I am running a large job on our HPC that uses the Rocks distribution.

I get after a long time a crash which I am not able to trace down to the reason for. But I have found the following messages posted around the time when the crash occurred( Feb 01 23:05). Note that my username (a8h) appears on the later unmount messages (a8h)

Do you have any idea of the following messages - what they tell me? I appreciate any help

The following was found in the file /var/log/messages

Feb 1 23:01:39 ceres-server-0 rpc.mountd[1768]: authenticated unmount request from 10.1.1.206:746 for /state/partition1/apps (/state/partition1)
Feb 1 23:02:36 ceres-server-0 rpc.mountd[1768]: authenticated unmount request from 10.1.1.207:748 for /state/partition1/apps (/state/partition1)



Feb 1 23:21:48 ceres-server-0 rpc.mountd[1768]: authenticated unmount request from 10.1.1.215:980 for /state/partition1/home/a8h (/state/partition1/home)
Feb 1 23:22:06 ceres-server-0 rpc.mountd[1768]: authenticated unmount request from 10.1.1.219:882 for /state/partition1/home/a8h (/state/partition1/home)
Feb 1 23:22:19 ceres-server-0 rpc.mountd[1768]: authenticated unmount request from 10.1.1.200:954 for /state/partition1/home/a8h (/state/partition1/home)



( I get many more with different IP (10.1.1.XXX.YYY) for the case with reference to my directory a8h )

Hello AKJT, & welcome to LQ! Useful reading: How_To_Ask_a_Question There's also one on secure questions.

Also, it's best if you treat us like we don't know much about you, so that you have to explain everything. Because, actually, it's true, and you didn't explain much

Anyhow, with so few details, only vague answers.Some vague IPs (probably on the network) are telling your system to unmount that particular directory. It could be inconvenient if you are running an all-night job. Possible causes are: cron jobs; backup scripts; fsck; (Being Paranoid) intrusion.

Anyhow possible fixes are: Close eth0, or whatever nic they are coming in on; Leave a local or even remote terminal open with a user logged in in that particular directory. You can have that open terminal in a locked room. Attempts to unmount the drive will fail with 'drive busy' type of errors unless the '-f' option is used. If you have a firewall, there are options there.

Ongoing, I would investigate why these commands are coming in and which user is behind them.