[Gluster-users] NFS client blocking reboot

Anand Avati anand.avati at gmail.com
Sun Jul 28 17:16:14 UTC 2013

On Sun, Jul 28, 2013 at 9:57 AM, Marcus Bointon
<marcus at synchromedia.co.uk>wrote:

> On 28 Jul 2013, at 18:12, Anand Avati <anand.avati at gmail.com> wrote:
> What is your typical workload, and what kind of tests did you compare
> native client perf against NFS perf?
> Low load, two web servers sharing a content area of < 500M. Even with a
> single client, performance with native client is slow, i.e. more than about
> 1 request/sec for a 100k image was enough to make the web server saturate
> iowait. NFS performance is not amazing, but it's definitely less bad (and
> there are lots of references in the list archives on NFS being better for
> small-file performance). Local disk only is orders of magnitude faster.
> I've not tested whether the performance has improved under 3.4, but frankly
> having it actually work reliably is more important.

You might want to give the native client another shot by setting "option
max-file-size 128KB" in the quick-read section of the client volfile in
/var/lib/glusterd/<volname>/*fuse*.vol (there will be two). Unfotunately
this is not settable through the CLI.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20130728/5fe3fdb6/attachment.html>

More information about the Gluster-users mailing list