Registries MemoryLimit is too low and deployment fails on OpenShift

[l0rd]

Describe the bug

Currently the memory max limits for the plugin and devfile pods are set to 32Mi. On some OpenShift clusters that is not allowed: the minimum is 256Mi.

Che version

7.0.0-rc-4.0

Steps to reproduce

Use the operatorhub to deploy Che on openshift online

Expected behavior

Che is deployed

Runtime

OS 4

Installation method

OperatorHub

Environment

Cloud: openshiftworkshop.com

[skabashnyuk]

Can we set a limit to 0? WDYT?

[l0rd]

Discussing with @davidfestal that has debugged the problem it looks like the solution would be to set memory request to 16Mi and limit to 256Mi.

@skabashnyuk I am not sure that this should be fixed Helm side as well (since it's an OpenShift specific problem) but it maybe a good idea to keep consistency on both installers (i.e. it's not a blocker).

[davidfestal]

More precisely the idea is the keep the memory request (= min memory) to 16Mi, but increase the memory limit (= max memory) to 256Mi to satisfy a common Openshift cluster configuration.

[davidfestal]

So +1 to do that for Openshift only at first.

[skabashnyuk]

Removed team/platform since helm is not in the scope

[l0rd]

@skabashnyuk as I said in the comment above helm and operator should be consistent. Helm is in scope by the helm part is not a blocker. You can create a new issue about that or use this one and re-apply your team label

[ibuziuk]

PR merged - eclipse-che/che-operator#64
Should be fixed with the next operator release