Summary
* The JEM module has been updated. Fixed: wrong interpretation of config
params enclosed in single quotes; catch multiple similar config params
when stripping online mode (Tim dos Santos)
* Sending attemptNr to the PanDA server on each updateJob call to
alleviate a book problem with jobs being retried (requested by Tadashi
Maeno)
* Added a special error code (1194, ERR_NOFILEVERIFICATION) for cases
where all file verifications (checksums/file size) failed, i.e. if e.g.
lcg-get-checksum and subsequent alternative checksum/file size commands
failed to read the remote file due to an SE problem, even though the
original stage-out seemed to have worked
* Corrected a bad variable name in the pilot fixing a problem with thrown
exceptions in failed jobs in multi-job mode (causing the multi-job pilot
to crash when more than four jobs failed due to stage-out problems)
* Pilot is now aborting if copytool=lcgcp2 and lfcregister are not set
(for non-US sites in the US several sites are using the lcgcp2 site
mover in combination with LFC registration by the pilot)
* Added special check for storm sites related to server based LFC
registration (pilot resets copy tools to defaults if not set correctly in
the schedconfig; if lfcregister='server', copytool='storm' and
copytoolin='' then the pilot will reset copy tools to copytool='lcgcp2'
and copytoolin='storm'). Requested by Ikuo Ueda et al
* Updated usage of function isinstance() in the pilots' timed command
class to allow for unicode strings; problem seen when python 2.6 was used.
Solution provided by Rodney Walker
* Added a 1 m sleep before attempting file removal after a failed
stage-out. Requested by Rodney Walker et al
Version info
Sending attemptNr with node structure in getNodeStructure() (PandaServerClient)
Removing attemptNr unless command is updateJob in httpConnect() (pUtil)
Modified argparser() to return a queuename instead of useless pworkdir (runJob)
Now failing job if file verification was not successful in put_data() (lcgcpSiteMover, lcgcp2SiteMover)
Added new error code 1194, ERR_NOFILEVERIFICATION (PilotErrors)
Corrected number_of_jobs -> number_of_failed_jobs in runMain() fixing a problem with thrown exceptions in failed jobs in multi-job mode (when >4 failures) (pilot)
Pilot is now aborting if copytool=lcgcp2 and lfcregister is not set in handleQueuedata() (pilot)
Added special check for storm sites in handleQueuedata() (pilot)
Added new version of JEMstub from Tim
Sending -h queuename to child in runMain() (pilot)
Corrected str -> basestring in isinstance() call in _child(). Solution provided by Rodney Walker (timed_command)
Adding a 1 m sleep before file removal in removeFile() (SiteMover)
No comments:
Post a Comment