On Thu, 8 Nov 2007 08:19:42 -0800 Garrick Staples wrote: > Sorry, I ment to respond to the first one. No worry, and many thanks for your reply. > > we're getting sporadic errors when jobs finishes running in a WN and > > has to copy its output to submitter host. > > > > We've configured ssh in our submitter/executer in order to avoid > > requesting password, so for example: > > > > [EMAIL PROTECTED] ~]# su - ops006 > > [EMAIL PROTECTED] ~]$ ssh ce07 date > > Scientific Linux CERN Release 3.0.8 (SL) > > Tue Nov 6 12:09:05 CET 2007 > > [EMAIL PROTECTED] ~]$ > > Non-interactive shells shouldn't print anything other than the > command output. It tends to confused programs that do their own > remote shells (like scp). > > For example, I can't do this on your cluster: > date=`ssh ce07 date` I tried and:
[EMAIL PROTECTED] ~]$ date=`ssh ce07 date` Scientific Linux CERN Release 3.0.8 (SL) [EMAIL PROTECTED] ~]$ echo $date Thu Nov 8 17:29:51 CET 2007 anyway, I don't mind removing issue.net/mod files. Done. > > But looking job's log in WN we find: > > Oct 24 02:20:48 td237 pbs_mom: req_cpyfile, Unable to copy file > > [EMAIL > > PROTECTED]:/home/ops006/.lcgjm/globus-cache-export.Q18475/globus-cache-e > > xport.Q18475.gpg to globus-cache-export.Q18475.gpg > > The actual copy file error message should be in syslog. This grep is done in /var/log/messages. Are you refering to this? # grep "Unable to copy file" /var/log/messages|wc -l 132 Cheers, Arnau _______________________________________________ mauiusers mailing list mauiusers@supercluster.org http://www.supercluster.org/mailman/listinfo/mauiusers
