You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As a good practice we should set resource limits on each container sf-operator deploys. Without
them, containers are not constrained in CPU and MEM usage. Furthermore some cluster force
the use of a LimitRange resource for a namespace and this applies the same set of default values
on every container whatever it is.
https://kubernetes.io/docs/concepts/configuration/manage-resources-containers/https://kubernetes.io/docs/concepts/policy/limit-range/
So with this change:
- All containers get a default Resources Limit which is (request.mem: 128Mi, request.cpu: 100m, limit.mem: 256Mi, limit.cpu: 500m)
- git-server - remove the nodepexporter container because the volume is not supposed to increase in size as we are (control plan) on control of the content stored into that storage
- git-server - Add probes
- Add a SetContainerLimitsLowProfile function used to set low limits for sidecar or init containers when it seems to make sense.
- Add a SetContainerLimitsHighProfile function used to set limits for containers that require more resources.
- Update CRD to add a LimitsSpec for MariaDB, Zookeeper, Nodepool launcher and builder, Zuul web, scheduler, executor, merger.
This allows to customize the limits.
TODO: Add in documentation the constraints about the requests/and limits.
Change-Id: I743d38b09a0ba830511c56558f75d083272d3450
0 commit comments