On Thu, Jun 03, 2010 at 10:08:05AM +0200, RaSca wrote:
> Il giorno Mer 02 Giu 2010 15:10:14 CET, Florian Haas ha scritto:
> [...]
> >... and then mounted the exported /mnt directory on an Ubuntu lucid box.
> >Then, in two parallel sessions, I started jobs that wrote to and read
> >from the NFS mount. On the NFS cluster, I sent one node into standby and
> >watched failover go completely flawlessly. The jobs continued to read
> >from/write to NFS with an interruption on the order of seconds.
> >Here are my NFS mount options, in case you're interested:
> >192.168.122.110:/mnt on /tmp/nfsmount type nfs
> >(rw,relatime,vers=3,rsize=65536,wsize=65536,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,mountaddr=192.168.122.110,mountvers=3,mountproto=tcp,addr=192.168.122.110)
> >So, I'll file this under "works for me". If it doesn't work for you,
> >please collect an hb_report.
> >Cheers,
> >Florian
> 
> Hi Florian,
> I followed your suggestions, but the problem is the same.
> There is only one difference between our architectures and is that
> you're mounting an iscsi share while I'm using drbd. The rest of the
> configuration is identical: I've mounted with "noatime" the ext3
> filesystem and from the client the nfs export is mounted with the
> same your options.
> The point is: is this an exportfs bug? It is a drbd related issue?

No, and no.

> Those seems to be the differences between this solution and the one
> that you (and also the creator of the exportfs RA) implemented with
> success.

I think the difference is that your client workload includes locking,
and theirs do not.

Please try with "nolock" on the client side,
and see if that changes something.

-- 
: Lars Ellenberg
: LINBIT | Your Way to High Availability
: DRBD/HA support and consulting http://www.linbit.com

DRBD® and LINBIT® are registered trademarks of LINBIT, Austria.
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to