Re: [HTCondor-users] Debugging a DedicatedScheduler?

[Steffen Grunewald <Steffen.Grunewald@xxxxxxxxxx>]

Good morning everyone,

I had put the whole thing under my pillow for the weekend, and
had another look at another (working, parallel-wise) pool today.

Then it dawned on me that Condor has some, um, peculiar handling
of macro names (in particular, of capitalization and underscores),
and that this might have produced an interpreter loop while
substituting $(DEDICATED_SCHEDULER) for DedicatedScheduler -
can someone confirm that this may cause issues? (If so, please
put it into the manual, in friendly but determined LARGE RED LETTERS.)

Indeed, shortening the intermediate macro name to $(DEDICATED_SCHED)
seems to have me brought a bit further (now the condor_status test 
returned a lot of empty strings which are always better than "undefined"),
but, and here's the culprit that I overlooked all the time:
There has to be a "DedicatedScheduler@" string in front of the 
machine name "of course" (why this reminds me of garters and
belts, somehow?) *and* it's a good idea to enclose everything in
quotes:

DedicatedScheduler = "DedicatedScheduler@$(DEDICATED_SCHED)"
DEDICATED_SCHED    = $(MASTER_MACHINE).$(PUBLIC_DOMAIN_NAME)
(no quotes in the latter case?!)

seems to do the trick - at least

# condor_status -af DedicatedScheduler | uniq -c
    665 DedicatedScheduler@...

and the long-idle jobs finally started running (!!!)

Thanks for your patience and support,
  Steffen

-- 
Steffen Grunewald * Cluster Admin * steffen.grunewald(*)aei.mpg.de
MPI f. Gravitationsphysik (AEI) * Am Mühlenberg 1, D-14476 Potsdam
http://www.aei.mpg.de/ * ------- * +49-331-567-{fon:7274,fax:7298}