SGE support broken

[bastiaans]

After going through the instructions, I attempted to execute
jobmond.py. When I did that, I received the following error message:
cluster1:/usr/local/sbin # /usr/local/sbin/jobmond.py -c /etc/jobmond.conf
Traceback (most recent call last):
 File "/usr/local/sbin/jobmond.py", line 814, in ?
   main()
 File "/usr/local/sbin/jobmond.py", line 807, in main
   gather.daemon()
UnboundLocalError: local variable 'gather' referenced before assignment

An examination of the code reveals that the SGE data gathering code
was commented out on line 792. Uncommenting it had the following
effect:
cluster1:/usr/local/sbin # /usr/local/sbin/jobmond.py -c /etc/jobmond.conf
 File "/usr/local/sbin/jobmond.py", line 797
   debug_msg( 0, "fatal error: BATCH_API set to 'sge' but python
module 'sge_drmaa' is not installed' )

                               ^
SyntaxError: EOL while scanning single-quoted string

Commenting out everything but "gather = SgeDataGatherer()" gave me the
following error:
cluster1:/usr/local/sbin # /usr/local/sbin/jobmond.py -c /etc/jobmond.conf
Traceback (most recent call last):
 File "/usr/local/sbin/jobmond.py", line 814, in ?
   main()
 File "/usr/local/sbin/jobmond.py", line 800, in main
   gather = SgeDataGatherer()
 File "/usr/local/sbin/jobmond.py", line 419, in __init__
   self.initSgeJobInfo()
 File "/usr/local/sbin/jobmond.py", line 426, in initSgeJobInfo
   self.qstatparser = SgeQstatXMLParser( SGE_QSTAT_XML_FILE )
NameError: global name 'SGE_QSTAT_XML_FILE' is not defined

At this point, I decided to search my systems for references to drmaa.
I saw several references to C++ example and header files related to
it. Is the sge_drmaa module supposed to be provided by Job Monarch or
Sun Grid Engine? 

[bastiaans]

reported by Jeremy Stout

[bastiaans]

made some fixes in changeset r347

[bastiaans]

I downloaded the trunk version and tried it. When I started the
script, it failed to daemonize. If I comment out lines 326-328 and
start the program, I get the following output:

cluster1:/usr/local/sbin # /usr/local/sbin/jobmond.py -c /etc/jobmond.conf
Traceback (most recent call last):
 File "/usr/local/sbin/jobmond.py", line 813, in ?
   main()
 File "/usr/local/sbin/jobmond.py", line 806, in main
   gather.daemon()
 File "/usr/local/sbin/jobmond.py", line 334, in daemon
   self.run()
 File "/usr/local/sbin/jobmond.py", line 341, in run
   self.jobs = self.getJobData( self.jobs )
TypeError: getJobData() takes exactly 1 argument (2 given)

If I remove the "self.jobs" argument from lines 342 and 343, the
script daemonizes itself. However, I don't think it will produce the
desired output if I try to do that.

Thank you for your help. 

[bastiaans]

done some additional fixes in changeset r348, to make datagatherer superclass more uniform for different types of datagatherer's

[bastiaans]

I can't fix this, I don't have access to SGE systems

[anonymous]

patch for SGE 6.0

[d.love@…]

I've attached diffs tested with SGE 6.0u11. Unfortunately, they're against svn v. 450, and I can't currently test. I hope to set up jobmonarch with SGE 6.1 before long.

[bastiaans]

I will try to backport your patch into current trunk and send it out to the testers.

right now I'm working on release 0.3.1 with some small bugfixes first though.

[d.love@…]

untested merge of changes against revision 496

[bastiaans]

this is now committed, awaiting test user results.

without any significant bug reports it will be included in version 0.4

[ramonb]

my username changed

[ramonb]

On hold until I can receive confirmation from Dave Love or other SGE users.

[davel]

I belatedly added some fixes to the trunk and 0.3 branch to make them consistent with what I've had running under a few versions of SGE 6.0. The result may work with 6.1, and should be easily fixable if it doesn't. The output has changed radically in 6.2, though, and that currently won't work. I'll see if I can fix it later, though there's no relevant doc, unfortunately.