ACF2 Backup failed with IEC020I 001-3 anbd S001 abend and it has locked the ACF2 Logon IDS backup file making it unavailable

Products

• ACF2
• ACF2 - z/OS
• ACF2 - MISC

Issue/Introduction

ACF2 Backup failed & it has locked the ACF2 Logon IDS backup file with an exclusive enqueue.

IEC020I 001-3,ACF2,ACF2,BACKLID,xxxx,yyyyyy,ACF2.LOGONIDS.BACKUP
IEC020I NON-ACCEPTABLE ERROR
CCSR010E MAINBKUP S001 at 00xxxxxx LMOD IGG019AR CSECT N/A +N/A ACF2 N/A ACF2
IKJ56861I DATA SET ACF2.LOGONIDS.BACKUP NOT UNALLOCATED, DATA SET IS OPEN
 
How can the file be released so it can be used?

Environment

Release:
Component: ACF2MS

Resolution

The ACF2 backup got an abend from an I/O error. Similar abends like this happened when in a shared environment and two systems tried to run the backup at the same time. When the ACF2 backup runs the code checks to see if the backup dataset is already allocated in the ACF2 proc. If it is not, then ACF2 allocates the dataset with DISP=OLD and the dataset is opened. If during the backup processing you get a S001 abend, ACF2 will detect this and bypass the close. ACF2 will attempt to unallocate the dataset since it  did allocate it during the backup process. The unallocate fails and as seen that by the message that comes out right after the S001 abend

IKJ56861I DATA SET ACF2.LOGONIDS.BACKUP NOT UNALLOCATED, DATA SET IS OPEN

A long time ago ACF2 would abend after this occurrence which was deemed to be bad thing, so code was written for ACF2 to stay up and running. But the enqueue on the backup DSN stays in place until cleared. The enqueue can only be cleared by bouncing the ACF2 address space, or doing an IPL. With ACF2 staying up, this leaves the choice of when to do this to the site, instead of an abend forcing that at the time.
 
If a site bounces(stop and start) ACF2, it is recommended doing this at a slow time as the master console will get security prompts like this:
 
ACF90CCC 
name, ACF2 NOT ACTIVE: REPLY 'U' TO CONTINUE, 'C' TO CANCEL, OR 'W' TO WAIT AND RE-START ACF2 

There are multiple ACF9xxxx messages. Reply W for wait. Then that task will wait till ACF2 becomes active again.