[Pacemaker] 2 Node Clustering, when primary server goes down(shutdown) the secondary server restarts

kamal kishi kamal.kishi at gmail.com
Tue Oct 28 06:24:54 UTC 2014


Hi,

 I know, no fencing configuration creates issue.
But the current scenario is due to fencing??
The syslog isn't revealing much about the same.
I would love to configure fencing but currently need some solution to
overcome the current scenario, if you say fencing is the only solution then
I might have to do it remotely.

OS -> UBUNTU 12.04 (64 bits)
DRBD -> 8.3.11

Thanks for the quick reply

On Tue, Oct 28, 2014 at 11:19 AM, Digimer <lists at alteeve.ca> wrote:

> On 28/10/14 01:39 AM, kamal kishi wrote:
>
>> Hi all,
>>
>>        Facing a strange issue which I'm not able to resolve as I'm not
>> sure where what is going wrong as the logs is not giving away much to my
>> knowledge.
>>
>> Issue -
>> Have configured 2 Node Clustering, have attached the configuration
>> file(New CRM conf of BIC.txt).
>>
>> If Server2 which is primary is shutdown(forcefully by turning off the
>> switch), Server1 restarts within few seconds and starts the resources.
>> Even though the Server1 restarts and starts the resources the time taken
>> to recover is too long to convince the clients and the current working
>> is erroneous is what I feel.
>>
>> Have attached the syslog with this mail.(syslog)
>>
>> Do go through the same and let know a solution to resolve the same as
>> the setup is in clients place.
>>
>> --
>> Regards,
>> Kamal Kishore B V
>>
>
> You really need fencing, first and foremost. This will cause the survivor
> to put the lost node into a known state and then safely begin taking over
> lost services. Do your nodes have IPMI (or iRMC, iLO, DRAC, etc)? If so,
> setting up stonith is easy.
>
> Once it is setup, configure DRBD to use the fence-handler
> 'crm-fence-peer.sh' and change the fencing policy to
> 'resource-and-stonith'. Without this, you will get split-brains and
> fail-over will be unpredictable.
>
> Once stonith is configured and tested in pacemaker and you've hooked
> DRBD's fencing into pacemaker, see if you problem remains. If it does, on
> both nodes, run: 'tail -f -n 0 /var/log/messages', kill a node and wait for
> things to settle down. Share the log output here.
>
> Please also tell us your OS, pacemaker, drbd and corosync versions.
>
> --
> 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?
>
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org
>



-- 
Regards,
Kamal Kishore B V
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20141028/942935a5/attachment.htm>


More information about the Pacemaker mailing list