DQM errors when submitting jobs to logical queue via uxsubjob

Document ID : KB000123733
Last Modified Date : 26/12/2018
Show Technical Document Details
Issue:
Some DQM related errors may appear in universe.log when submitting many jobs to a Logical Queue pointing to multiple Physical Batch Queues via uxsubjob.

Example, launching  2000 jobs via the command uxsubjob against a Logical Queue with remote Physical queues may have 5 or 6 Jobs causing the errors in universe.log. 
Nevertheless, the submission works fine.

Errors in universe.log of the logical queue node: 
|ERROR|X|DQM|pid=p.t| u_dqm_conf_exec_dist | dqm_lecture_directe_job returns 3 [XXX] 
|ERROR|X|DQM|pid=p.t| owls_dqm_job_conf | u_dqm_conf_exec_dist returns 3 
|ERROR|X|DQM|pid=p.t| process_fnc_conf | process_conf in error: 3 
|ERROR|X|DQM|pid=p.t| u_dqm_process_client_requ | Error handling job request: owls_dqm_handle_job returns 104 

Errors in universe.log in remote physical queue node: 
|ERROR|X|DQM|pid=p.t| u_dqm_trt_que_physical | u_dqm_conf_exec returns 3 [XXX]
 
Environment:
Unix / Linux
Cause:
The root cause is related to how the load balancing is performed by DQM in case of uxsubjob submission
Resolution:
The Error messages will no longer be displayed in universe.log, they will only appear in TRACE level in case the Main Log Level is set to DQM_

Update to a fix version listed below or a newer version if available.

Fix version: Dollar Universe 6.9.61
Component: Application.Server
Release Date: 15th January 2019