[ Pasting Tirthankar's review reply here, as it shows up only on the web-view of the thread in the forum ]
> Hi Sambit, > > Logically your changes look fine. > But I am concerned about the CMM_TRACE message. > Is that message really required. > What value add does it provide. > CMM debug buffer is limited, putting > unnecessary information just causes clutter. > > You could potentially change it to CMM_DEBUG_TRACE. > > Thanks, > Tirthankar <http://blogs.sun.com/tirthankar> Hi Tirthankar, Thanks for the review. The CMM_TRACE message is being printed only when the node is shutting down due to which the CMM does not need to to a ping health check. That will record whether CMM attempted to do the ping health check or not; helpful for knowing what was going on. To reiterate, note that the message is not printed in every reconfiguration. Its printed only when the CMM is shutting down (shutdown reconfigurations before the nodes go down). So I think its ok to leave the trace message as is. Thanks & Regards, Sambit Sambit Nayak wrote: > Hi everyone, > > Please review the fix for bug 6817566 > (http://bugs.opensolaris.org/view_bug.do?bug_id=6817566) > *Synopsis* - > CMM: Halting this node because door_ki_open of > /etc/cluster/qd_userd_door returned error 22. > > You can access the webrev at : > http://cr.opensolaris.org/~samnayak/webrev-6817566/ > > > Explanation of the fix > -------------------------------- > The ping check is done by CMM automaton in begin state of every > reconfiguration. > CMM automaton executes the ping using a userland daemon qd_userd. > > During a shutdown, the daemon could have been gone by the time the > automaton tries to do a ping check in begin state. The door_ki_open() > for qd_userd done by automaton in begin state can fail then. > > The automaton does not need to do the ping check when the node is > shutting down (shutdown intent or commit flag is set in CMM). That is > the fix implemented. > > > Thanks & Regards, > Sambit > > _______________________________________________ > ha-clusters-discuss mailing list > ha-clusters-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/ha-clusters-discuss
