How to configure affinity for inClusterCache and inClusterDatabase?
It seems that in the current version we can not specify the affinity for the inClusterCache (redis) to ensure that the PODs are scheduled on different hosts. A similar configuration also seems to be missing for the inClusterDatabase.
Did we oversee this or is it true that this can not be configured at the moment?
Thanks!
@Martin-Weiss
They can not be configured at the moment. We'll check and see if the affinity of inClusterCache and inClusterDatabase can be supported in the next release (possible v1.2).
Reference for psql: https://github.com/zalando/postgres-operator/blob/b1fab703ee1676cb6b6f00b43bf785cb7a95e30a/manifests/complete-postgres-manifest.yaml#L182
Reference for Redis: https://github.com/spotahome/redis-operator/blob/b3dad57f5c4c94fe48824deacf391ab82232b3d5/example/redisfailover/node-affinity.yaml#L8
They can not be configured at the moment. We'll check and see if the affinity of
inClusterCacheandinClusterDatabasecan be supported in the next release (possible v1.2).
Thanks! For PGSQL there is an option in the operators helm chart : enable_pod_antiaffinity: true that does part of what we need. We might also have to check for supporting zones/regions..
For Redis - I can adjust the CRD object but after changing something in the harbor CRD the Redis CRD is "reset".. so here I assume the harbor operator needs to add the parameters to the redis CRD..
Beside the pqsql and redis deployments affinity settings - is there also a plan to add affinity settings for all other components?
@Ghostbaby please check if underlining postgres-operator and redis-operator support this param
@bitsf copy
support for postgres-operator need more work, need update configmap
would be great to support placement also for minio operator.