Opensearch v1.3.5
We are part way through a blue green upgrade (Adding extra volumes to ec2 instances so need to delete the old volumes as reducing their size). So we have the cluster running with everything migrating from old to new using routing allocation require. This was all running fine until an old node dropped of due to a network issue (couldn’t contact master node), it recovered but soon after a second one dropped off for a network issue (couldn’t contact master node again) but it couldn’t recover it’s state, so all it’s shards became un-assigned. Is this due to the node attribute not being in the routing allocation require (as we are migrating everything off this node), or is there something else I need to look at?