Recently i found an issue about pod registration on consul client.
I used the official consul helm chart. My pods are well registered to consul client and consul clients (daemonset) are well registered to consul server.
The problem is, say,
pod-replica1 , pod-replica2 -------> consul client on node1
pod-replica3 -------> consul client on node2
If consul client on node1 crashes and restarts again, pod-replica1 and pod-replica2 (and other pod on this node) seem not registering automatically to the new consul client !
How can i solve this issue ?
Is there possible that when consul client dies,it sends crash event to the pods registered onto it and ask them to restart ?