Brian,
thanks for the input. I'll have to figure out exactly how to
perfomance test it. Fortunately, the environment we are going to be
using this in serves a fixed number of servers (at least for the NFS)
that is around 150 or so. None of the data provided by our current
solution is high performance, but I will want to make sure this at least
keeps pace with that solution.
Thanks,
Kevin
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Brian Long
Sent: Monday, June 09, 2008 11:47 AM
To: Red Hat Enterprise Linux 5 (Tikanga) discussion mailing-list
Subject: RE: [rhelv5-list] NFS / CIFS cluster question
Kevin,
I would make sure to do extensive performance testing emulating hundreds
or thousands of clients before you choose one or the other. I've heard
GFS has had issues when compared to standard single-node filesystems
like ext3 so you might consider verifying the cluster FS will meet your
needs.
/Brian/
On Mon, 2008-06-09 at 09:21 -0700, Collins, Kevin [Beeline] wrote:
> Thanks... I googled and played around a bit further after sending the
> original message. One thing I found was that if I use GFS and manage
> my own /etc/exports (which I prefer to do), then the clustering is
> easy as I really only have to have a clustered resource for a virtual
> IP address.
>
> Kevin
>
>
> ______________________________________________________________________
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] On Behalf Of Zavodsky, Daniel
> (GE Money)
> Sent: Monday, June 09, 2008 12:52 AM
> To: Red Hat Enterprise Linux 5 (Tikanga) discussion mailing-list
> Subject: RE: [rhelv5-list] NFS / CIFS cluster question
>
>
>
> If you want to use the cluster in an active/passive configuration,
> then you can even use normal non-clustered LVM, you just have to
> set-up the cluster accordingly. Or like you said, use CLVM, it is
> easier to maintain and safer (less prone to admin errors) in a
> clustered environment.
>
> Regards,
> Daniel
>
>
>
> ______________________________________________________________________
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] On Behalf Of Collins, Kevin
> [Beeline]
> Sent: Friday, June 06, 2008 10:54 PM
> To: [email protected]
> Subject: [rhelv5-list] NFS / CIFS cluster question
>
>
>
> I am preparing to build a 2-node cluster to server NFS and CIFS
> (Samba) to a farm of servers (primarily non-Linux). I have the systems
> clustered, and CLVM is working fine on my SAN disks.
>
> Now it occurs to me that I have 2 ways to do the shared filesystems:
>
> 1) Use GFS (which I have no experience with), which allows me to have
> simultaneous read/write access on all cluster nodes (not really a
> requirement)
>
> 2) Use CLVM, and let the cluster manage who can write to the
> filesystems (i.e., which node they get mounted on, exported from, etc)
>
> Since I am implementing this as an HA cluster, not a load-balancer, do
> I really care about GFS? Is it simpler to set up an NFS/CIFS cluster
> with GFS? Only the active node has any reason to ever have write
> access to the disks.
>
> Can someone enlighten me, please? I'm open to opinions, especially
> fromanyone with experience in doing the same thing.
>
> Thanks,
>
> Kevin
>
> _______________________________________________
> rhelv5-list mailing list
> [email protected]
> https://www.redhat.com/mailman/listinfo/rhelv5-list
--
Brian Long | |
. | | | . | | | .
' '
C I S C O
_______________________________________________
rhelv5-list mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/rhelv5-list
_______________________________________________
rhelv5-list mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/rhelv5-list
