Last night, we had a started task DB00DTS go end of memory but the associated EOM rule did not send out proper notifications because it appears it did not finish running.
It looks like I have an SVC dump of the associated DB00DTS1 address space with OPSMVS and CONSOLE in there as well, which was taken at the same time the EOM rule failed to execute. I will upload that dump in case it is useful.
From the log I see
Date Time Job Name ----+----1----+----2----+----3----+----4----+----5----+----6----+----7----+----8
20MAY 18:13:31 *MASTER* OPS9999T EOM RETURNED PROCESS BLOCK AT X'2749EF00' FOR ASID: X'00C6'
20MAY 18:13:31 *MASTER* OPS1000J NOTIFY_STATUS=ENABLE
20MAY 18:13:31 JES2 $HASP310 DB00DTS TERMINATED AT END OF MEMORY
So I'm wondering if this is something that OPSMVS should handle better or if we need to figure out another way of notifying in case of these EOM abends?
Release : 13.5
Component : OPS/MVS
That should not be. OPSLOG and SYSCHK1 datasets need to be treated like paging volumes, since any reserve on the packs can cause a system outage, or other strange occurrences. That could explain the original abend with the EOM being terminated if referencing global variables that are paged out.
This (and opslog) are the only explanation for why an EOM rule is terminated.