ARM policies

The Automatic Restart Manager policies are user-defined rules that control ARM functions that can control any restarts of a queue manager.

ARM functions are controlled by a user-defined ARM policy. Each z/OS® image running a queue manager instance that is to be restarted by ARM must be connected to an ARM couple data set with an active ARM policy.

IBM® provides a default ARM policy. We can define new policies, or override the policy defaults by using the administrative data utility (IXCMIAPU) provided with z/OS. The z/OS MVS Setting up a Sysplex manual describes this utility, and includes full details of how to define an ARM policy.

Figure 1 shows an example of an ARM policy. This sample policy restarts any queue manager within a sysplex, if either the queue manager failed, or a whole system failed.

//IXCMIAPU EXEC PGM=IXCMIAPU,REGION=2M
//SYSPRINT DD SYSOUT=*
//SYSIN  DD *
DATA TYPE(ARM)
DEFINE POLICY NAME(ARMPOL1) REPLACE(YES)
RESTART_GROUP(DEFAULT)
ELEMENT(*)
RESTART_ATTEMPTS(0)  /* Jobs not to be restarted by ARM  */
RESTART_GROUP(GROUP1)
ELEMENT(SYSMQMGRMQ*)  /* These jobs to be restarted by ARM */
/*

For more information see:

• Defining an ARM policy
• Activating an ARM policy
• Registering with ARM

Defining an ARM policy

Set up your ARM policy as follows:

• Define RESTART_GROUPs for each queue manager instance that also contain any CICS® or IMS subsystems that connect to that queue manager instance. If we use a subsystem naming convention, you might be able to use the '?' and '*' wild-card characters in your element names to define RESTART_GROUPs with minimum definition effort.
• Specify TERMTYPE(ELEMTERM) for your channel initiators to indicate that they will be restarted only if the channel initiator has failed and the z/OS image has not failed.
• Specify TERMTYPE(ALLTERM) for your queue managers to indicate that they will be restarted if either the queue manager has failed or the z/OS image has failed.
• Specify RESTART_METHOD(BOTH, PERSIST) for both queue managers and channel initiators. This tells ARM to restart using the JCL it saved (after resolution of system symbols) during the last startup. It tells ARM to do this irrespective of whether the individual element failed, or the z/OS image failed.
• Accept the default values for all the other ARM policy options.

Activating an ARM policy

To start your automatic restart management policy, issue the following z/OS command:

SETXCF START,POLICY,TYPE=ARM,POLNAME= mypol

When the policy is started, all systems connected to the ARM couple data set use the same active policy.

Use the SETXCF STOP command to disable automatic restarts.

Registering with ARM

IBM MQ registers automatically as an ARM element during queue manager startup (subject to ARM availability). It deregisters during its shutdown phase, unless requested not to.

At startup, the queue manager determines whether ARM is available. If it is, IBM MQ registers using the name SYSMQMGR ssid, where ssid is the four character queue manager name, and SYSMQMGR is the element type.

The STOP QMGR MODE(QUIESCE) and STOP QMGR MODE(FORCE) commands deregister the queue manager from ARM (if it was registered with ARM at startup). This prevents ARM restarting this queue manager. The STOP QMGR MODE(RESTART) command does not deregister the queue manager from ARM, so it is eligible for immediate automatic restart.

Each channel initiator address space determines whether ARM is available, and if so registers with the element name SYSMQCH ssid, where ssid is the queue manager name, and SYSMQCH is the element type.

The channel initiator is always deregistered from ARM when it stops normally, and remains registered only if it ends abnormally. The channel initiator is always deregistered if the queue manager fails.