CEMT SET CONNECTION

Description

You can use the CEMT SET CONNECTION command to change some attributes that define an MRO or ISC over SNA connection.

When a connection has been defined as INDIRECT, no information is given unless the direct connection to which it refers is installed.

To install a new connection definition when one is already installed, you must set the connection OUTSERVICE and, if it is z/OS® Communications Server-connected, release it before using the CEDA INSTALL commands for your new definition.

Options

(value)

One or more names (1 - 4 characters) defined for an interregion communication (IRC) or an intersystem communication (ISC) connection.

ACQUIRED (APPC only)

Specifies whether CICS® is to acquire a session with the logical unit represented by the CONNECTION name. To get more detailed information about the availability status of the connection elements, use the CEMT INQUIRE MODENAME command. A connection cannot be both ACQUIRED and OUTSERVICE.

ALL

Any changes you request are made to all resources of the specified type that you are authorized to access.

BACKOUT (APPC PARALLEL-SESSION, CICS-TO-CICS MRO, AND LU61 ONLY)

All UOWs shunted because of the failure of this connection are to be backed out. The normal resynchronization process is to be partially overridden: decisions are taken for any units of work that are indoubt because of a failure of the connection, but the decisions are recorded and any data inconsistencies are reported when the connection is next acquired.

CANCEL

Automatic initiate descriptors (AIDs) queuing for the specified connection are to be canceled.

AIDs representing scheduled and allocated requests waiting in the local CICS system for the specified connection are canceled. However, TD AIDs with an associated triggered task already started are not be canceled. In addition, the following CICS system AIDs are not purged unless FCANCEL is specified.

Table 1. System AIDs requiring FCANCEL to remove them

Description	Tranid
Remote delete AIDs
Remote scheduler AIDs	CRSR
LU6.2 service manager 1 AIDs	CLS1
LU6.2 service manager 3 AIDs	CLS3
Remote scheduler PURGE AIDs	CRSQ
Resource manager resync AIDs	CRSY
Autoinstalled terminal delete AIDs	CATD
Autoinstalled terminal restart AIDs	CATR

When a canceled SCHEDULE request is found to have a precursor in a remote CICS system; that is, the AID was originally scheduled in a remote system, the remote AID is canceled asynchronously.

Message DFHTF0100 is written to CSMT to indicate how many AIDs have been deleted for the connection and how many remain.

An ‘AIDS CANCELED' message appears on the CEMT panel whenever AIDs are deleted using the CANCEL option of the CEMT SET CONNECTION command.

COMMIT (APPC parallel-session, CICS-to-CICS MRO, and LU61 only)

All UOWs shunted because of the failure of this connection are to be committed. The normal resynchronization process is to be partially overridden: decisions are taken for any units of work that are indoubt because of a failure of the connection, but the decisions are recorded and any data inconsistencies are reported when the connection is next acquired.

ENDAFFINITY (APPC and LU6.1 only)

Specifies, where CICS is a member of a z/OS Communications Server generic resource group, that z/OS Communications Server is to end an affinity owned by CICS. This option is valid only for APPC and LU6.1 connections. The connection must be out of service and, for APPC, in NORECOVDATA state.

Note:

1. There is no facility in z/OS Communications Server for inquiring on affinities, so CICS has no certain knowledge that an affinity exists for a given connection. Whenever there is a possibility that an affinity has been created that you must end explicitly, CICS issues message DFHZC0177. This message gives the NETNAME of the suspect connection.
2. If a request to end an affinity is rejected by z/OS Communications Server because no such affinity exists, CICS issues message DFHZC0181.
3. A response of 'NETID 0 USE PRFRM' means that you cannot use SET CONNECTION to end the affinity because the value of NETID in the installed connection is 0. (NETID is the name by which the network containing the connected LU is known to z/OS Communications Server.) To end the affinity, you must use the PERFORM ENDAFFINITY command, on which you must specify the correct NETID.

   This response can also result from repeatedly hitting the ENTER key after the affinity has been ended.

4. Generic resources and affinities are described in Intercommunication concepts and facilities.

FCANCEL

All AIDs, including system AIDs, queuing for the specified connection are to be canceled. See Table 1 for a list of those system AIDS that require FCANCEL to remove them. This can lead to unpredictable results and should be used only in exceptional circumstances.

Note: FCANCEL does not remove transient data AIDs with an associated triggered task. These aids may be removed by purging the associated task.

An ‘AIDS CANCELED' message appears on the CEMT panel whenever AIDs are deleted using the FCANCEL option of the CEMT SET CONNECTION command.

FORCEPURGE (z/OS Communications Server only)

All transactions running on sessions on the connected system are immediately terminated abnormally. This can lead to unpredictable results and should be used only in exceptional circumstances.

In some extreme cases (for example, if an error occurs during backout processing), CICS might terminate abnormally.

FORCEPURGE replaces PURGE FORCE which is retained only for compatibility purposes. You should use FORCEPURGE in new applications.

FORCEUOW (APPC PARALLEL-SESSION, CICS-TO-CICS MRO, AND LU61 ONLY)

All UOWs shunted because of the failure of this connection are to be forced to back out or commit, as specified in the ACTION option of the TRANSACTION definition. The normal resynchronization process is to be partially overridden: decisions are taken for any units of work that are indoubt because of a failure of the connection, but the decisions are recorded and any data inconsistencies are reported when the connection is next acquired.

INSERVICE

The system is in service; that is, it is available for use.

For an MRO connection, all sessions are placed in service and the following occurs:

• If both the issuing system and system sysid have IRC open, and the issuing system has status INSERVICE within system sysid, connections are established and made available between the two systems.
• Otherwise, the status of the system (and the status of the underlying sessions) is set INSERVICE, so that when both systems have IRC open and are INSERVICE with respect to each other, connections are established. Note that INQUIRE indicates that systems (and their underlying sessions) are INSERVICE even though no connections can be used.
• The status of the underlying sessions for a system is always the same as that for the system itself.

For an ISC system, the following occurs:

• LU6.1 system (that is, links to sysid are through intersystem communication): all underlying sessions are placed in service.
• APPC system: causes the SNASVCMG sessions to be placed in service thereby enabling the connection subsequently to be established.

KILL

Terminate the task. System and data integrity is not guaranteed. The KILL option extends the PURGE and FORCEPURGE options. Use this option only after first attempting a purge or forced purge. The KILL option does not guarantee integrity of any kind, but in some situations you can use it to free up a stalled region, thus enabling the region to continue processing. In some cases, for example, if a task is killed during backout processing, CICS terminates abnormally.

NORECOVDATA (APPC only)

Specifies that the normal resynchronization process is to be overridden. NORECOVDATA forces indoubt units of work (according to the transaction definitions), targets any resyncs that were outstanding for the connection, and erases the logname previously received from the partner system. The state of the connection is reset.

Note: You should use SET CONNECTION NORECOVDATA only in exceptional circumstances. It erases recovery information and may compromise data integrity for units of work that have updated resources on remote systems.

Examples of circumstances in which you might need to use it are:

• You need to discard a connection, or issue a SET CONNECTION ENDAFFINITY command, and it is not possible for the quiesce protocols with the partner system to be completed. (Neither action is possible for an APPC connection if recovery data is outstanding.)
• An operational or logic error results in a logname mismatch for the connection. The connection state must be reset to allow the exchange lognames process to complete.

Note: If you specify NORECOVDATA you cannot specify COMMIT, BACKOUT, FORCEUOW, RESYNC, or NOTPENDING.

NOTPENDING (APPC and CICS-to-CICS MRO only)

Specifies, for either of the following kinds of connection, that the normal resynchronization process is to be overridden:

• A connection to a CICS TS for z/OS, Version 5.2 partner that has performed an initial start
• A connection to a pre- CICS TS for z/OS, Version 5.2 partner that has performed a cold start.

NOTPENDING forces all indoubt units of work (according to the transaction definition) that were created by the connection before the initial (or cold) start of the partner. It also forgets any resyncs (waitforget UOW-links) that are outstanding for the connection, and created before the initial (or cold) start of the partner.

The PENDING condition indicates the existence of recovery information (either shunted UOWs or decisions remembered for the partner) on a connection that has experienced a lognames mismatch with its partner. For a CICS TS for z/OS, Version 5.2 partner, a lognames mismatch indicates that the partner has performed an initial start. For a pre-CICS TS for z/OS, Version 5.2 partner, a lognames mismatch indicates that the partner has performed a cold start. In either case, the recovery protocol has been corrupted by a loss of log data at the partner.

It is not possible to set a connection to NOTPENDING state (forcing indoubt and erasing NOFORGET UOWs) until this system has made contact with the partner and received a new logname from it.

Decisions for a whole connection can be forgotten, but that does not affect the memory of a decision for any other connection involved in the UOW.

Note:

1. If you specify NOTPENDING you cannot specify COMMIT, BACKOUT, FORCEUOW, RESYNC, or NORECOVDATA. For advice on which option to use, see CEMT SET CONNECTION - notes on RES ync.
2. NOTPENDING has no effect on MRO connections to pre-CICS TS for z/OS, Version 5.2 systems. However, the cold start of a pre-CICS TS for z/OS, Version 5.2 MRO partner causes the SET CONNECTION NOTPENDING function to be performed automatically, session by session.

The exchange lognames function and the resynchronization function are described in Defining links for multiregion operation and the Systems Network Architecture—LU6.2 Reference: Peer Protocols manual.

OUTSERVICE

The system is out of service; that is, it is not available for use.

For an MRO connection, all sessions are placed out of service (immediately if PURGE is specified, or when tasks have terminated if it is not) and the following occurs:

• If the connection is currently ACQUIRED, the sessions are broken (quiesced). The connection cannot be used until it is once again placed INSERVICE.
• If the connection is currently RELEASED, the status of the connection is set OUTSERVICE and it cannot be used until it is INSERVICE again.
• The status of the underlying sessions for a connection is always the same as that for the connection itself.

For an ISC system, the following occurs:

• LU6.1 system—all underlying sessions owned by the system are released and placed out of service: immediately if PURGE or FORCEPURGE is specified; or when tasks have terminated if neither PURGE nor FORCEPURGE is specified.

  If the response to an INQUIRE CONNECTION command shows OUTSERVICE, it does not imply that the connection has been explicitly set as SET OUTSERVICE; in particular circumstances, you cannot reinstall this connection.

• APPC system—this option is valid only if the system is released. All sessions owned by the ISC system are then out of service and released.

PURGE (z/OS Communications Server only and IRC only)

Transactions running on the connected system are abnormally terminated. Transactions are terminated only if system and data integrity can be maintained. A transaction is not purged if its definition Specifies SPURGE=NO.

RELEASED (APPC only)

Specifies whether CICS is to release a session with the logical unit represented by the CONNECTION name. To get more detailed information about the availability status of the connection elements, use the CEMT INQUIRE MODENAME command.

RESYNC (MRO TO CICS TS for z/OS, Version 5.2 AND LATER SYSTEMS AND APPC ONLY) (APPC PARALLEL-SESSION, CICS-TO-CICS MRO, AND LU61 ONLY)

Any UOWs shunted because of the failure of this connection are to be retried (that is, exchange lognames resynchronization for this connection is to be attempted). This process should normally be started automatically when a connection is acquired or when a UOW is unshunted. The normal resynchronization process is to be partially overridden: decisions are taken for any units of work that are indoubt because of a failure of the connection, but the decisions are recorded and any data inconsistencies are reported when the connection is next acquired.

Note:

1. The COMMIT, BACKOUT, FORCEUOW and RESYNC operations are synchronous with setting the state of the UOW; that is, an INQUIRE UOW following SET CONNECTION BACKOUT, COMMIT, FORCEUOW, or RESYNC returns the new UOW states.
2. Specifying one of these options unshunts all units of work that have failed because of a failure of the connection. Before issuing SET CONNECTION FORCEUOW, you may want to use the SET UOW command to specify commit or backout for each indoubt unit of work explicitly, rather than letting it default, Local procedures determine the importance of the data and the method of using the INQUIRE UOW, INQUIRE UOWENQ, and INQUIRE UOWLINK commands to establish the correct actions.
3. You can specify only one of the BACKOUT, COMMIT, FORCEUOW, RESYNC, NOTPENDING, and NORECOVDATA options. SET CONNECTION NORECOVDATA should be used only in exceptional circumstances.
4. To force all indoubt units of work caused by a failure of the connection in the same direction, use SET CONNECTION COMMIT or SET CONNECTION BACKOUT.
5. The BACKOUT, COMMIT, FORCEUOW, or RESYNC options of SET CONNECTION and SET UOW do not clear resync information. If you want to do this, you must use SET CONNECTION NOTPENDING or SET CONNECTION NORECOVDATA.
6. You can issue BACKOUT, COMMIT, FORCEUOW, or RESYNC commands before issuing SET CONNECTION NOTPENDING or SET CONNECTION NORECOVDATA.