Review: Asimov Monitor
A number of conceptual issues with running asimov monitor
have come up in reviewing that I'd like to capture for understanding if not for improvements:
-
When an asimov run finishes and you type asimov monitor
you will get a response that is just the event name, e.g.GW150914
. It would be useful if at the completion of the run this added a word similar tocompleted
, i.e.GW150914 completed
to assure that everything is at the point where it is safe to doasimov stop
-
Possibly coupled to the above is that when you issue the asimov start
after submitting an asimov run, the asimov condor job stays in thecondor_q
indefinitely until eitherasivom stop
orcondor_rm
is used to remove it. If all monitoring has completed would it be reasonable to expect theasimov start
condor job to shutdown? -
On occasions I have forgotten to do an asimov stop
from a previous run, but in starting a new asimov run gone through the steps and issued a newasimov start
resulting in a secondasimov --chain
being placed in thecondor_q
. Does this have any bad consequences? Would the originalasimov --chain
have automatically detected the new run and perform the monitoring function on it? If only oneasimov --chain
is needed in thecondor_q
could an attempt to start another be detected and halted?
Note: These sorts of questions (and others I may have overlooked on the topic) will likely be raised by future users of asimov start|monitor|stop
Edited by Kent Blackburn