[PVE-User] GlusterFS Performance tweak

Gilberto Nunes gilberto.nunes32 at gmail.com
Tue Jul 5 16:58:35 CEST 2016

Well... I am not a gluster expertise, but I add this parameters in
/etc/glusterfs/glusterd.vol in hope to improve performance:

option performance.flush-behind on
option performance.strict-write-ordering on
option performance.strict-o-direct on
option performance.force-readdirp on
option performance.low-prio-threads 64
option performance.least-prio-threads 64
option performance.enable-least-priority on
option performance.least-rate-limit 10
option performance.high-prio-thread 64
option performance.normal-prio-threads 64
option performance.nfs.write-behind on
option performance.nfs.read-ahead on
option performance.nfs.io-cache on
option performance.nfs.quick-read on
option performance.nfs.stat-prefetch on
option performance.nfs.io-threads on
option performance.write-behind-window-size 100MB
option performance.cache-refresh-timeout 10
option performance.cache-size 1GB
option cluster.stripe-block-size 512KB
option performance.io-thread-count 32

Perhaps this can help me in any way....
I just wanna some advice...


2016-07-05 11:19 GMT-03:00 Gilberto Nunes <gilberto.nunes32 at gmail.com>:

> Hello list
> I have PVE 4.2 and a server act as a GlusterFS storage...
> I have just one VM, a KVM Ubuntu machine.
> The VM itself works fine.
> This VM is our Zimbra Mail Server.
> The VM image reside in the glusterfs server, seting up without any
> performance tunning, just the defaults options.
> But, now, after to migrate amost 1 000 mail account from out old postfix
> based server, I encounter some delay when use Zimbra via webclient (Nginx).
> Is there some tweak performance that I can do in gluster server or in PVE
> server (storage.cfg) to improve write data??
> Thanks for any advice...
> --
> Gilberto Ferreira
> Skype: gilberto.nunes36


Gilberto Ferreira
+55 (47) 9676-7530
Skype: gilberto.nunes36

More information about the pve-user mailing list