Step 8: Configure ADNR to automatically restart in case of application or system failure (optional)
Automatically restarting ADNR minimizes the period of time that name servers do not accurately reflect the status of sysplex resources. This can be accomplished using automation software or by defining an automatic restart manager (ARM) policy. For more information about defining ARM policies, see z/OS MVS Setting Up a Sysplex.
ADNR registers with ARM using the following values:
ELEMTYPE=SYSTCPIP
ELEMNAME=EZBADNRelemsuffix
TERMTYPE=ALLTERM
The elemsuffix value is specified in the arm_element_suffix configuration statement. ADNR registers the element name EZBADNR concatenated with the elemsuffix value. If there is no arm_element_suffix statement, the ELEMNAME is EZBADNR. For example, if the following statement is configured:
arm_element_suffix SYS1
ADNR registers ELEMNAME=EZBADNRSYS1.
When ADNR registers with ARM using these values, then if ADNR fails or the system fails, ADNR can be restarted on any system in the sysplex.
An ARM policy or other automation software should be in place to quickly restart the TCP/IP stack that ADNR is running on if the TCP/IP stack fails. ADNR continues to run after its TCP/IP stack fails, and reconnects to its GWM after the TCP/IP stack recovers.
Although this step is optional, performing it provides high availability to your target applications. In the event that ADNR fails, the name servers it manages are no longer updated with sysplex resource availability information. As a result, client connections might fail because the name server might resolve their requests to application instances that are not available, or be unable to resolve DNS requests to any application when active application instances might actually exist. When ADNR is restarted, it again accurately reflects the availability of sysplex resources after its convergence period. The convergence period is twice the interval at which the Advisor normally updates ADNR with new information. The interval is determined by the update_interval statement in the Advisor's configuration file.
- Each ADNR instance should have a unique ARM element name within a sysplex.
- Establish an ARM policy with TCP/IP at a lower level than ADNR, so that TCP/IP is restarted before ADNR is restarted. For more information, see z/OS MVS Setting Up a Sysplex.
- To enable ADNR to continue operating on another TCP/IP stack in a Common INET (CINET) environment in the case of TCP/IP stack failure, or to restart on another system in case of system failure, configure ADNR with a unique application-instance DVIPA. The unique application-instance DVIPA is coded on the host_connection_addr keyword of the gwm statement. For more information about unique application-instance DVIPAs, see Configuring the unique application-instance scenario.
RDEFINE FACILITY IXCARM.SYSTCPIP.EZBADNR* UACC(NONE)
PERMIT IXCARM.SYSTCPIP.EZBADNR* CLASS(FACILITY) ID(ADNR) ACCESS(UPDATE)
SETROPTS REFRESH RACLIST(FACILITY)