[Linux-cluster] Restarting GFS2 without reboot
Digimer
lists at alteeve.ca
Tue Nov 26 15:42:14 UTC 2013
On 26/11/13 07:41, Vladimir Melnik wrote:
> On Tue, Nov 26, 2013 at 12:17:14PM +0000, Steven Whitehouse wrote:
>>> Okay, one node has gone down, but why the second node can't keep working with
>>> the filesystem? :-( That's what surprises and scares me at the same time.
>> Well the second node will need to ensure quorum, so you should have a
>> two node set up configured. That will require some kind of tie-breaker
>> so I'm guessing that you are using qdisk for that? This is why it would
>> help if you posted your config, as otherwise I'm left guessing,
>
> I'm using corosync instead of qdiskd. And here's my cluser.conf, it
> looks really simple:
>
> <?xml version="1.0"?>
> <cluster name="ckvm1_pod1" config_version="5">
> <clusternodes>
> <clusternode name="***.host1.ckvm1.***" votes="1" nodeid="1">
> <fence>
> <method name="single">
> <device name="host1_ipmi"/>
> </method>
> </fence>
> </clusternode>
> <clusternode name="***.host2.ckvm1.***" votes="1" nodeid="2">
> <fence>
> <method name="single">
> <device name="host2_ipmi"/>
> </method>
> </fence>
> </clusternode>
> </clusternodes>
> <fencedevices>
> <fencedevice name="host1_ipmi" agent="fence_ipmilan" ipaddr="***" login="***" passwd="***"/>
> <fencedevice name="host2_ipmi" agent="fence_ipmilan" ipaddr="***" login="***" passwd="***"/>
> </fencedevices>
> <rm>
> <failoverdomains/>
> <resources/>
> </rm>
> </cluster>
Stephen was right, you don't have qdisk and you haven't configured for
2-node configuration.
Either add a qdisk or disable quorum with:
<cman expected_votes="1" two_node="1"/>
You have a SAN, so using qdisk is the better idea.
--
Digimer
Papers and Projects: https://alteeve.ca/w/
What if the cure for cancer is trapped in the mind of a person without
access to education?
More information about the Linux-cluster
mailing list