Implementing and Managing APPC Protected Conversations

64 Implementing and Managing APPC Protected Conversations

The update actions you can do against the UOR depend on the state of the UOR itself. You

can back out or commit any “in doubt” UOR, while you can only remove interest for the other

states. Removing interest is possible only if the resource manager of the UOR is not active. In

terms of APPC, the LU must be inactive. You can inactivate the LU with the VTAM command

v net,id=LUname,inact,f. After the LU activation, with the command v net,id=LUname,act, it

will restart in warm mode without any memory of the UOR you removed. It is your

responsibility to do what is needed on every subsystem involved in the protected

conversation to solve the UOR’s status.

4.2.2 Solving LUs warm/cold or name mismatch problems

If, during an exchange log name transaction, the local LU or partner LU detects a warm/cold

log status mismatch, APPC/MVS issues operator message ATB210E. Messages ATB70052I

and ATB80129I may be returned to the TP. The log status mismatch may be caused by:

򐂰 The wrong level of log data at the local or partner LU

򐂰 A cold log start at one of the partners

If the cold log status is valid for one of the logical units, and if the warm partner is an

APPC/MVS managed logical unit of work, then to resolve the warm/cold mismatch, take one

of the following actions against the warm partner (listed in order of increasing potential

disruption):

򐂰 Restart the warm APPC/MVS LU after removing all interests for the APPC/MVS LU using

the RRS ISPF panels.

򐂰 Delete the LU from the APPC/MVS configuration by issuing a SET command for a parmlib

member with an LUDEL statement for the LU.

򐂰 Remove all interests for the cold status partner using the RRS ISPF panels as described

in 4.2.1, “Solving unit of recovery problems” on page 62.

򐂰 Add the LU to the APPC/MVS configuration by issuing a SET command for a parmlib

member with an LUADD statement for the LU. The APPC/MVS LU will now restart;

however, this will be without incomplete logical units of work.

򐂰 Attempt to initiate a protected conversation between the affected LUs.

If a CICS is involved in the mismatch problem, you have to solve the problem using the CEMT

transaction against UOR and CONNection. CICS (we are interested only in APPC

communications to non-CICS subsystems and not to the classic DPL between CICS

subsystems) maintains for itself the log name information. Example 4-13 shows a sample

syslog output for a mix of successful and unsuccessful exchange log names.

Example 4-13 Syslog sample for a mix condition on an exchange log name.

ATB217I EXCHANGE LOG NAME PROCESSING INITIATED BY LU USIBMSC.SCSIM8IA

787

WITH LU USIBMSC.SCSCPA8K HAS FAILED ON 11/02/2004 AT 17:48:57.

THE LOCAL LU WILL TRY AGAIN TO COMPLETE A SUCCESSFUL EXCHANGE LOG NAME

WITH LU USIBMSC.SCSCPA8K.SOME LOGICAL UNITS OF WORK MIGHT NOT

BE AUTOMATICALLY RESOLVED BY RESYNCHRONIZATION AND NO NEW PROTECTED

CONVERSATIONS MAY BE ALLOCATED BETWEEN THE TWO LOGICAL UNITS

UNTIL AN EXCHANGE LOG NAME TRANSACTION COMPLETES.

Tip: Before acting to remove the UOR interest, go through the RRS panels to discover

which partners are involved in the conversation. Looking at Example 4-12, LUWID

information on the “RRS Unit of Recovery Work IDs” is very useful. Of course a thorough

knowledge of the application flow is a plus.