Because server instances register with the default restart
group, automatic restart management (ARM) attempts to restart the
entire default group on another system in the sysplex when a system
failure occurs. To create a restart group other than this default
group, you need to follow certain rules.
If you want to create a restart group other than this default group,
you need to comply with the following rules and restrictions that
apply for z/OS® ARM policies. For more information about
how to actually change these policies, see,
z/OS MVS™ Setting
Up a Sysplex (SA22-7625).
- To change the policy, you need to know the existing element names
for the server instances and how to name new elements for additional
instances. The element names for these server instances are formed
by concatenating the cell short name and the servers specific short
name.
If you have a cell named PLEX1 and server named BBOS001, for
example, the ARM element name would be PLEX1BBOS001.
Since wildcard
characters can be used in the ARM policy, it is possible to exclude
an entire group of servers by using a common naming scheme within
your cell.
The following section of the ARM policy will prevent
any node agents from starting, for example, assuming that each node
agent in your cell has a name that adheres to the form BBON
xxx:
RESTART_GROUP(WEBSPHERE)
ELEMENT(PLEX1BBON*)
RESTART_ATTEMPTS(0,150)
RESTART_TIMEOUT(600)
READY_TIMEOUT(1200)
TERMTYPE(ALLTERM)
RESTART_METHOD(BOTH,PERSIST)
This ARM policy will also
prevent the node agent from restarting in place. This specification
can be modified by changing the RESTART_METHOD and TERMTYPE parameters.
See z/OS MVS Setting
Up a Sysplex (SA22-7625) for more information.
- If you create a restart group, keep the following in the same
restart group and set the restart order for the elements as indicated:
- RRS
- DB2® with IRLM, IMS™, CICS®,
and other transaction or resource managers if used by your application
servers in the restart group
- Your server instances
Either set up the location service daemon and node agent for
restart-in-place or remove them from your ARM policy. Since WebSphere® Application Server must be running
on all systems that might be used to perform recovery, the application
servers will use the location service daemon and node agent that are
already running on the alternate system. If the location service daemon
attempts to restart on the alternate system, it will fail. If the
node agent restarts on the alternate system, it will have no recovery
work to do.