Y.Lyublev wrote:
> Dears.
> Last days exists strange situation when submit the jobs
> (on LCG-CE and gLiteCE) with use different RBs (WMSLB).
> Upgrade of midleware (glite-yaim-3.0.1) was made on WNs only.
>
> 1. The jobs submitted from local UI to local (g)CE with use default
> local (g)RB are executed successfully.
>
> 2. But the jobs, received from outside of, ended orderly, but
> resubmited newly by RB. The user gets the messages that jobs
> abended after 3 attemps.
Always look at the output of this command for a failed job:
edg-job-get-logging-info -v 1 job_ID
or
glite-job-logging-info -v 2 job_ID
In your case the relevant error is this:
---------------------------------------------------------------------
Event: Done
- exit_code = 127
- host = wn63.itep.ru
- reason = Cannot download env.sh from
gsiftp://rb105.cern.ch/var/edgwl/SandboxDir/xR/
https_3a_2f_2frb105.cern.ch_3a9000_2fxR8zeX8UfKVqMTWjdXECnA/input/
- source = LRMS
- status_code = FAILED
---------------------------------------------------------------------
Your WN cannot do a globus-url-copy from/to the RB.
The GOC Wiki has tips:
http://goc.grid.sinica.edu.tw/gocwiki/Cannot_download_.BrokerInfo
|