiding this comment.
is it worth seeing if a newer version of this works...?
I think so. I added a commit to put all three versions of C-P-A in the VHD but ideally we would all use 1.7.1 (or the 1.7.1-hotfix?). I'll try that next.
There was a problem hiding this comment.
Actually, since this is a new regression in v0.49.0, let's defer the version bumps to a separate PR.
mboersma
force-pushed
the
fix-cluster-prop-auto
branch
from
5dbc98e
to
63fc761
Compare
|
/lgtm |
* fix: update cluster-proportional-autoscaler path at MCR * chore: add other cluster-prop-auto versions to Linux VHD * fix: remove DNSAutoscalerAddonName after rebase fail * chore: remove unreferenced c-a-p version from Linux VHD * refactor: remove kubernetesImageBase prefix from calico c-p-a
|
Hi @jackfrancis @mboersma when will this been release? :) |
|
@andyliuliming we are in the process of validating master for a v0.50.0 release by end of week |
Reason for Change:
The cluster-proportional-autoscaler image reference at mcr.microsoft.com was incorrect. Now that MCR is the default, this could cause calico cluster deployments to fail.
Issue Fixed:
Fixes #3079
Requirements:
Notes:
I did the before and after tests locally to confirm this fixes things. This probably should have been caught somewhere in Jenkins e2e, however. I will follow up.