nfs caching issue

Tue May 27 11:05:25 EDT 2008

Hello,

My application is running on debian vserver and is triggering conversion 
of file (stored on other storage server) in an incremental way. For 
this, it is looking the previous file size (from last conversion, stored 
in an "offset" file and compare it with actual size. From 
investigations, we see randomly in time some old values (see below on 
left you can see the time we checked and then the size of the file with 
ls system call). The stupid value from ls seems to be an old file system 
descriptor (file descriptor from 2:39 when the check is done at 12:5 or 
later. At a certain moment, it jumps to another value (see below, then 
it moves to file descriptor of 20:38).

So it looks a cache issue of NFS (provide old file descriptor). Weird 
thing is that between 2 checks (like below first error at 12:5 and then 
12:32), we have a lot of checks that are ok (so either not using the 
caching, either another cache) and then error pops up again with same 
"stupid" value.

This occurs on lot of different files and when an error occurs, it is on 
several files at same time (during few seconds), which would give a hint 
like: server overloaded=> use NFS cache that is corrupted.

Does someone has an idea what could it be?

Could you redirect me if I'm posting something on wrong place?

21.5.2008 12:5:17 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 34622013 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 12:32:16 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 34950014 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 12:58:41 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 35258560 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 13:9:49 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 35258560 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 13:25:52 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 35572550 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 14:13:26 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 36165993 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 14:40:32 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 36470064 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 15:7:42 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 36771739 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 15:34:14 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 37094016 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 16:1:39 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 37412906 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 16:28:32 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 37740376 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 17:7:10 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 38064941 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 17:28:55 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 38401762 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 17:55:56 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 38761058 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 19:23:51 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
28440786, but higher offset 39759058 | ls=-rw-r--r--  1 root users 
28440786 2008-05-21 02:39 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 20:59:42 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
40800051, but higher offset 40870081 | ls=-rw-r--r--  1 root users 
40800051 2008-05-21 20:38 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 21:55:30 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
40800051, but higher offset 41554014 | ls=-rw-r--r--  1 root users 
40800051 2008-05-21 20:38 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 23:1:16 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
40800051, but higher offset 42271328 | ls=-rw-r--r--  1 root users 
40800051 2008-05-21 20:38 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
21.5.2008 23:36:37 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
40800051, but higher offset 42681325 | ls=-rw-r--r--  1 root users 
40800051 2008-05-21 20:38 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
22.5.2008 1:22:16 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
43246993, but higher offset 43902409 | ls=-rw-r--r--  1 root users 
43246993 2008-05-22 00:07 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
22.5.2008 2:18:29 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
43246993, but higher offset 44514934 | ls=-rw-r--r--  1 root users 
43246993 2008-05-22 00:07 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat
22.5.2008 3:7:52 : 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat has size 
43246993, but higher offset 45117137 | ls=-rw-r--r--  1 root users 
43246993 2008-05-22 00:07 
/mnt/linuxtesters/testlog/12117CA.B76872/PROB1.001.22.dat

-- 
MOREL Jerome
Xfab coordinator
Phone: +32 57 22 61 27 (127)
Mobile: +32 485 88 94 20
e-mail: jem at melexis.com
www.melexis.com <http://www.melexis.com>

Think about the environment, before printing this message!

*Disclaimer*
"The contents of this e-mail, including any attachments, are strictly 
confidential and are intended solely for the addressee. If you are not 
the intended addressee, or have otherwise received this e-mail in error, 
please immediately inform the sender by return e-mail and delete this 
e-mail, including any attachments. Any copying, distribution or 
dissemination of this e-mail and its attachments for any purpose is 
strictly prohibited. The author of this e-mail has taken every effort to 
ensure that the information provided is accurate and up-to-date, based 
upon the information available at the time of writing. MELEXIS accepts 
no liability for any loss or damage, howsoever caused, due to errors 
contained in this e-mail, nor for any loss or damage due to the timing 
of its transmission or receipt. While every precaution is taken to 
ensure that e-mail messages and attachments are free of computer 
viruses, MELEXIS accepts no liability for any loss or damage caused by 
the presence of a virus."

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://linux-nfs.org/pipermail/nfsv4/attachments/20080527/d37fd55c/attachment.htm 