In data 15 aprile 2010 alle ore 07:11:15, Tim Serong <[email protected]>  
ha scritto:

> Hi,
>
> It must be broken-NFS-on-DRBD-week (refer to the linux-ha-dev thread
> at  
> http://lists.linux-ha.org/pipermail/linux-ha-dev/2010-April/017296.html)
Sorry if it's my fault :)

>
> On 4/15/2010 at 07:27 AM, Davide D'Amico <[email protected]>  
> wrote:
>> Everything seems fine (drbd-fence-by-handler-ms-drbd0 locations seems
>> right?), but if I reboot the primary node (nfs01), I notice that:
>> - centos is unable to umount cleanly the /data directory;
>> - after the reboot the cluster is in a split-brain situation.
>
> What happens if you just try to stop the filesystem resource (or put the  
> node
> nfs01 on standby)?  Does it stop/unmount cleanly?  Anything in the logs  
> about
> why it can't unmount?  Any processes still accessing the filesystem?
Well, If I do (on nfs01 node):
crm node standby nfs01.local
everything runs fine (resource control passes on nfs02)

and when I'll do:
crm node online nfs01.local the control returns on nfs01.local node.

Could my problems related to centos init.d scripts (I could try using crm  
node offline nfs01.local
in S00cpuspeed /etc/rc3.d/init.d script)?
In this case, is there a distribution more 'in sync' with cluster  
management?

Thanks,
d.
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to