I was mistaken...
Genesha wasn't working. Glusters NFS was still on. :-/
A Centos 7 box can, but not XenServer... I think thats the last nail :-/ back to the shelf Gluster goes sorry :-(, it works well for generic file server purposes but now a storage server (or cluster) _must_ handle VM images :-/ and with usable speed. I'll probably test it again in a year or so. I keep coming back since I first tested it years ago (and deployed to a few customers) Gluster is cool. maybe oVirt will have a more viable UI then too. But I am testing Proxmox next so ;-).
Genesha wasn't working. Glusters NFS was still on. :-/
so i fixed that....
XenServer (7.1) can't seem to mount NFS-Genesha, nfs v3 or v4, or using the xe command. (sr_backend_failure_73)
Travis Eddy
Smartware IT - Number 2
On Tue, Apr 4, 2017 at 6:38 PM, Travis Eddy <travis@xxxxxxxxxxxxxxx> wrote:
Hello,
The default (+epel) ones that get searched by yum?
the package is there after you run
yum install centos-release-glusterworks if you run " yum install centos-release-gluster " firstAlso is Ganesha is only working with " -o nfsvers=3 "So Ganesha is working! :-) annndddd its just as slow as the Gluster internal NFS :-((image showing network moving ~31Mb/sec and disk going ~5MB/sec)
So at this point I guess it something in Gluster it self causing the slow down. (again bland kernal NFS goes 100MB+ on the same hardware )Thanks for the help getting Ganesha running :-) :-)
Any advice why Gluster isn't working so well as storage for Virtual hosts?Email - Travis@xxxxxxxxxxxxxxxOn Tue, Apr 4, 2017 at 6:05 PM, Kaleb S. KEITHLEY <kkeithle@xxxxxxxxxx> wrote:
I see glusterfs-ganesha-3.10.0 and -3.10.1 in the Storage SIG repo at https://buildlogs.centos.org/centos/7/storage/x86_64/gluster -3.10/
What repos are you looking at?
On 04/04/2017 02:21 PM, Travis Eddy wrote:
Centos 7.3 fresh installed this morning.
current new this morning problem is the guides want
package: glusterfs-ganesha and it doesn't exist anymore in the repos
Before we would get 'file not found' errors, or permission denied. There
where some differences in how various guides would explain the path,
Pseudo, and volume options. We tried several variations. exporting via
kernal nfs or gluster internal always worked. (selinux is in premissive)
$ sudo yum search ganesha
Loaded plugins: fastestmirror
Loading mirror speeds from cached hostfile
* base: centos.aol.com <http://centos.aol.com>
* epel: mirror.cogentco.com <http://mirror.cogentco.com>
* extras: mirror.rackspace.com <http://mirror.rackspace.com>
* updates: mirror.cs.vt.edu <http://mirror.cs.vt.edu>
====================================== N/S matched: ganesha
=======================================
nfs-ganesha-nullfs.x86_64 : The NFS-GANESHA's NULLFS Stackable FSAL
nfs-ganesha-proxy.x86_64 : The NFS-GANESHA's PROXY FSAL
nfs-ganesha-utils.x86_64 : The NFS-GANESHA's util scripts
nfs-ganesha-vfs.x86_64 : The NFS-GANESHA's VFS FSAL
nfs-ganesha-xfs.x86_64 : The NFS-GANESHA's XFS FSAL
nfs-ganesha.x86_64 : NFS Server running in user space
nfs-ganesha-mount-9P.x86_64 : a 9p mount helper
Name and summary matches only, use "search all" for everything.
your guide is
: http://blog.gluster.org/2015/10/linux-scale-out-nfsv4-using- nfs-ganesha-and-glusterfs-one- step-at-a-time/
? correct? I think i'm going to start a list.
Thanks for your help :-)
Travis Eddy
Smartware IT - Number 2
Cell (texts welcome) - 703.981.5765
Email - Travis@xxxxxxxxxxxxxxx
On Tue, Apr 4, 2017 at 2:09 PM, Kaleb S. KEITHLEY <kkeithle@xxxxxxxxxxGluster-users@xxxxxxxxxxx <mailto:Gluster-users@gluster.<mailto:kkeithle@xxxxxxxxxx>> wrote:
On 04/04/2017 01:54 PM, Travis Eddy wrote:
Thank you for the response.
I'm Pretty sure the problem is with the Gluster's NFS and how
XenServer
interacts with it. As I feel like its not running in async mode.
Everything is at-least 10x faster using bland kernel NFS with
the async
option... 90+MB/sec (saturate gigabit) vs 5~7MB/sec I'm pretty sure
there is a default option in the gluster nfs that needs changing.
I tried the " nfs.trusted-sync off" option.
as well as the "tuned-adm profile rhgs-random-io " (which isn't
aviable
on cent btw)
and using the gluster group virt and the redhat options
Is there another way to ensure the Gluster NFS sync is in async
mode?
Also currently following the Gluster and Centos guides GlusterFS and
Ganesha isn't working.
What exactly didn't work?
E.g. my blog article details setting up nfs-ganesha, from a very
simple non-gluster setup, to a simple setup with gluster, and
gluster with pacemaker HA.
--
Kaleb
_______________________________________________
Gluster-users mailing listorg >
http://lists.gluster.org/mailman/listinfo/gluster-users
<http://lists.gluster.org/mailman/listinfo/gluster-users >
_______________________________________________ Gluster-users mailing list Gluster-users@xxxxxxxxxxx http://lists.gluster.org/mailman/listinfo/gluster-users
