RE: Which lock manager to use on a 2-node GFS Cluster?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi,

Thanks Riaan/Bob for the explanations. It helps me a lot and thanks to
Ceslo, you asked what I finally wanted.

Thanks and regards

- Hirantha

> -----Original Message-----
> From: linux-cluster-bounces@xxxxxxxxxx [mailto:linux-cluster-
> bounces@xxxxxxxxxx] On Behalf Of Celso K. Webber
> Sent: Thursday, August 17, 2006 12:48 AM
> To: linux clustering
> Subject: Re:  Which lock manager to use on a 2-node GFS
> Cluster?
> 
> Hi Riaan,
> 
> Thank for your excelent explanation!
> 
> Now I know I'll have to run an Oracle unsupported environment ... :)
> 
> Would someone please confirm if the environment suggested by Hirantha is
> supported by Red Hat?
> 
> * 2 GFS/RAC nodes;
> * 2 DLM on RAC/GFS nodes;
> * fibre channel external storage.
> 
> The above environment is what we usually encounter on our cluster
> projects. Sometimes Oracle Database is installed as active-passive (not
> a RAC Cluster). It would be useful to know if Red Hat supports it, even
> if Oracle does not.
> 
> Last question: does Oracle supports a 2-node cluster using DLM for
> Oracle Database (no RAC, no GFS) when used with RHCS?
> 
> Thanks again!
> 
> Celso.
> 
> 
> Riaan van Niekerk escreveu:
> > hi Hirantha
> >
> > short answer: "No"
> >
> > If you are willing to settle for "should work just fine" instead of
> > "certified by Oracle" you can try running DLM on 2 RAC/GFS nodes, then
> > "yes"
> >
> > To have your solution certified/supported by Oracle, for a minimal
> > configuration you need 5 machines:
> > a) 2 GFS/RAC nodes
> > b) 3 external GULM lock servers
> >
> > We also found these system requirements somewhat excessive (having more
> > lock servers than actual GFS-accessing cluster nodes).
> >
> > more notes:
> > - supported GFS with RAC is GULM only - DLM is not supported by Oracle,
> > because it is embedded by default.
> >
> > - external lock servers - that means the the GULM service cannot run on
> > the nodes accessing the GFS. The 3 GULM nodes and the 2 GFS/RAC nodes
> > must be completely separate
> >
> > - the reason for the separation of lock servers from GFS/RAC nodes is
> > that Oracle RAC is designed to be able to with a single node up, whereas
> > DLM  needs a quorum of 50%+1 to operate. Also, external lock servers
> > allow you to reboot any RAC node without causing the lock server to
> switch.
> >
> > - specs for the locks servers are as entry-level as you can get for
> > modern server hardware, e.g. 1CPU, 1GB RAM.
> >
> > - Red Hat Knowledge base article for RAC on GFS:
> > http://kbase.redhat.com/faq/FAQ_78_3853.shtm
> > It mentions GFS 6.0, (slightly out of date) but is just as relevant for
> > GFS 6.1 / RHEL 4). According to this article, "Red Hat does not charge
> > for Red Hat GFS nodes that are used only for the external lock server
> > configuration"
> >
> > - the Cluster/GFS FAQ has some URLs for RAC on GFS:
> > http://sources.redhat.com/cluster/faq.html#gfs_oraclerac
> >
> > - MetaLink Note  329530.1 is the definitive source for what exactly is
> > supported by Oracle, w.r.t. RAC and GFS. Unfortunately it is only
> > available via Metalink.
> >
> > Celso - answers to your questions below
> ...
> --
> *Celso Kopp Webber*
> 
> celso@xxxxxxxxxxxxxxxx <mailto:celso@xxxxxxxxxxxxxxxx>
> 
> *Webbertek - Opensource Knowledge*
> (41) 8813-1919
> (41) 3284-3035
> 
> --
> Linux-cluster mailing list
> Linux-cluster@xxxxxxxxxx
> https://www.redhat.com/mailman/listinfo/linux-cluster

--
Linux-cluster mailing list
Linux-cluster@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-cluster

[Index of Archives]     [Corosync Cluster Engine]     [GFS]     [Linux Virtualization]     [Centos Virtualization]     [Centos]     [Linux RAID]     [Fedora Users]     [Fedora SELinux]     [Big List of Linux Books]     [Yosemite Camping]

  Powered by Linux