Considerations when upgrading an existing processor to a new processor type

When you upgrade from an existing processor to a new processor type, you have to bring the current processor down. You have to plan this disruptive action to minimize the potential impact on the operating systems that are running on remaining processors.

As the HCD administrator you might have to consider the following steps as part of changing a processor configuration:
  • Creating a new work IODF by initiating a change in your active production IODF. HCD will copy all processor and operating system configurations to a new work IODF and then make it available for further modification.
  • Changing the processor ID if necessary.
  • Changing the current type and model of the outgoing processor to the type and model of a new processor.
  • If necessary, updating a processor serial number and description.
  • If necessary, updating the network name and CPC name fields in the SNA address.
  • Updating a local system (processor) name if you wish to make it different from the processor ID.
  • Building an IOCDS with the I/O configuration for a new processor using the option "Write IOCDS in preparation of the upgrade".

IBM® recommends that you do not include a new processor's type and/or model into a network name in the SNA address or into a local system name in the corresponding field. IBM recommends that you introduce naming conventions that may remain the same in definitions of a future processor configuration. The use of names, not dependent on a processor type-model, will help develop consistent operational instructions and make configurations of peer processors independent of names of currently installed processors. Use the CHPID Mapping Tool (CMT).

Note:

Definitions of some interprocessor connection types include specifications of local system (processor) names of peer processors. One example may be an Infiniband-based CF connection; a definition of an Infiniband channel path must include a local system name of a processor on the other side of the CF connection. If system names in the definitions of Infiniband channel paths do not match local system names of peer processors, expect a CF connectivity problem (loss of a signal).

A mismatch could be a result of a mistake or a typo in the process of entering a local system name for a peer processor on one of the sides. You may fix an error/typo in the incorrect local system name dynamically, in nondisruptive mode. Initiate the processor configuration change dialogue against the processor configuration you need to fix, follow the dialogue of creating a new work IODF, change the value in the Local system name field, build a new production IODF, and activate in hardware and software mode as required.