-
I switched my use of drmaa_synchronize() to serially running drmaa_wait() for a list of jobs. On occasion I still get the same problem. I tried stepping through the code in gdb, and found that the infinite loop is the do-while loop in lsfdrmaa_job_update_status(). It appears that lately_submitted is set to 1 so the chunk of code that I'm supposing should be run is never run.
167.
2009-12-04 21:56:56 UTC in FedStage DRMAA for LSF
-
Using DRMAA for LSF 1.0.3.
Using -C in a native specification results in the error "Error in native specification".
It appears that in native.rl, capital C is not one of the allowed arguments:
short_with_arg = ([LcFWkPGgqRmnJbtuUMDSvfa] | 'i' 's'? | [oe] 'o'?
| 'w' [at]? | 'E' 'p'? | 'sp' | 'sla' | 'ext' 'sched'?
| 'jsdl' '_strict'? | 'We' |...
2009-10-19 22:45:11 UTC in FedStage DRMAA for LSF
-
When submitting several jobs to drmaa_synchronize() where the job run
times vary by several hours, drmaa_synchronize() often does not
return. I rebuilt DRMAA for LSF with --enable-debug. Here is the
relevant part of the debug log:
d #5e3c [ 13866.94] * lsb_openjobinfo( 40662[0], NULL, NULL, NULL, NULL, ALL_JOB ) =1
d #5e3c [ 13866.94] * lsb_readjobinfo(...) =0x2afec5ee29e0: more=0
t...
2009-10-06 23:13:39 UTC in FedStage DRMAA for LSF
