yes, thanks for the hint. in our case the task would run to long, but for others this might be a valid solution.
Original Message:
Sent: Sep 24, 2024 03:42 AM
From: Stephan Mannert
Subject: Possible bug in V 21.0.11 (and maybe lower)
Hi Matthias,
thanks for the information.
Another workaround might be a WAIT in the activate_uc_object ;)
:SET &COUNTER# = 1:SET &HND# = PREP_PROCESS_AGENTGROUP(HOSTG.TEST,,BY_RULE):PROCESS &HND#: SET &AGENT# = GET_PROCESS_LINE(&HND#,1): IF &COUNTER# EQ 10: PRINT "Activate next job for &AGENT# with WAIT": SET &WAIT# = "WAIT": SET &COUNTER# = 1: ENDIF: PUT_READ_BUFFER AGENT# = &AGENT#: SET &RUNID# = ACTIVATE_UC_OBJECT(JOBP.TEST, &WAIT#)! (Re)set values: SET &WAIT# = "": SET &COUNTER# = ADD(&COUNTER#,1): ENDIF:ENDPROCESS:CLOSE_PROCESS &HND#
Best regards
Stephan
Original Message:
Sent: Sep 20, 2024 03:03 AM
From: Matthias Schelp
Subject: Possible bug in V 21.0.11 (and maybe lower)
Good Morning,
I would like to inform you about a possible bug I found in version 21.0.11 which can be harmfull. The 'allow x simultaneous executions' feature doesn't work with values > 1. At least when the Job/Workflow/.. is activated by script ( e.g. activate_uc_object() )
: S &CNT# = 0: WHILE &CNT# < 10: PRINT &CNT#: S &RC# = ACTIVATE_UC_OBJECT(JOBP.TEST): S &CNT# = ADD(&CNT#,1): ENDWHILE
If you set a limit of max. 2 parallel executions on JOBP.TEST and start the job 10 times, it would run the first 2 while the rest is waiting (as expected). As soon as the first 2 executions ended it will start the 8 remaining jobp all at once ignoring the max parallel setting. In our case it was used for loadbalancing and lead to 100% cpu and an application crash.
We used QUEUE objects as a workaround but SYNC should be possible, too. Hope this helps anyone.
I did open a ticket and they were able to reproduce. Not sure which versions are affected. We updated from 21.0.7 -> 21.0.11 so everything in between may have the bug. If anyone has one of these versions a short test will be appreciated.
Regards,
Matthias