RRS automation and recovery scenarios

(ZOS) RRS automation and recovery scenarios

Use this table for RRS automation and recovery scenarios.
Task RRS automation and recovery scenarios
Startup
Ensure System Logger has been started before RRS.
RRS will display error messages indicating that System Logger must be started first if we try to start RRS without starting System Logger. Ensure RRS is started before WebSphere for z/OS . RRS does not start by itself. RRS will start automatically only if it was registered with the Automatic Restart Manager (ARM) and if ARM is running. To start RRS, issue the start command:
start atrrrs,sub=master
RRS doesn't restart itself if we issue the cancel command, so we need to restart it manually if it was canceled or if ARM isn't running.
Shutdown
Shutdown RRS in the reverse order that we started RRS. Shutdown WAS for z/OS, then RRS, followed by System Logger. There is no controlled way to bring down RRS. The best approach is:

Quiesce WAS for z/OS.
Shutdown WAS for z/OS.
Cancel RRS.
You may want to bring down the DB2 we are using for WAS for z/OS before canceling RRS.
To cancel RRS, issue the command:
setrrs cancel
Handling in-flight and indoubt transactions if there is a failure Refer to the RRS system management panels to display in-flight and resolve indoubt transactions. We can display the resource managers on the RM panels in RRS, display all units of recovery (UR), filter the URs, and then resolve the indoubts. We cannot resolve in-flights. We can display all RRS-managed transactions.
If we are using the IMS™ Connector for Java, this process applies only if IMS Connector for Java, IMS Connect, and the IMS subsystem are configured locally on the same z/OS system image on which the WebSphere for z/OS J2EE server runs. The local configuration is the only configuration in which IMS Connector for Java runs as an RRS-transactional connector.
How to determine if RRS is running
Use the display command:
d a,atrrs
atrrs is the name of the default RRS procedure shipped with WAS for z/OS. Use the procedure name used to start RRS. The address space comes from the procedure.
What happens to WebSphere for z/OS if RRS goes down? RRS is a required subsystem, so WAS for z/OS will not run without it. If RRS goes down, WAS for z/OS will get fatal errors. We need to get RRS started, then restart WAS for z/OS.
What happens to other subsystems if RRS goes down? RRS is the WAS for z/OS transaction monitor. If we cancel RRS, we will have problems with any subsystems using it (for example, WAS for z/OS, DB2, IMS). We should understand the implications before we cancel RRS.
Where to find more information z/OS MVS™ Programming: Resource Recovery
Diagnose problems (using diagnosis tools)
Automation and recovery scenarios and guidelines

Task	RRS automation and recovery scenarios
Startup	Ensure System Logger has been started before RRS. RRS will display error messages indicating that System Logger must be started first if we try to start RRS without starting System Logger. Ensure RRS is started before WebSphere for z/OS . RRS does not start by itself. RRS will start automatically only if it was registered with the Automatic Restart Manager (ARM) and if ARM is running. To start RRS, issue the start command: start atrrrs,sub=master RRS doesn't restart itself if we issue the cancel command, so we need to restart it manually if it was canceled or if ARM isn't running.
Shutdown	Shutdown RRS in the reverse order that we started RRS. Shutdown WAS for z/OS, then RRS, followed by System Logger. There is no controlled way to bring down RRS. The best approach is: Quiesce WAS for z/OS. Shutdown WAS for z/OS. Cancel RRS. You may want to bring down the DB2 we are using for WAS for z/OS before canceling RRS. To cancel RRS, issue the command: setrrs cancel
Handling in-flight and indoubt transactions if there is a failure	Refer to the RRS system management panels to display in-flight and resolve indoubt transactions. We can display the resource managers on the RM panels in RRS, display all units of recovery (UR), filter the URs, and then resolve the indoubts. We cannot resolve in-flights. We can display all RRS-managed transactions. If we are using the IMS™ Connector for Java, this process applies only if IMS Connector for Java, IMS Connect, and the IMS subsystem are configured locally on the same z/OS system image on which the WebSphere for z/OS J2EE server runs. The local configuration is the only configuration in which IMS Connector for Java runs as an RRS-transactional connector.
How to determine if RRS is running	Use the display command: d a,atrrs atrrs is the name of the default RRS procedure shipped with WAS for z/OS. Use the procedure name used to start RRS. The address space comes from the procedure.
What happens to WebSphere for z/OS if RRS goes down?	RRS is a required subsystem, so WAS for z/OS will not run without it. If RRS goes down, WAS for z/OS will get fatal errors. We need to get RRS started, then restart WAS for z/OS.
What happens to other subsystems if RRS goes down?	RRS is the WAS for z/OS transaction monitor. If we cancel RRS, we will have problems with any subsystems using it (for example, WAS for z/OS, DB2, IMS). We should understand the implications before we cancel RRS.
Where to find more information	z/OS MVS™ Programming: Resource Recovery