[Linux-cluster] Cluster stability with missing qdisk

emmanuel segura emi2fast at gmail.com
Fri Feb 10 17:00:53 UTC 2012


I understand what you say and that's right

One solution can be play with qdisk cluster timeout

man qdisk for more info

2012/2/10 Jan Huijsmans <Jan.Huijsmans at interaccess.nl>

> We're using it on multipath, it's a failure to write to the device that's
> killing the cluster.
> All other devices work, so the cluster can function as it should, would it
> be for the reboot by the cluster software.
>
> I would like to prevent the cluster rebooting nodes just because the qdisk
> isn't responding (due to slow storage, failure on the quorum location such
> as powerloss and other, non application related, errors)
>
> When both nodes are up and the application is able to run, there should be
> no reboot in my opinion.
>
> With regards,
> Inter Access BV
>
> ing. J.C.M. (Jan) Huijsmans
> Designer / Technical Consultant UNIX & Linux
> Infrastructure Professional Services UNIX
>
> E-mail:  jan.huijsmans at interaccess.nl
>
> Tel:     035 688 8266
> Mob.:    06 4938 8145
>
> Hoofdkantoor:
> Colosseum 9, 1213 NN Hilversum
> Postbus 840, 1200 AV Hilversum
> K.v.K. Hilversum 32032877
> ________________________________________
> From: linux-cluster-bounces at redhat.com [linux-cluster-bounces at redhat.com]
> On Behalf Of emmanuel segura [emi2fast at gmail.com]
> Sent: 10 February 2012 15:16
> To: linux clustering
> Subject: Re: [Linux-cluster] Cluster stability with missing qdisk
>
> Why you don't use qdisk on multipath, than can resolv your problem
>
> 2012/2/10 Jan Huijsmans <Jan.Huijsmans at interaccess.nl<mailto:
> Jan.Huijsmans at interaccess.nl>>
> Hello,
>
> In the clusters we have we use a qdisk to determine which node had the
> quorum, in case of a split brain situation.
>
> This is working great... until the qdisk itself is hit due to problems
> with the SAN. Is there a way to have a stable cluster,
> with qdisks, where the absence of (1) qdisk won't kill the cluster all
> together. At this moment, with the setup with 1 qdisk,
> the cluster is totally depending on the availability of the qdisk, while,
> IMHO, it should be expendable.
>
> We have now a triangle setup, with 2 data centers and 1 extra 'quorum'
> location for the IBM SAN. In the SAN setup there
> are 3 quorum devices, 1 in each data center and the 3rd on the quorum
> location. When 1 location fails (one of the data
> centers or the quorum location) it is still up and running.
>
> Is it possible to copy this setup and use 3 qdisks, so when 1 qdisk fails
> the cluster stays alive? I would set the vote
> value of all components (systems and qdisks) to 1, so the cluster would
> keep running with 2 systems and 1 qdisk
> or 1 system with 2 qdisks. (it'll be dead with only 3 qdisks, as the
> software will die with both systems ;) )
>
> I've heard of setup's with 3 systems, where the 3rd was just for the
> quorum, so you this one can die, but on this
> occasion it won't help us, as there are no systems on the 3rd location.
> (and it's not supported by Red Hat, when I'm
> correctly informed)
>
> With regards,
> Inter Access BV
>
> ing. J.C.M. (Jan) Huijsmans
> Designer / Technical Consultant UNIX & Linux
> Infrastructure Professional Services UNIX
>
> E-mail:  jan.huijsmans at interaccess.nl<mailto:jan.huijsmans at interaccess.nl>
>
> Tel:     035 688 8266<tel:035%20688%208266>
> Mob.:    06 4938 8145<tel:06%204938%208145>
>
> Hoofdkantoor:
> Colosseum 9, 1213 NN Hilversum
> Postbus 840, 1200 AV Hilversum
> K.v.K. Hilversum 32032877
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com<mailto:Linux-cluster at redhat.com>
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
>
> --
> esta es mi vida e me la vivo hasta que dios quiera
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>



-- 
esta es mi vida e me la vivo hasta que dios quiera
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20120210/9cadff5f/attachment.htm>


More information about the Linux-cluster mailing list