[Users] Re: NFS stall problem
Cédric Jeanneret
cedric.jeanneret at camptocamp.com
Tue Feb 2 01:35:32 EST 2010
Hello,
Hm, in fact lenny includes nfsv4. But we had same troubles with nfsv3
(they were even worth). I'll check the links and keep this thread
updated.
Thank you for your concern.
Best regards,
C.
On Mon, Feb 1, 2010 at 11:35 PM, Gordan Bobic <gordan at bobich.net> wrote:
> NFSv2?! You are joking, right? NFSv3 has been stable for over a decade. Last
> time I saw NFSv2 in the wild was in the '90s.
>
> If you said NFSv4 was unstable, that I might have been inclined to agree
> with.
>
> Gordan
>
> Rafael Ruiz wrote:
>>
>> Hello, NFSv3 is know to have stability issues on linux, so by default it
>> uses NFSv2 wich is more stable but doesn't support files large than 2GB. As
>> I can see you are using NFSv3. I would recommend you to Try using NFSv4, the
>> link
>> http://wiki.linux-nfs.org/wiki/index.php/General_troubleshooting_recommendations
>> would help on setting this up.
>>
>> This would help too
>> http://www.citi.umich.edu/projects/nfsv4/OLS2001/tsld004.htm
>>
>> I hope this help.
>>
>> Best regards,
>>
>>> Hello,
>>>
>>> we're currently using openvz (2.6.18-128.2.1.el5.028stab064.4) on debian
>>> lenny nodes, with debian lenny or etch VE (kernel was alienized from latest
>>> openvz rpm).
>>>
>>> We have an nfs server on a hardware node (so NOT in a VE) which shares
>>> some cartographic datas (means: LOT of small files, aka tiles).
>>>
>>> For now, we're mounting NFS shares in VE in this way :
>>> nfs-server:/path/to/share on /path/to/mount/point type nfs
>>> (ro,nfsvers=3,rsize=32768,wsize=32768,hard,proto=tcp,timeo=600,retrans=2,sec=sys,addr=nfs-server)
>>>
>>> Share is done as follow :
>>> /path/to/share IP(ro,all_squash,anongid=65534,sync,no_subtree_check)
>>>
>>> Problem :
>>> sometimes (well, about once a day, or each two days), nfs share seems to
>>> be stalled. That means :
>>> - unable to access datas from VE
>>> - unable to umount share in VE
>>>
>>> The only thing we can do is a restart of the VE.
>>> As far as I can see, the others VE having an nfs share on the same HN
>>> don't have any problem (well, for a while, than it stalls too).
>>>
>>> We don't have any logs in syslog nor kern.log, either on HN or in VEs.
>>>
>>> NFS-SRV is in a VLAN, and distributes files accross up to 3 VLANS.
>>> firewall rules are correct...
>>>
>>>
>>> Any idea ?
>>>
>>> I've read on wiki that "it's better to mount nfs share on HN, then use a
>>> mount -o bind". We can't afford that, as we need to move VE quiet often, and
>>> this without shutting them down (vzmigrate --online)...
>>>
>>> Thank you in advance.
>>>
>>> Best regards,
>>>
>>> C.
>>> -------------- next part --------------
>>> A non-text attachment was scrubbed...
>>> Name: signature.asc
>>> Type: application/pgp-signature
>>> Size: 197 bytes
>>> Desc: not available
>>> Url :
>>> http://openvz.org/pipermail/users/attachments/20100201/b23258ad/signature-0001.bin
>>>
>>> ------------------------------
>>>
>>> _______________________________________________
>>> Users mailing list
>>> Users at openvz.org
>>> https://openvz.org/mailman/listinfo/users
>>>
>>>
>>> End of Users Digest, Vol 52, Issue 1
>>> ************************************
>>>
>>
>> _______________________________________________
>> Users mailing list
>> Users at openvz.org
>> https://openvz.org/mailman/listinfo/users
>
> _______________________________________________
> Users mailing list
> Users at openvz.org
> https://openvz.org/mailman/listinfo/users
>
More information about the Users
mailing list