CICS

 View Only

 Related to CF

Rani Singh's profile image
Rani Singh posted 08/13/26 01:35 AM

When we fail over a Coupling Facility (CF), we lose our shared temporary storage queue servers. I’d like to understand whether we can adjust a CF structure definition parameter or update the ARM policy to reduce the outage time.

Andy Clifton's profile image
Andy Clifton

Greetings Rani,

If you are talking about a planned fail over, you can trigger a rebuild of the structure onto another CF without loss of data using an operator command for example:

'SETXCF START,REBUILD,STRNM=DFHXQLS_CST1S,LOC=OTHER'

However, data in a SIMPLEX CF structure is at risk from failure in the structure or the CF regardless whether marked recoverable in CICS and such a failure will cause the contents  of the structure to be lost. I'd suggest you use duplex structures so the system maintains a second copy of the structure and will switch to the alternate.

There is a slight "gotcha" if you have two CFs, your structures can be duplex but if you loose a CF (planned or unplanned) the structures are still there but unless there is a third CF with enough space in the CFRM policy for the structure it becomes simplex (so at risk if that CF or structure fails) until the CF is restored.

The CICS CF severs can also be ARM enabled.

https://www.ibm.com/docs/en/cics-ts/6.x?topic=processes-system-managed-list-structure-duplexing in the CICS Documentation has some information about CF Duplexing and from https://www.ibm.com/docs/en/zos/3.2.0?topic=rebuild-controlling-use-duplexing-process, you need the DUPLEX keyword in the CFRM parameter but https://www.ibm.com/docs/en/zos/3.2.0?topic=resources-installation-guide-duplexing-rebuild may be a good starting point.

There is a tool to cause errors in a CF structure (for testing) documented at https://community.ibm.com/community/user/viewdocument/injerror-a-tool-to-inject-errors-i?CommunityKey=2a2f855c-5950-4a9d-8485-86645982646a&tab=librarydocuments

Scott Fagen's profile image
Scott Fagen IBM Champion

To augment Andy's answer, there are a number of CICS features that may be using the coupling facility.  If CICS is accessing Db2, there could be even more.  Solving the CICS TSQ issue in a vacuum is likely to keep you exposed to other issues should you lose a CF or need to move (rebuild) a structure during active CICS execution.  You might consider looking at the entire infrastructure to determine how to implement a complete HA strategy, rather than just solving a local problem that is "one of n" during a global issue.

Rani Singh's profile image
Rani Singh
Current Symptom: When an environment Coupling Facility (CF) failover occurs, our shared TS queue servers drop connections or lose availability briefly before recovering. We want to determine if we can adjust our structure definitions or modify our z/OS ARM policy rules to achieve a faster, non-disruptive recovery during a CF failover.
Technical Findings: We ran an MVS XCF display command against our shared TS pool structures and verified that our active CFRM policy currently has DUPLEX: DISABLED configured.