Closed samof76 closed 1 year ago
Here is the query...
count(redis_instance_info{k8s_cluster_name="$cluster", redisfailovers_role="master"}) by (node)
@ese and @Wouter0100 may must randomize the allotement instead of picking the zeroth pod.
IMHO this depends a lot on the specific use case and the architecture where you are integrating your Redis clusters.
From my point of view, different Redis clusters are independent so it is not relevant if they share the same node for their masters while they slaves are well distributed across failure domains.
The schedule constraints should be managed by Kubernetes and the operator should look for mechanisms to leverage like topologySpreadConstraints
, I think maybe a more reliable labeling of the pods to mark the Redis role of Redis instances could help on this task
I agree with @ese.
Ok makes sense
This issue is stale because it has been open for 45 days with no activity.
This issue was closed because it has been inactive for 14 days since being marked as stale.
The master distribution, is fairly skewed, and might be cause of worry when running multiple redis-failover deployments and many masters get crowded to a single node.