We would like clarification on SO10450 and SO10458:
1) Is the only pre-requisite for the problem to happen that multiple events to run simultaneously?
2) What kind of events (AOF rules)?
3) Which address space held the latch (OPSMAIN)?
4) The SQL tables in question are OPS/MVS SQL tables, right?
Release : 13.5 and 13.0
Component : OPS/MVS
It is possible for a task to go thru EOT when multiple events were in flight simultaneously, resulting in process block cleanup that may miss the fact that one of the processes held a latch.
Once the latch is left on it will be held until the address space terminates. If it is a latch used to protect SQL tables, then depending upon automation present, this may cause all automation to cease once all process blocks are held.
The system may grind to a halt, with no real work being dispatched.
The impact will be that an IPL may need to be performed.