[Linux-cluster] unfreeze a node of the cluster and cause reboot remaining nodes

[aneesh vs]

Hello,

Read "man clusvcadm" to know about freeze feature . clusvcadm -Z doesn't
freeze a node, instead it will freeze monitoring the service.
Please refer https://access.redhat.com/kb/docs/DOC-43505 and
http://sources.redhat.com/cluster/wiki/ServiceFreeze more details about this
feature.

The service going in failed may have many reasons. If any of resource's
status check fails, rgmanager will use recovery policy and try to restart or
relocate. But as you said, if there is any issue in unmounting fs resource,
rgmanager will move service to failed state. If you set self_fence=1 in fs
resource section and if force_unmount to fs resource is not successful, node
will do self fence.


>>During servera reboot, servicea can not failover to standby node and the
service state shows "recoverable"
You atleasr need paste snip of /var/log/messages from all nodes for this
time to explain exactly what is happening.

>>After servera rebooted successfully, servicea can run on servera but then
serverb and serverc reboot togeter.
Need to check logs to know reason for this.



On Tue, Jan 11, 2011 at 2:49 PM, KC LO <kclo2000 at gmail.com> wrote:

> Dear all,
>
> We have set up a 3 + 1 cluster which is 3 active node and 1 standby nodes
> and quorum disks.
>
> clustat
> Member Status: Quorate
>  Member Name                             ID   Status
>  ------ ----                             ---- ------
>  servera                                1 Online, rgmanager
>  serverb                                2 Online, rgmanager
>  serverc                                3 Online, rgmanager
>  standby                               4 Online, Local, rgmanager
>  /dev/emcpowers                    0 Online, Quorum Disk
>
>  Service Name                 Owner (Last)                   State
>  service:servicea              servera                   started
>  service:serviceb              serverb                   started
>  service:servicec              serverc                   started
>
> Any server failure and cause server relocate to the standby server and
> basically all cluster functions properly.
>
> However, when I type clusvcadm -Z servera, it can sucessfully freeze the
> nodes.  However, if I type clusvcadm -U servera to unfreeze the node, it
> will check the status of the running application under cluster monitoring.
> But don't know why it return status failed while the application is running
> properly.  It will then try to stop the application and reported that it
> failed to unmount the partition and cause servera rebooted.  During servera
> reboot, servicea can not failover to standby node and the service state
> shows "recoverable".  After servera rebooted successfully, servicea can run
> on servera but then serverb and serverc reboot togeter.
>
> Do you have any idea?
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20110111/45192f50/attachment.htm>