MQWP table filling up with EXSTAJ messages when too many unix jobs
Article ID: 387079
Updated On:
Products
Issue/Introduction
After a reconnect of agents, while many jobs are started (perhaps as a part of UC_EXECUTOR_END or UC_EXECUTOR_START), the MQWP table gets flooded with EXSTAJ events. This impacts performance and could cause an outage. On Oracle, the MQWP table can be checked with SQL like:
with tmp as (
select utl_raw.cast_to_varchar2( dbms_lob.substr(MQWP_MSG, 8, 3 )) msg from mq1wp
) select msg, count(msg) as total from tmp group by msg
Environment
Automic Automation version: 21.0.12 or lower
Cause
The cause of the EXSTAJ build up is a defect where the agent process would send continuous ESXTAJ messages to the AE if the AE would not respond because it was too busy with functions like agent onboarding.
Resolution
This has been fixed with 21.0.13 for the UNIX agent with defect DE154684:
A problems has been fixed where the native Unix Agent overloaded the message queue by resending end status messages.
Feedback
