During ZDU process, core dumps and job failures happen on postgres db

Products

• CA Automic Workload Automation - Automation Engine
• CA Automic One Automation

Issue/Introduction

When using the Zero Downtime Upgrade (ZDU) to go from 12.3 to 21.0.5 or earlier OR from any service pack within 21.0 to a later service pack or hotfix within 21.0 with postgres as a database for the Automic backend, the following may happen:

• opening views in the Admin perspective cause WP crashes (Signal 11 & dump)
• opening the detail view of tasks causes crashes (Signal 11 & dump)
• Jobs begin to fail, at first they continue to run, but soon after the load some executions fail with: 20221212/133703.211 - U00020331 Runtime error in object 'HEADER.UNIX', line '00012': U00003705 Variable 'UC_EX_JOB_MD_IP_ADDR' not found.
• somehow the dumps also kill set traces, no tracelog generation for the servers that dump are generate

4. after changing the Database back to the previous state the system works again.

Full steps to reproduce:

Customer has a system with patch level 12.3.9 + HF.2 and wants to ZDU to 21.0.4 + HF.1

1. Start ZDU - Result everything works fine

2. Start the V21 ucybdbld utility and load the DB -> load finishes fine

The issues above then start to happen

This also happens within 21.0, e.g. using ZDU to go from 21.0.1 to 21.0.4 HF1

Environment

Release : 21.0.5 and earlier

Resolution

There is a bug that appears to be with the base database.  This will not be fixed for upgrades from 12.3, but is planned to be fixed with 21.0.6.  Until customers are on a fixed version, do not use the ZDU if the backend database for Automic is on postgres.