LKML Archive on
help / color / mirror / Atom feed
From: Jaco Kroon <>
Subject: Re: NFS deadlock
Date: Thu, 20 May 2004 10:10:32 +0200	[thread overview]
Message-ID: <> (raw)
In-Reply-To: <>

[-- Attachment #1: Type: text/plain, Size: 5044 bytes --]

Oh, sorry.

The one box at home is running 2.6.5 currently with the intent of 
upgrading to 2.6.6 as soon as I can find the time.  It is using an ext3 
file system underlying.  The same goes to the single client it serves to.

The one at the office that serves up to a hundred or so clients 
currently runs 2.6.4 with a patch for the dpt_i2o driver, it has an ext3 
partition but the one being served up via nfs is reiserfs.  The clients 
are running 2.6.5 at the moment (anything between 0 and 320 clients max 
at any time, usually between 20 and 50 clients depending on lab usage).  
The other server is using 2.4.24 (waiting for the dpt_i2o driver in the 
2.6 kernel) with ext3 file systems and once again reiserfs for the nfs 
exported part of the file system.  Variety of clients in this case, from 
2.4.20 kernels, right through to 2.6.6 kernels.

The machine that died yesterday is also running a 2.6.5 kernel, ext3 
file system.  It's two clients is the first of the two servers above and 
the other runs kernel 2.6.6 as well.

What affects the regulularity of the crashes seems to be the load placed 
on it by clients.  In my case at home the client is considerably faster 
that the server, which will enforce a relatively high load.  I wish I 
had more time to check this out. I'm suspecting some kind of race 
condition that gets triggered by either heavy system load or a heavy 
skew between speeds on the client/server.  I might be totally wrong 
though ...

Transfers in our case is always between linus and linux (at least as far 
as we can control it, we are not aware of any other clients and would 
probably manage to get such a person expelled should we find him).

The client lock-ups we've experienced as well.  It eventually times out 
after a *long* time, we usually bounce the server before that happens. 
This can be explained and is in my oppinion quite normal.

What does sysstat and sar do?  How can I use them to analyse the problem?

Jaco wrote:

>How are your boxes locking up, I have nfs in use every day,
>does rpc die?
>what kernel are you using?
>and are you transfering linux to linux, or to some other platform.
>The only time I had problems was when my client locked up
>because I disconnected the server, and it hung the client,
>the only solution (based on the way I connected), was to reboot.
>To make matters worse, I rean a script that used du every day, and
>so there were 12+ instances of du, all trying to run about.
>I would suggest using a program like sysstat, or sar, to help you
>analyse the issues at hand.
> -sam
>>Hello there
>>I've once again got problems with the kernel locking up.  I'm now
>>convinced that it has something to do with NFS.
>>Previously weve had 2 machines that locked up, plus my one at home,
>>resulting in three machines.  Sometimes they would recover by themselves
>>after some time, other times they could be left for 2 days or so without
>>recovering.  All three of these use NFS to export files to other
>>machines, it's the only thing we can find they have in common, other
>>that x86 architecture, but then other machines would be dying as well.
>>It should be noted that none of these runs on the newest hardware, but
>>that should not matter, neither does any of our other servers.  We have
>>a 3rd NFS server, which doesn't take nearly as heavy load via NFS.  I've
>>been wondering why it hasn't locked up either, and this morning (right
>>now in fact) it has decided that it is it's turn and is currently
>>If anybody else is experiencing similar problems, or have possible work
>>arounds, it would be appreciated if you could share your knowledge.
>>This message and attachments are subject to a disclaimer. Please refer to
>> for full details.
>>Hierdie boodskap en aanhangsels is aan 'n vrywaringsklousule onderhewig.
>>Volledige besonderhede is by
>> beskikbaar.
>To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>the body of a message to
>More majordomo info at
>Please read the FAQ at

"The strength of the Constitution lies entirely in the determination of each
citizen to defend it.  Only if every single citizen feels duty bound to do
his share in this defense are the constitutional rights secure."
-- Albert Einstein
This message and attachments are subject to a disclaimer. Please refer to for full details.
Hierdie boodskap en aanhangsels is aan 'n vrywaringsklousule onderhewig. Volledige besonderhede is by beskikbaar.

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/x-pkcs7-signature, Size: 3174 bytes --]

  reply	other threads:[~2004-05-20  8:10 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2004-05-18 10:15 Jaco Kroon
2004-05-19 16:09 ` samg
2004-05-20  8:10   ` Jaco Kroon [this message]
2004-05-20 14:46     ` Sam Gill

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \ \ \ \ \
    --subject='Re: NFS deadlock' \

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).