fix: add autoscaling eviction annotation to catalog pods

[exdx]

Signed-off-by: Daniel Sover dsover@redhat.com

Description of the change:
Closes #2666

Because the catalog pods are standalone and not backed by a Deployment/ReplicaSet, it causes issues when draining the underlying node, when then affects autoscaling. This additional annotation on the catalog pods will enable the cluster autoscaler to evict catalog pods when draining a node.

Motivation for the change:
Enable cluster autoscaling to work

Reviewer Checklist

• Implementation matches the proposed design, or proposal is updated to match implementation
• Sufficient unit test coverage
• Sufficient end-to-end test coverage
• Docs updated or added to /doc
• Commit messages sensible and descriptive
• Tests marked as [FLAKE] are truly flaky
• Tests that remove the [FLAKE] tag are no longer flaky

[perdasilva]

/approve @exdx could you please rebase?

[openshift-ci]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: exdx, perdasilva

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [perdasilva]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[exdx]

/approve @exdx could you please rebase?

done -- I think this is a reasonable stop-gap solution for our upstream users until we come up with a better story around how we want to support catalog pods being evicted from nodes during a drain or node scale down event. If we come up with a more robust solution, this annotation still should not be prohibitive.

[elmiko]

from the autoscaler side, +1, thank you for adding this.

[oceanc80]

/lgtm

[exdx]

Rebasing off HEAD of master

[oceanc80]

/lgtm

[kfox1111]

Does this make node draining work too? In that past that hasnt worked. Though I haven't tested on the newest olms.

[exdx]

Does this make node draining work too? In that past that hasnt worked. Though I haven't tested on the newest olms.

Unfortunately no, the catalogsource pod preventing node draining is still an issue. We are tracking that effort in an RFE https://issues.redhat.com/browse/RFE-2737.