Glusterd2 services behavior when external etcd services are went down

[rmadaka]

-> If we stop the external etcd services, glusterd2 services will be in running state but glustercli commands will not work. -> If we bring back the external etcd services immediately, then glustercli commands will start working.

But here issue is that: -> Once we stop the external etcd services, if we are not bringing back the external etcd services for some hours, glusterd2 services will stop running. -> After some hours if we are bringing up the external etcd services, glusterd2 services will not come up. Even restart of glusterd2 services also not able to bring back the glusterd2 services. Only option to bring up the glusterd2 servcies is node reboot (where glusterd2 is running).

-> After node reboot glusterd2 servcies will start running.

[prashanthpai]

-> After some hours if we are bringing up the external etcd services, glusterd2 services will not come up. Even restart of glusterd2 services also not able to bring back the glusterd2 services. Only option to bring up the glusterd2 servcies is node reboot (where glusterd2 is running).

This is a known issue. Etcd can't come back up when there is quorum loss. I'm certain other issues capture this.

Can you try this with bringing one etcd down in a 3 node cluster ?