I've been debugging some modifications I'm making to seg_sge_module.c and my
experience might be useful.
What I do is:
export SEG_SGE_DEBUG=9
$GLOBUS_LOCATION/libexec/globus-scheduler-event-generator -s sge -t 1198261100
>& tmpfile
Where SEG_SGE_DEBUG is a bitmap selection on what to print out. For =9, it's
suppose to print INFO and TRACE messages & spit 'em out to stderr. I would
think that this functionality is independent of the build flags since it's just
message passing.
However, I have had some problems with the output not appearing after making
minor changes to the code and not being able to get WARN messages out. I
haven't figured out why yet - though I'm guessing pilot error.
- Jeff
> For Esteban's benefit, will it work to enable BUILD_DEBUG for just the
> seg_sge_module.c? If so, where would the debug output go?
>
>
>
> Also,
> On Dec 21, 2007 4:35 PM, Rodolfo Edelmann <[EMAIL PROTECTED]> wrote:
>
> > DEBUG monitoring.SchedulerEventGenerator [Thread-
> 9,startSegProcess:231]> Starting seg process
> > DEBUG monitoring.SchedulerEventGenerator [Thread-
> 9,startSegProcess:250]> executing command:
> > DEBUG monitoring.SchedulerEventGenerator [Thread-
> 9,startSegProcess:253]> ->/usr/local/globus-4.0.5/libexec/globus-
> scheduler-event-generator
> >
> > DEBUG monitoring.SchedulerEventGenerator [Thread-
> 9,startSegProcess:253]> ->-s
> > DEBUG monitoring.SchedulerEventGenerator [Thread-
> 9,startSegProcess:253]> ->sge
> > DEBUG monitoring.SchedulerEventGenerator [Thread-
> 9,startSegProcess:253]> ->-t
> > DEBUG monitoring.SchedulerEventGenerator [Thread-
> 9,startSegProcess:253]> ->1190799979
> > DEBUG monitoring.SchedulerEventGenerator [Thread-9,run:128]
> getting seg
> > input
> > DEBUG monitoring.SchedulerEventGenerator [Thread-9,run:133] Seg
> input> buffer is not ready
> > ERROR monitoring.SchedulerEventGenerator [Thread-9,run:198] SEG
> Terminated> with Fault: globus_xio: Operation was canceled
> >
> > So, it is the same problem but not the same solution.
> >
> > globus-scheduler-event-generator is not returning anything so the
> parsing> can't be complete.
> >
> > When I saw that, I start editing "seg_sge_module.c" and putting
> some> printf in :
> >
> > rp = strptime(fields[0],"%s", &tm);
> > if (rp == NULL || (*rp) != '\0')
> > {
> > printf("Unable to extract timestamp from first
> field"); //Add
> > by me
> > SEG_SGE_DEBUG(SEG_SGE_DEBUG_WARN,
> > ("Unable to extract timestamp from first
> field in line
> > '%s'\n",
> > state->buffer + state->buffer_point));
> > goto free_fields;
> > }
> >
> > and in:
> >
> > stamp = mktime(&tm);
> > if (stamp == -1)
> > {
> > printf("mktime generated invalid timestamp\n"); //Add
> by me
> > SEG_SGE_DEBUG(SEG_SGE_DEBUG_WARN,
> > ("mktime generated invalid timestamp\n"));
> > goto free_fields;
> > }
> >
> > Try that, and tellme how it go.
> >
> > Sorry the patch didn't work
> >
> >
> > Good luck and Merry Christmas. ;D
> >
> >
> >
> > --
> > Rodolfo E. Edelmann
> > Lab. Computación
> > Fac. Cs. Ex.Fis y Nat.
> > Univ.Nac.Córdoba
> >
>
>
>
> --
> Otheus
> [EMAIL PROTECTED]
> +43.699.1049.7813
>
