Tell me more ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.
up vote 0 down vote favorite

and thank you for any help with this problem. I will find some questions I can answer well to pay the community back either way.

I keep getting the following error on my test cluster syslog:

May 20 04:41:28 MediaBox pbs_mom: LOG_ERROR::sys_copy, command '/usr/bin/scp 
-rpB /var/spool/pbs/spool/169104.davidbook.ER david@DavidBook://Test.e169104' 
failed with status=1, giving up after 4 attempts

I have thoroughly configured and tested ssh/scp for access both directions with no password.

When I search Google for the error, I find that most people have user@domain:/DIRECTORY/... in the error message. Like I said, I can scp both ways without a password, but if I try to omit the directory like "david@DavidBook://Test.e169104" above, then it doesn't work.

I guess there must be some way to tell it which directory to use, and that it isn't properly set that way be default (there were also some file permission problems).

The scripts run fine if I run them manually on the "client" node.

I am using Mandriva 2010.2 with Torque 2.4.8.

Here are some more logs each of which may or may not be useful.

mom_logs: (on client node)

05/20/2012 04:41:19;0001;   pbs_mom;Job;TMomFinalizeJob3;job not started, Failure job exec failure, after files staged, no retry (see syslog for more information)
05/20/2012 04:41:19;0080;   pbs_mom;Svr;preobit_reply;top of preobit_reply
05/20/2012 04:41:19;0080;   pbs_mom;Svr;preobit_reply;DIS_reply_read/decode_DIS_replySvr worked, top of while loop
05/20/2012 04:41:19;0080;   pbs_mom;Svr;preobit_reply;in while loop, no error from job stat
05/20/2012 04:41:19;0080;   pbs_mom;Job;169104.davidbook;obit sent to server
05/20/2012 04:42:33;0002;   pbs_mom;Svr;pbs_mom;Torque Mom Version = 2.4.8, loglevel = 0
05/20/2012 04:47:33;0002;   pbs_mom;Svr;pbs_mom;Torque Mom Version = 2.4.8, loglevel = 0

server_logs: (on server node) (doesn't look useful)

05/20/2012 04:41:35;000d;PBS_Server;Job;169096.davidbook;Email 'o' to david@DavidBook failed: Child process 'sendmail -f adm david@DavidBook' returned 127 (errno 10:No child processes)
05/20/2012 04:41:36;0100;PBS_Server;Job;169092.davidbook;dequeuing from batch, state COMPLETE
05/20/2012 04:41:36;0100;PBS_Server;Job;169094.davidbook;dequeuing from batch, state COMPLETE
05/20/2012 04:41:36;0100;PBS_Server;Job;169098.davidbook;dequeuing from batch, state COMPLETE
05/20/2012 04:41:36;0100;PBS_Server;Job;169095.davidbook;dequeuing from batch, state COMPLETE
05/20/2012 04:41:36;0100;PBS_Server;Job;169099.davidbook;dequeuing from batch, state COMPLETE
05/20/2012 04:41:36;0100;PBS_Server;Job;169096.davidbook;dequeuing from batch, state COMPLETE
05/20/2012 04:41:36;0040;PBS_Server;Svr;davidbook;Scheduler was sent the command term
05/20/2012 04:41:40;0002;PBS_Server;Svr;PBS_Server;Torque Server Version = 2.4.8, loglevel = 0
05/20/2012 04:46:40;0002;PBS_Server;Svr;PBS_Server;Torque Server Version = 2.4.8, loglevel = 0
05/20/2012 04:51:36;0040;PBS_Server;Svr;davidbook;Scheduler was sent the command time
05/20/2012 04:51:40;0002;PBS_Server;Svr;PBS_Server;Torque Server Version = 2.4.8, loglevel = 0
share|improve this question
I seem to have found the answer. The key was to use insert "$usecp :/ /home/david/spool" into mom_priv/config, since that directory is shared over the network. I still don't know why scp didn't have a destination directory, but this circumvented the problem. I don't know if this answer is 100% correct because I don't like the ":/" part, but it works, and $usecp is a good idea. $usecp is useful for avoiding using scp when the nodes have a shared mount. The rest of my problem is that the group submitting the jobs does not exist on the compute node, so I'm off to fix that and such. – David W. Allor May 20 '12 at 12:07
I guess that on a real cluster I should have used a directory outside of the /home share, for all users. Also, I tried using autofs, but it was not reliable. – David W. Allor May 20 '12 at 12:10

Know someone who can answer? Share a link to this question via email, Google+, Twitter, or Facebook.

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.