Re: [Condor-users] writing hawkeye modules

[obi <graziano@xxxxxxxxxxx>]

On Mon, Feb 06, 2006 at 10:29:24PM -0600, Nick LeRoy wrote:
> On Monday 06 February 2006 7:20 pm, obi wrote:
> > more information. Changing the configuration to
> >
> > STARTD_CRON_NAME = HAWKEYE
> > HAWKEYE_JOBLIST = NWS
> > HAWKEYE_NWS_EXECUTABLE = /bin/cat
> > HAWKEYE_NWS_ARGS = "/home/graziano/test"
> > HAWKEYE_NWS_PERIOD = 60s
> >
> > seems to make everything a bit happire: now I got
> >
> > 2/6 17:17:29 Completed initial benchmarks.
> > 2/6 17:17:29 Cron: Initializing job 'NWS' (/bin/cat)
> >
> > in the logs but nothing shows up in the classad (as per codor_status -l).
> >
> > Are there any more logs to check?
> 
> In the StartLog, you should see messages something like this (Make sure you 
> have D_FULLDEBUG) turned on:

My apologies to be so dense: where should I turn D_FULLDEBUG on? I found
at least 2 instances of it in the config file (STORK, CREDD and QUILL I
think). Is there a STARTD_DEBUG?


> 2/6 22:16:35 Cron: 'NWS' (pid 10318) exit_status=0
> 2/6 22:16:35 NWS: 51 lines in Queue
> 2/6 22:16:35 Replacing ClassAd for 'NWS'
> 2/6 22:16:35 Cron: STDOUT closed for 'NWS'
> 2/6 22:16:35 Cron: STDERR closed for 'NWS'

Definetely not in the logs ...

> 
> If you don't see that, then '/bin/cat' isn't exitting as expected...  
> Unfortunately, the 6.7.14 version that you're running doesn't log the command 
> line before exec-ing it (IIRC); it'd be nice to verify that the file name is 
> being passed to /bin/cat correctly (although, as far as I know, the command 
> line generation, excepting for the documented bugs in 6.7.{15,16}, has always 
> worked properly...  Still, my memory could be faulty; try running it with 
> D_FULLDEBUG and see if you learn anything more interesting...
> 
> 
> I do see another problem, however.  When Hawkeye invokes a module, it invokes 
> it like this:
> "/path/to/exececutable name <args>"
> 
> So, in this case, you should be running this, which isn't what you expect:
> "/bin/cat NWS /home/graziano/test"
> 
> However, /bin/cat should still do "The Right Thing(tm)" (according to my 
> exhaustive 3 seconds of testing).  I'll look at it some more when I'm back in 
> the office tomorrow.  Let me know what D_FULLDEBUG reveals.

I had another command complain loudly about the input (and now I
understand why) and everything was in the log. There is nothing with cat.
And no cat in the list of process (in case it got stuck there). 

I'll try harder tomorrow.

thanks
graziano

> 
> -Nick
> 
> -- 
>            <<< Why, oh, why, didn't I take the blue pill? >>>
>  /`-_    Nicholas R. LeRoy               The Condor Project
> {     }/ http://www.cs.wisc.edu/~nleroy  http://www.cs.wisc.edu/condor
>  \    /  nleroy@xxxxxxxxxxx              The University of Wisconsin
>  |_*_|   608-265-5761                    Department of Computer Sciences
> _______________________________________________
> Condor-users mailing list
> Condor-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/condor-users
> 

-- 
+-----------------------+--------------------------+
| Graziano Obertelli    | CS Dept. Rm 102          |
| graziano@xxxxxxxxxxx  | University of California |
| (805) 893-5212        | Santa Barbara, CA 93106  |
+-----------------------+--------------------------+