Vaguely NFS related problem

Vaguely NFS related problem

[Jeff L. Smith]

We have been running a couple of machines (PIII with Mylex Extreme 2000 
SCSI RAID, RH7.3, 2.4.18 kernel, 64 nfsd's) which only act as NFS 
fileservers.  When either of these servers start getting busy, I start 
seeing a lot of the following messages:

Dec  8 15:27:58 localhost kernel: RPC: Unable to allocate resbuf from cache!

I does not seem to cause any problems (things keep running with no 
obvious corruption) but this surly cannot be good.  Is there something I 
need to tune or a FAQ I could review?

Jeff


-- 
Jeff Smith                                  Atheros Communications, Inc.
Hardware Manager                            529 Almanor Avenue
(408) 773-5257                              Sunnyvale, CA  94086



-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility 
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: Vaguely NFS related problem

[Neil Brown]

On Wednesday December 11, jeff@atheros.com wrote:
> We have been running a couple of machines (PIII with Mylex Extreme 2000 
> SCSI RAID, RH7.3, 2.4.18 kernel, 64 nfsd's) which only act as NFS 
> fileservers.  When either of these servers start getting busy, I start 
> seeing a lot of the following messages:
> 
> Dec  8 15:27:58 localhost kernel: RPC: Unable to allocate resbuf from cache!

This message is not part of any 'vanilla' release kernel.  It must be
a redhat-special.  Ask redhat.

NeilBrown


-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility 
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: Vaguely NFS related problem

[Jeff L. Smith]

I'm running a kernel I built from kernel.org source.  I found the 
message in linux-2.4.18/net/sunrpc/svcsock.c.


printk (KERN_WARNING "RPC: Unable to allocate resbuf from cache!\n");

Jeff

Neil Brown wrote:
> On Wednesday December 11, jeff@atheros.com wrote:
> 
>>We have been running a couple of machines (PIII with Mylex Extreme 2000 
>>SCSI RAID, RH7.3, 2.4.18 kernel, 64 nfsd's) which only act as NFS 
>>fileservers.  When either of these servers start getting busy, I start 
>>seeing a lot of the following messages:
>>
>>Dec  8 15:27:58 localhost kernel: RPC: Unable to allocate resbuf from cache!
> 
> 
> This message is not part of any 'vanilla' release kernel.  It must be
> a redhat-special.  Ask redhat.
> 
> NeilBrown
> 
> 
> -------------------------------------------------------
> This sf.net email is sponsored by:
> With Great Power, Comes Great Responsibility 
> Learn to use your power at OSDN's High Performance Computing Channel
> http://hpc.devchannel.org/
> _______________________________________________
> NFS maillist  -  NFS@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/nfs


-- 
Jeff Smith                                  Atheros Communications, Inc.
Hardware Manager                            529 Almanor Avenue
(408) 773-5257                              Sunnyvale, CA  94086



-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility 
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: Vaguely NFS related problem

[Neil Brown]

On Wednesday December 11, jeff@atheros.com wrote:
> I'm running a kernel I built from kernel.org source.  I found the 
> message in linux-2.4.18/net/sunrpc/svcsock.c.
> 
> 
> printk (KERN_WARNING "RPC: Unable to allocate resbuf from cache!\n");

I think you should check again.  I did and I still cannot find that
message:

% bzcat linux-2.4.18.tar.bz2 | grep 'Unable to allocate resbuf from cache'
%

Are you sure they were *unpatched* kernel.org sources?

I suggest you unroll 2.4.20 and try that.

NeilBrown


-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility 
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: Vaguely NFS related problem

[Jeff L. Smith]

You are correct.  The only patch I applied was Trond's 
linux-2.4.18-NFS_ALL.dif and that is where the check is.  The actual 
patch was linux-2.4.18-svc_tcp.dif.  I'm not doing NFS-over-TPC, but 
this seems to be biting me anyway.

I plan to upgrade to 2.4.20 as soon as I can take the fileservers down 
long enough, but that will be a few weeks.  But then that begs the 
question, should I apply Trond's 2.4.20 patches?

Thanks,
Jeff

Neil Brown wrote:
> On Wednesday December 11, jeff@atheros.com wrote:
> 
>>I'm running a kernel I built from kernel.org source.  I found the 
>>message in linux-2.4.18/net/sunrpc/svcsock.c.
>>
>>
>>printk (KERN_WARNING "RPC: Unable to allocate resbuf from cache!\n");
> 
> 
> I think you should check again.  I did and I still cannot find that
> message:
> 
> % bzcat linux-2.4.18.tar.bz2 | grep 'Unable to allocate resbuf from cache'
> %
> 
> Are you sure they were *unpatched* kernel.org sources?
> 
> I suggest you unroll 2.4.20 and try that.
> 
> NeilBrown


-- 
Jeff Smith                                  Atheros Communications, Inc.
Hardware Manager                            529 Almanor Avenue
(408) 773-5257                              Sunnyvale, CA  94086



-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility 
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: Vaguely NFS related problem

[Trond Myklebust]

>>>>> " " == Jeff L Smith <jeff@atheros.com> writes:

     > I plan to upgrade to 2.4.20 as soon as I can take the
     > fileservers down long enough, but that will be a few weeks.
     > But then that begs the question, should I apply Trond's 2.4.20
     > patches?

FYI: There are no further NFS server-related patches in my 2.4.20
patchsets.

The only server patches I included earlier were beta-versions of the
NFS over TCP related stuff ('cos they conflicted with some of the
client changes). Now that the finalized TCP code has been merged into
the mainstream kernel by Neil, I've dropped them.

Cheers,
  Trond


-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility 
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

Re: Vaguely NFS related problem

[Jeff L. Smith]

I should also add that I realize this an RPC problem, not and NFS 
problem (hence vaguely NFS related), but this is being tickled by the 
NFS server so I was hoping someone else had seen this.

Jeff


-------- Original Message --------
Subject: Re: [NFS] Vaguely NFS related problem
Date: Thu, 12 Dec 2002 13:12:34 +1100
From: Neil Brown <neilb@cse.unsw.edu.au>
To: "Jeff L. Smith" <jeff@atheros.com>
CC: nfs@lists.sourceforge.net
References: <3DF7EA94.20108@atheros.com>

On Wednesday December 11, jeff@atheros.com wrote:
 > We have been running a couple of machines (PIII with Mylex Extreme 2000
 > SCSI RAID, RH7.3, 2.4.18 kernel, 64 nfsd's) which only act as NFS
 > fileservers.  When either of these servers start getting busy, I start
 > seeing a lot of the following messages:
 >
 > Dec  8 15:27:58 localhost kernel: RPC: Unable to allocate resbuf from 
cache!

This message is not part of any 'vanilla' release kernel.  It must be
a redhat-special.  Ask redhat.

NeilBrown


-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

-- 
Jeff Smith                                  Atheros Communications, Inc.
Hardware Manager                            529 Almanor Avenue
(408) 773-5257                              Sunnyvale, CA  94086



-------------------------------------------------------
This sf.net email is sponsored by:
With Great Power, Comes Great Responsibility 
Learn to use your power at OSDN's High Performance Computing Channel
http://hpc.devchannel.org/
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs