Ceph fs stability

Ceph fs stability

[Niko!]

Hi!

we are using ceph 0.48 on three nodes to provide rbd images for further 
four kvm nodes (not kernel mapped) with no big issues and would mount 
the ceph fs on the kvm nodes just to store xml virtual machine 
definitions in order to have them immediately in the case an host 
crashes (we are in multimds configuration). As ceph fs is not production 
ready what are the possible problems? May the fs corrupt the rbd pool or 
damages will be limited only to the data/metadata pools (acceptable for 
us)? may the fs hang the entire cluster or kvm client nodes?

Regards.

     Niko

Re: Ceph fs stability

[Josh Durgin]

On 08/02/2012 10:10 AM, Niko! wrote:
> Hi!
>
> we are using ceph 0.48 on three nodes to provide rbd images for further
> four kvm nodes (not kernel mapped) with no big issues and would mount
> the ceph fs on the kvm nodes just to store xml virtual machine
> definitions in order to have them immediately in the case an host
> crashes (we are in multimds configuration). As ceph fs is not production
> ready what are the possible problems? May the fs corrupt the rbd pool or
> damages will be limited only to the data/metadata pools (acceptable for
> us)? may the fs hang the entire cluster or kvm client nodes?
>
> Regards.
>
>      Niko

The fs won't corrupt other pools or make them inaccessible, but if you
use the kernel client it could potentially lock up the node. If you
really want the fs interface, you can limit the failure further by
mounting with fuse instead of the kernel ceph module.

Since you're just storing some some vm definitions, why not use rados
objects directly, e.g. with rados get/put to retrieve and store them?

Josh

Re: Ceph fs stability

[Smart Weblications GmbH - Florian Wiessner]

Am 02.08.2012 19:10, schrieb Niko!:
> Hi!
> 
> we are using ceph 0.48 on three nodes to provide rbd images for further four kvm
> nodes (not kernel mapped) with no big issues and would mount the ceph fs on the
> kvm nodes just to store xml virtual machine definitions in order to have them
> immediately in the case an host crashes (we are in multimds configuration). As
> ceph fs is not production ready what are the possible problems? May the fs
> corrupt the rbd pool or damages will be limited only to the data/metadata pools
> (acceptable for us)? may the fs hang the entire cluster or kvm client nodes?
> 



Don't use it for now, it is too buggy - you sometimes cannot delete files or
rename them, or the fuse client crashes..


what we do instead is use a mapped rbd device with ocfs2 - works stable.

-- 

Mit freundlichen Grüßen,

Florian Wiessner

Smart Weblications GmbH
Martinsberger Str. 1
D-95119 Naila

fon.: +49 9282 9638 200
fax.: +49 9282 9638 205
24/7: +49 900 144 000 00 - 0,99 EUR/Min*
http://www.smart-weblications.de

--
Sitz der Gesellschaft: Naila
Geschäftsführer: Florian Wiessner
HRB-Nr.: HRB 3840 Amtsgericht Hof
*aus dem dt. Festnetz, ggf. abweichende Preise aus dem Mobilfunknetz
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html