08/31/2022, 10:30 AM
Hello Colleagues! I've created a HA cluster of 4 nodes with the PostgreSQL datastore. I'm not planning to add agent nodes, just want to run all payload on the master nodes. I've noticed however that a 4-node cluster is running only one coredns pod, one local-path-provisioner pod etc. What if a node where they are running fails? Should I 1) scale these Deployments up to 3-4 pods 2) run the kube-system pods as DaemonSets instead of Deployments somehow 3) leave everything as it is and wait 5 minutes for the system pods to be rescheduled to a healthy node?