[Lustre-discuss] Mount Failure

[Chris Exton]

I am having a few problems with Lustre and I can't seem to find the
answer to my problem on the web so I wondered if you could help?

 

Firstly I would like to give you some information on the kit we have and
how I have configured it so far, apologies if I get some of naming
schemes incorrect.

 

I installed two servers for (MGS/MDT) with RHEL 5.4, Lustre 1.8.1.1 and
the OFED Infiniband driver. These servers are connected to a SUN J4200
which has 12 disks configured as one RAID 10 Lun. 

 

I used the following command to create the lustre filesystem:

 

mkfs.lustre --fsname=lustre --reformat --mgs --mdt
--failnode=lmds02 at o2ib0 /dev/md0

 

This filesystem mounts without a problem with the following command:

 

mount -t lustre /dev/md0 /mnt/test/mdt

 

I then installed a further two servers for (OSS) with RHEL 5.4, Lustre
1.8.1.1 and the OFED Infiniband driver. These servers are connected to a
SUN J4400 which has 24 disks configured as 2x RAID 6 8+2 and 2x RAID 1
for the Journals.

 

I used the following command to zero the Journal:

 

mke2fs -O journal_dev -b 4096 /dev/md0

 

and then issued the following command to create the filesystem and use
the external Journal:

 

mkfs.lustre --fsname=lustre --reformat --ost --mgsnode=lmds01 at o2ib0
--mgsnode=lmds02 at o2ib0 --failnode=lmds02 at o2ib0 --mkfsoptions="-J
device=/dev/md0" /dev/md1

 

The problem occurs when I try to mount the above filesystem:

 

[root at lss01 ~]# mount -t lustre /dev/md1 /mnt/ost1

mount.lustre: mount /dev/md1 at /mnt/ost1 failed: Cannot send after
transport endpoint shutdown

 

I have had a look in /var/log/messages and I get quite a lot of errors
regarding connection refused etc, please see attached file for more
info.

 

I wondered if the above was Infiniband related so I did the above
configuration again using tcp0 only over the GigE network and this
worked without a problem.

 

Any help you can offer would be much appreciated this is my first Lustre
installation so it's quite possible I have configured something
incorrectly.

 

Kind Regards

 

Chris

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20091112/0d8e385d/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: messages
Type: application/octet-stream
Size: 1410055 bytes
Desc: messages
URL: <http://lists.lustre.org/pipermail/lustre-discuss-lustre.org/attachments/20091112/0d8e385d/attachment.obj>