Re: [Condor-users] wishlist for notification messages

[Dan Christensen <jdc@xxxxxx>]

Derek Wright <wright@xxxxxxxxxxx> writes:

> On Sat, 21 May 2005 10:18:04 -0400  Dan Christensen wrote:
>
>> I was wondering if there could be a few changes to the format/content
>> to make it easier for both people and machines to parse.
>
> those are all nice suggestions, thanks.  i'll see if we can work those
> into an upcoming development release.  the only potential snag is if
> people already are parsing the emails, they'll scream about changing
> the format.  all you pesky users make our job so much harder... ;)

And as of today I'm now one of the pesky users who is parsing the
current format.  :-)  But I (and maybe others who are parsing the
current format) still would be happy to have the format improved,
because my parsing is fragile, and I have to guess at some things
where the information isn't easily available.

One of the reasons I suggested several changes at once is exactly
because I figured if you're going to break backwards compatibility,
you may as well do lots of changes at once.  And the reason I
posted to condor-users instead of the bugs list is because I
thought other people might have some suggestions about other 
possible changes to the notification messages.  Anyone?

One of the things I proposed was a line like:

Command: /home/spin/integrate/lorentz.condor -i 15 -w 3 -x 1000000000 -d 9 -f /home/spin/integrate/test-lorentz/7/7 -g -w 0 -y 7 7 7 7 7 7 7 7 7 7

I was wondering about a couple of things.  First, would it be better
to separate out the executable from the arguments, like is done for
the input to condor_submit?

Second, how does Condor handle arguments (or even executables) with
embedded spaces?  E.g. if I want to run

  grep "foo bar" filename

can I do this?  And will the notification e-mail contain the necessary
quotes?  (They don't appear in argv, of course, so if Condor just
prints out argv it loses information...)

> instead of having a script parse emails to resubmit jobs that failed
> with a certain code/signal, you should probably check out the
> "on_exit_remove" setting in your submit description file:
>
> http://www.cs.wisc.edu/condor/manual/v6.6.9/condor_submit.html
> (search for "on_exit_remove").

Thanks for that.  It happens rarely enough that I like to investigate
manually before restarting, but I may try that option too.  I'll also
probably start another thread about this when I have a bit more time.

Best,

Dan