Launcher/Supervisor with deadlock on u_fmev60

Products

• CA Automic Dollar Universe

Issue/Introduction

We have a $U 6.10.111 client installed on an AIX Server.
There were no new jobs launched and $U was frozen. A stop-reorg-restart allowed the situation to be solved each time it occurred, and jobs were running after the restart.

After investigation, we saw that uxioserv used a lot of CPU when $U was frozen.

Since the upgrade to 6.10.111 on an AIX 7.2 node, the Launcher and Supervisor remain stuck with a wake-up time in the past
This has occurred on 7 days in 2 months during the same 5-minute timeslot (:55 to :00).
No new job launched at the top of the hour (:00) and $U was frozen. Multiple uxjobend processes remained using a lot of CPU during the time.

From the ps output, we obtain that the issue is that a uxjobend was looping using a lot of CPU

In universe.log there are no messages in the 5-minute timeslot but when uxiotrclck was executed, we can see that the launcher got stuck during the timeslot (~:57) and supervisor ~10 minutes later.

Environment

DUAS v6.10.111-v7.01.02

Cause

This is an issue with the Supervisor process of Dollar Universe

Resolution

A cleanup process was implemented at the end of the supervisor cycle unit to release any remaining locks.

 

Solution:

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

Fix version:

Component(s): Application Server

Dollar.Universe 7.01.11 - Planned release September 2025

 

Additional Information

The reference for this fix is DE122484