I have two managers deployed, active-stdby, on a couple of the monitors. The managers were stopped with the orchestrator, and as expected, cannot be started without a manager running. I started a manager instance manually using ‘systemctl start ceph-fsid@xxxxxxxxxxxx.service<mailto:ceph-fsid@xxxxxxxxxxxx.service>’ and the manager starts fine and is stable, but after a few seconds, is gracefully terminated, saying it received sigterm. I paused the orchestrator and started the manager again. Now it stays running. As soon as I resume the orchestrator, the manager is terminated. Why would the orchestrator be terminating managers constantly? The configuration shows 2 managers should be running, not 0. This is what ‘ceph orch ls’ shows for the 10 seconds before the manager is terminated: mgr 1/2 16s ago 18h count:2 Cephadm logs show: 2021-07-28T18:03:53.508042+0000 mon.host1 [INF] Activating manager daemon host2.ijsxjg 2021-07-28T18:03:54.326785+0000 mon.host1 [INF] Health check cleared: MGR_DOWN (was: no active mgr) 2021-07-28T18:03:54.326941+0000 mon.host1 [INF] Cluster is now healthy 2021-07-28T18:03:54.381231+0000 mon.host1 [INF] Manager daemon host2.ijsxjg is now available 2021-07-28T18:04:38.354264+0000 mon.host1 [INF] Manager daemon host2.ijsxjg is unresponsive. No standby daemons available. 2021-07-28T18:04:38.355071+0000 mon.host1 [WRN] Health check failed: no active mgr (MGR_DOWN) Starting the other mgr gives same result. Jim _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |