Re: [condor-users] Limitations in DAGman

[Corey Shields <cshields@xxxxxxxxxxx>]

On Sat, 2003-09-06 at 23:06, Mark Silberstein wrote:
> Hi
> I used DAGman with multiple job submission ( simply since I didn't read
> the manual and didn't know of that limit of one job ). There was one
> problem of specifying a log file. The point is that DAGman monitors all
> the events using log file. So if your DAGman nodes write to different
> log files, DAGman fails. However if you force all of them to report to
> the same log, everything seems to work fine. Although officially this
> behavior is not supported, it worths trying on your DAG.

Mark,

Thanks!   We tried this and have run into problems.  It seems to only
like the first queuing of a "queue 100" submission.   We tried a .dag
with a parent and a child, both of which were set to "queue 100".   The
first submit got queued okay, but then the dag segfaulted before it
could catch the finished product and start the next one.   Can you check
your dag logs and let me know if you see anything like this?

9/8 16:13:33 Submitting Condor Job A ...
9/8 16:13:33 submitting: condor_submit  -a 'dag_node_name = A' -a
'dagman_job_id
 = 9669.0' -a 'submit_event_notes = DAG Node: $(dag_node_name)'
9440.submit 2>&1
9/8 16:13:40    assigned Condor ID (0.0.0)
9/8 16:13:40 Registering condor_event_timer...
9/8 16:13:41 Event: ULOG_SUBMIT for Condor Job A (9670.0.0)
9/8 16:13:41 Event: ULOG_SUBMIT for Condor Job A (9670.1.0)
9/8 16:13:41 Unrecognized submit event (for job "A") found in log (none
expected
)
9/8 16:13:41 Event: ULOG_SUBMIT for Condor Job A (9670.2.0)
9/8 16:13:41 Unrecognized submit event (for job "A") found in log (none
expected
)

(repeats for about 40 more of the jobs before dying).

Cheers,

-Corey

-- 
Corey Shields - IU Unix Systems Support Group
http://ussg.iu.edu/~cshields

My PGP/GPG public encryption key is at:
http://www.ussg.iu.edu/cshields/cshields_pub_key.asc
GPG fingerprint: 78A8 E5EB E455 0A90 F392 59BC A6AF F8A3 A304 1453

Attachment: signature.asc
Description: This is a digitally signed message part