automount failures when many clients try to mount in a short period

[shilong han]

Hi,

I have the same problem, and I increased the number of NFS to 64, but it
happend intermittent.

Could you please send me the result/solution on this topic?


http://lists.us.dell.com/pipermail/linux-poweredge/2005-April/020200.html

Buchan,

We increased the number of nfs threads to 32 since it was warning about
too many tcp connections before. We also export the filesystems with the
insecure option to permit higher port numbers. The clients are mounting
with the tcp option. Otherwise I think the rest should be very much
default settings.

I don't know all the details about the mounting procedure, but to me it
looks like rpc.mountd failing rather than nfsd. It could still be
related of course.

/Olle

Buchan Milne wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Olle Liljenzin wrote:
>
>>We have a new PE2850 running as nfs server for slightly above 100 active
>>clients.
>>
>>Typically 1-5 clients a day will sporadically fail to automount a
>>filesystem. The pattern we can see is that this happens when several
clients
>>try to mount at the same point in time. At the next try the mount will
>>succeed, but then the running processes are already crashed.
>>
>>The clients report the following message, not very informative, when a
mount
>>request fails:
>>
>>"Apr 14 20:57:34 townsville automount[28137]: failed to mount
>>/nfs/vm/master"
>>
>>The server doesn't write anything at all in the message file when the
mounts
>>fail.
>>
>>The server is running RHEL4 and has a large number of disks, fast
processors
>>and a lot memory.
>
>
> How may rpc.nfsd threads are you starting?
>
>
>>The network is switched gigabit ethernet and usually no
>>packages are lost. Load doesn't look like an issue. Most clients run
RHEL3.
>>
>>I have no clue why this happens or what I could do to fix the problem. Any
>>ideas are wellcome.
>
>
> Have you done any tuning of the NFS threads etc (see the nfs init script
> for variables you can set in /etc/sysconfig/nfs).
>
> You probably want to set RPCNFSDCOUNT a bit higher than the default of 8
...
>
> Regards,
> Buchan
>
> - --
> Buchan Milne                      Senior Support Technician
> Obsidian Systems                  http://www.obsidian.co.za
> B.Eng          RHCE (803004789010797),LPIC-1 (LPI000074592)
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.0 (GNU/Linux)
> Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
>
> iD8DBQFCX2YErJK6UGDSBKcRAoiHAJ4rlEGTd9fgeHjwB04dQ5ltMjTlnQCdHu0h
> dd39JDm2GSbx9hsA5u7Ujjw=
> =H/Y2
> -----END PGP SIGNATURE-----
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.us.dell.com/pipermail/linux-poweredge/attachments/20081110/c97d0e1e/attachment.htm