Hi , I am using Pacemaker/Corosync on RHEL 7.4 where facing the below issue randomly.
The corosync reforms the membership randomly even though the service was not restarted. I could observe the below log in the messages file before corosync reforms the membership, Log: May 9 18:02:15 node2 kernel: CPU24: Package temperature above threshold, cpu clock throttled (total events = 3295347) May 9 18:02:15 node2 kernel: CPU0: Package temperature above threshold, cpu clock throttled (total events = 3295402) May 9 18:02:15 node2 kernel: CPU5: Package temperature above threshold, cpu clock throttled (total events = 3295376) . . . May 9 18:02:15 node2 kernel: CPU6: Core temperature/speed normal May 9 18:02:15 node2 kernel: CPU27: Package temperature/speed normal May 9 18:02:15 node2 kernel: CPU3: Package temperature/speed normal May 9 18:02:15 node2 kernel: CPU28: Package temperature/speed normal May 9 18:02:15 node2 kernel: CPU4: Package temperature/speed normal May 9 18:02:15 node2 kernel: CPU30: Package temperature/speed normal May 9 18:02:15 node2 kernel: CPU6: Package temperature/speed normal May 9 18:02:15 node2 kernel: CPU24: Package temperature/speed normal . . . May 9 18:02:15 node2 corosync[30924]: [MAIN ] Corosync main process was not scheduled for 1684.8511 ms (threshold is 800.0000 ms). Consider token timeout increase. May 9 18:02:15 node2 corosync[30924]: [VOTEQ ] waiting for quorum device Qdevice poll (but maximum for 30000 ms) May 9 18:02:15 node2 corosync[30924]: [TOTEM ] A new membership (10.61.198.210:11924) was formed. Members May 9 18:02:15 node2 corosync[30924]: [QUORUM] Members[2]: 1 Is there any relation between the CPU clock throttling and corosync membership . With Regards Somanath Thilak J
_______________________________________________ Manage your subscription: https://lists.clusterlabs.org/mailman/listinfo/users ClusterLabs home: https://www.clusterlabs.org/
