@Kyriakos Akriotis some info on the operator

I had a look at your changes and a) inverting completely the rule when the controller will try to reconcile/reevaluate the quorum strikes rather odd and b) I’ve set a 1 minute interval between reconciliation loops not totally arbitrarily; while kubernetes itself checks for ConfigMap changes every 2mins, braft - that typesense is using as raft implementation - checks for node ip changes in the nodelist every 30sec. So I tried to meet it somewhere in the middle. I will give your changes a test-drive as well, but as said, flipping the ReconcileQuorum logic only when the ConfigMap is updated is odd. The ReconcileQuorum method reevaluates the whole quorum and adjusts the cluster ,and the quorum in a certain extent, based on the aggregated health of the cluster in that moment. If you skip it you miss the whole second part; you might have correct nodelist entries but nobody gives you a guarantee for what’s going on inside typesense pods. Anyways test it as well at your end and let me know how it went. I guess opening a Discussion thread in the upstream repo would fit better for that kind of very operator-specific matters than spamming this channel.