Bug 5930 - Make SchedulerEventGenerator.run() (Java) more robust
: Make SchedulerEventGenerator.run() (Java) more robust
wsrf scheduler interface
: 4.0.6
: Macintosh All
: P3 normal
: 4.0.7
Assigned To:
  Show dependency treegraph
Reported: 2008-03-18 15:31 by
Modified: 2008-03-20 16:48 (History)



You need to log in before you can comment on or make changes to this bug.

Description From 2008-03-18 15:31:48
Currently the Thread SchedulerEventGenerator stops processing if either
the restart of the c-SEG throws an IOException or the communication
between c-SEG and SchedulerEventGenerator fails with an IOException.
No error is printed and the run() method just terminates silently.
This causes that no job status information is available anymore for jobs 
to the correspondent LRM and no information about problems is printed to the
container log.
I think we ran into that with LEAD, since they didn't get information about
PBS job status and the PBS SEG related thread was missing in the thread dump.

Will make this a bit more robust and add more logging so that problems here
become more obvious.