C_PERIOD objects stop activating when empty hostgroup is found
Article ID: 263392
Updated On:
Products
Issue/Introduction
When running a task through an empty agentgroup kicked off with a C_PERIOD, the C_PERIOD stops activating tasks
1) Create a hostgroup with nothing in it (customer has APIs that clean up hostgroups to remove all agents that no longer exist) and
2) Create a job that uses that hostgroup as the agent
3) Execute recurring to run every 1 minute (all defaults)
4) Wait a few minutes and look at Process Monitoring
* * * notice that:
their are multiple child tasks for the C_PERIOD
C_PERIOD has a next check and a next run
Child tasks end with FAULT_OTHER as expected due to an empty hostgroup
5) Execute recurring a new run with "after the previous execution ends plus" set to 2 minutes
6) Wait for a few minutes and look at Process Monitoring
Expected behavior:
Same as "Default" recurring task - First child task should end with fault other, C_PERIOD should have next check and next run and there should be continuing executing child tasks that end with FAULT_OTHER
Actual behavior:
Only one child task runs
Child task ends with FAULT_OTHER due to empty hostgroup (as expected) - U00005013 Definition of the AgentGroup 'HOSTG.NAME' is empty.
C_PERIOD no longer has a next check and next run
C_PERIOD is now stuck in Active status
No more child tasks run
In some cases, the error U00005013 may be accompanied by a forced trace like this:
19 - 20230629/001057.159 - U00005013 Definition of the AgentGroup 'CATBATCH36HG' is empty.
19 - 20230629/001057.160 - U00003620 Routine 'JPEXEC_R' forces trace because of error.
19 - 20230629/001057.160 - U00003517 UCUDB: Invalid DB-STMT IND: '0' at OPC: 'CLST'
Environment
Release : 12.3, 21.0
Resolution
This has been fixed with the release of 21.0.6
Feedback
