Le 24/03/2011 16:03, Windsor Dave L. (AdP/TEF7.1) a écrit :
Hello Everyone,
I recently installed CentOS 5.5 x86_64 on a brand new ProLiant DL380 G7. I have
identical OS software running reock-solid on two other DL380 ProLiant servers, but they
are G6 models, not G7. On the G7, the installation went perfectly and the machine ran
great for about 2 weeks, when it just seemed to "stop". The system stopped
responding on the network, and there was no video on the console (or remote console via
iLO). It would not reboot or cold boot through iLO, I actually had to hold the power to
turn it off and then hit it again to power up.
This happened several times within a few days of each other. Each time, there
was no evidence in any logs of a problem - the system just seemed to stop or
lock up. We did have a CPU problem light appear on the front, so HP came in
and replaced the one 4-core CPU. Since then, it has run as long as two weeks,
but still crashes randomly. After the last reboot, I left the console in text
mode on vt1, and when it crashed again this morning this was displayed on the
screen:
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: ffff8100dc435cf0 CR3: 000000008a6ca000 CR4: 00000000000006e0
Process smbd (pid: 18970, threadinfo ffff81001529e000, task ffff81011f5347a0)
Stack: ffff81011e4e71c0 0000000000000000 ffff8100cf12a015 ffffffff80009c41
ffff81011e4e71c0 0000000100000000 000000030027ea9d ffff8100cf12a011
ffff81011e4e71c0 ffff81010d9cf300 ffff81011e4e71c0 ffff8101044099c0
Call Trace:
[<ffffffff80009c41>] __link_path_walk+0x3a6/0xf5b
[<ffffffff8000ea4b>] link_path_walk+0x42/0xb2
[<ffffffff8000cd72>] do_path_lookup+0x275/0x2f1
[<ffffffff80012851>] getname+0x15b/0x1c2
[<ffffffff800239d1>] __user_walk_fd+0x37/0x4c
[<ffffffff80028905>] vfs_stat_fd+0x1b/0x4a
[<ffffffff80039fa2>] fcntl_setlk+0x243/0x273
[<ffffffff80023703>] sys_newstat+0x19/0x31
[<ffffffff8005d229>] tracesys+0x71/0xe0
[<ffffffff8005d28d>] tracesys+0xd5/0xe0
Code: 00 00 00 00 00 00 00 00 70 4d 4f 9d 00 81 ff ff 98 e4 4b dc
RIP [<ffff8100dc435cf0>]
RSP<ffff81001529fd18>
CR2: ffff8100dc435cf0
<0>Kernel panic - not syncing: Fatal exception
This suggests that something happened in a Samba process. I have the Samba3x
packages installed since we are beginning to introduce Win7 clients into our
environment.
Googling "Kernel panic - not syncing: Fatal exception" and "CentOS" produced
many hits, but nothing that seemed to exactly match my problem. Since this is the only G7 server I
have here right now, I can't reproduce the problem on another machine. The G6s I have running the
identical version of CentOS have no problems.
I am trying to determine if this is pointing to a hardware or software issue.
Some of the Google results suggested using a Centosplus kernel - is this a good
idea?
The server is a HP DL380 G7 Server with 4 GB RAM (1 DIMM 1333 MHz), one 4-core CPU (2133
MHz), 4 built-in Broadcom "NetExtreme II BCM5709 II Gigabit Ethernet" NICs, and
a P410 Smart Array Controller. The P410 and the system BIOS have both been updated to
the latest levels to see if that fixes the crashes, with no change.
Any idea where I should look next?
Thanks for any help anyone can provide!
The fact that it appears after two weeks or so reminds me of a bug I saw
on linux PowerEdge mailing list, //the "blocked for more than 120
seconds" timeout bug.
I don't know if your problem is related, but if it is the case you
should see the message in your logs.
Do you have any high IO load, at least at some moments, on your server ?
See :
http://lists.us.dell.com/pipermail/linux-poweredge/2011-March/044515.html
In this case, using a newer kernel would be indeed it seems a good idea.
See if it can help...
Alain
//
--
==========================================================
Alain Péan - LPP/CNRS
Administrateur Système/Réseau
Laboratoire de Physique des Plasmas - UMR 7648
Observatoire de Saint-Maur
4, av de Neptune, Bat. A
94100 Saint-Maur des Fossés
Tel : 01-45-11-42-39 - Fax : 01-48-89-44-33
==========================================================
_______________________________________________
CentOS mailing list
CentOS@centos.org
http://lists.centos.org/mailman/listinfo/centos
