chore: update memory limit

[dkwon17]

Signed-off-by: David Kwon [email protected]

This PR increases the memory limit to match the memory limit for the che-operator deployment: https://github.com/eclipse-che/che-operator/blob/3830ac4e28983dbe928d0777f63c87bed43feb4b/deploy/deployment/openshift/objects/che-operator.Deployment.yaml#L132

After discussing with @ibuziuk , we plan to do a KIPO 1.0.1 release with this change.

Fixes: eclipse-che/che#21448

[github-actions]

Click here to review and test in web IDE:

[ibuziuk]

LGTM, @dkwon17 could you please clarify if the PR was tested it on a cluster with ~ 300 namespaces?

[dkwon17]

I was able to verify that the PR increases the KIPO pod's memory limit, but I wasn't able to reproduce the original problem on my test cluster even with 300+ namespaces, as the memory usage never exceeds 128mib in my case:

The namespaces were created by running oc new-project <namespace-name> 300 times

[dkwon17]

This is the bundle-image for this PR: quay.io/dkwon17/kubernetes-image-puller-operator-bundle:next, maybe we could try it on the che-dev cluster by running the following? I don't have access to the openshift-operators namespace to test it myself

 operator-sdk run bundle quay.io/dkwon17/kubernetes-image-puller-operator-bundle:next --namespace openshift-operators

[tolusha]

It is ok as workaround but we need to consider a robust solution.
Either use non caching client or limit cache by some label selector.