Somsak Sriprayoonsakul
2007-Dec-20 05:48 UTC
[Lustre-discuss] MDT Upgrade to 1.6.4.1 failed
Hello,
We just upgrade from Lustre 1.6.3 to Lustre 1.6.4.1. However, the
upgraded on MDT+MGS node failed severely. The node kernel panic
immediately when MDT+MGS partition is mounted. Attached is kernel dump
log, obtained from netdump.
Here''s our configuration
OS: CentOS4
Kernel used at that time: 2.6.9-55.0.9.EL_lustre.1.6.4.1smp
4 storage nodes:
storage-0-0 serve MDT+MGS and OST0, back-up for OST1
storage-0-1 serve OST1, back-up for MDT+MGS and OST0
storage-0-2 serve OST2, backup for OST3
storage-0-3 serve OST3, backup for OST2
Fail-over is done by heartbeat.
Lustre is run on Gigabit Network
There''s no client mount at all when the node panic. The node panic
right away when issue "mount /mnt/lfs/mdt". We do tunefs.lustre and
e2fsck on all lustre partitions, then mount again but the node still panic.
We downgrade back to 1.6.3 and everything back to normal. Do we need
to do anything before upgrade?
Best Regards,
--
-----------------------------------------------------------------------------------
Somsak Sriprayoonsakul
Thai National Grid Center
Software Industry Promotion Agency
Ministry of ICT, Thailand
somsak_sr at thaigrid.or.th
-----------------------------------------------------------------------------------
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: log
Url:
http://lists.lustre.org/pipermail/lustre-discuss/attachments/20071220/e77fa9e3/attachment.ksh
Somsak Sriprayoonsakul
2007-Dec-20 06:11 UTC
[Lustre-discuss] MDT Upgrade to 1.6.4.1 failed
I forgot one thing. The machine architecture is x86_64 Somsak Sriprayoonsakul wrote:> Hello, > > We just upgrade from Lustre 1.6.3 to Lustre 1.6.4.1. However, the > upgraded on MDT+MGS node failed severely. The node kernel panic > immediately when MDT+MGS partition is mounted. Attached is kernel dump > log, obtained from netdump. > > Here''s our configuration > > OS: CentOS4 > Kernel used at that time: 2.6.9-55.0.9.EL_lustre.1.6.4.1smp > 4 storage nodes: > storage-0-0 serve MDT+MGS and OST0, back-up for OST1 > storage-0-1 serve OST1, back-up for MDT+MGS and OST0 > storage-0-2 serve OST2, backup for OST3 > storage-0-3 serve OST3, backup for OST2 > Fail-over is done by heartbeat. > Lustre is run on Gigabit Network > > There''s no client mount at all when the node panic. The node panic > right away when issue "mount /mnt/lfs/mdt". We do tunefs.lustre and > e2fsck on all lustre partitions, then mount again but the node still > panic. > > We downgrade back to 1.6.3 and everything back to normal. Do we > need to do anything before upgrade? > > Best Regards, > > ------------------------------------------------------------------------ > > _______________________________________________ > Lustre-discuss mailing list > Lustre-discuss at clusterfs.com > https://mail.clusterfs.com/mailman/listinfo/lustre-discuss-- ----------------------------------------------------------------------------------- Somsak Sriprayoonsakul Thai National Grid Center Software Industry Promotion Agency Ministry of ICT, Thailand somsak_sr at thaigrid.or.th -----------------------------------------------------------------------------------
On Thu, 2007-12-20 at 12:48 +0700, Somsak Sriprayoonsakul wrote:> Hello,Hi Somsak,> We just upgrade from Lustre 1.6.3 to Lustre 1.6.4.1. However, the > upgraded on MDT+MGS node failed severely. The node kernel panic > immediately when MDT+MGS partition is mounted. Attached is kernel dump > log, obtained from netdump.Could I trouble you to open a bugzilla (bugzilla.lustre.org) ticket about this? Thanx, b. -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 189 bytes Desc: This is a digitally signed message part Url : http://lists.lustre.org/pipermail/lustre-discuss/attachments/20071220/3f8a1028/attachment-0002.bin
Somsak Sriprayoonsakul
2007-Dec-21 04:34 UTC
[Lustre-discuss] MDT Upgrade to 1.6.4.1 failed
Sure. I just submitted it. The bug ID is 14529 https://bugzilla.lustre.org/show_bug.cgi?id=14529 Thanks! Brian J. Murrell wrote:> On Thu, 2007-12-20 at 12:48 +0700, Somsak Sriprayoonsakul wrote: > >> Hello, >> > > Hi Somsak, > > >> We just upgrade from Lustre 1.6.3 to Lustre 1.6.4.1. However, the >> upgraded on MDT+MGS node failed severely. The node kernel panic >> immediately when MDT+MGS partition is mounted. Attached is kernel dump >> log, obtained from netdump. >> > > Could I trouble you to open a bugzilla (bugzilla.lustre.org) ticket > about this? > > Thanx, > b. > > > ------------------------------------------------------------------------ > > _______________________________________________ > Lustre-discuss mailing list > Lustre-discuss at clusterfs.com > https://mail.clusterfs.com/mailman/listinfo/lustre-discuss >-- ----------------------------------------------------------------------------------- Somsak Sriprayoonsakul Thai National Grid Center Software Industry Promotion Agency Ministry of ICT, Thailand somsak_sr at thaigrid.or.th -----------------------------------------------------------------------------------