With AIOps analytics deployment migration over to OCP prod (new projects are expected to be deployed there as well) a default resource limit is enforced on us. This limit is pretty strict:
$ oc describe limits aiops-prod-argo-limits
Name: aiops-prod-argo-limits
Namespace: aiops-prod-argo
Type Resource Min Max Default Request Default Limit Max Limit/Request Ratio
---- -------- --- --- --------------- ------------- -----------------------
Container memory - - 400Mi 1000Mi -
Container cpu - - 300m 500m -
This limit results in an OOM kill on version analysis notebook of Openshift SME.
Luckily the Min and Max is not set so we can override these defaults on per step/container bases. I've chose these (hopefully reasonable) values which should be safe and suitable for most projects for now. The same PR will be opened against project-template as well.
With AIOps analytics deployment migration over to OCP prod (new projects are expected to be deployed there as well) a default resource limit is enforced on us. This limit is pretty strict:
This limit results in an OOM kill on version analysis notebook of Openshift SME.
Luckily the
Min
andMax
is not set so we can override these defaults on per step/container bases. I've chose these (hopefully reasonable) values which should be safe and suitable for most projects for now. The same PR will be opened against project-template as well.This is a same PR as opened against Openshift SME https://github.com/aicoe-aiops/openshift-sme-mailing-list-analysis/pull/23