PowerHA: How to Prepare a "Cloned" System Before Adding Into Existing Cluster Device Domain.

Definition of a clone as considered by IBM i: A system where *SYSBAS has been copied using some storage mechanism, ie. PPRC or Flashcopy or some equivalent.

If the clone was done from a system that only contains SYSBAS, then there should be nothing to fear.

If the clone was done from a system that contained a cluster and/or an IASP already, then the cloned system should be cleaned up prior to joining a cluster.

To ensure proper clean up of the cloned system prior to adding into an existing cluster perform the following steps.

************************************************************************************************************************

Starting in R720, a new parameter was introduced in the CFGDEVASP command that could be leveraged to prep for a cloned node.

Steps for R720 and above:

1. After the clone has been performed, use WRKCLU or DSPCLUINF to determine if there is a cluster.  If there is a cluster, then ensure that everything in that cluster is already considered Inactive and use the DLTCLU command to get rid of that cluster information.

2. Use iNav, Navigator for IBM i or SST to determine if there is in IASP resource in existence.  If any IASP exists then it can be deleted by using iNav, Navigator for IBM i or using the command CFGDEVASP with the *DELETE option.

3. Next, since there should no longer be an IASP resource present, use the following command:

************************************************************************************************************************

Steps for R710 and older:

1. Use WRKCLU or DSPCLUINF to determine if a cluster is present on the cloned system. If a cluster exists use ensure that everything is Inactive in the cluster, then use the DLTCLU command to delete the cluster.

2. Use iNav, Navigator for IBM i or SST to determine if there is in IASP in existence. If any IASP exists then it can be deleted by using iNav, Navigator for IBM i or using the command CFGDEVASP with the *DELETE option.

After the above items have been deleted, there are a few more steps need to ensure that the system is considered clean.

The following steps perform the creation of a dummy cluster and device domain on the cloned system as a stand alone in it's own cluster, before properly removing and being ready to join into an actual cluster.

Some of the steps may be considered overkill but will ultimately result in the cleanup without any accidental side effects.

3. CRTCLU CLUSTER(TEMPCLU) NODE((NODEA (XX.XX.XX.XX))) - where TEMPCLU could be anything, NODEA could technically be anything and XX.XX.XX.XX would be a valid IP address to be used as a cluster heartbeat.

4. WRKCLU select option 6 to ensure that NODEA is Active. If not, select option 8 to Start it. Once Active proceed to next step.

5. ADDDEVDMNE CLUSTER(TEMPCLU) DEVDMN(TEMPDEVDMN) NODE(NODEA)

6. WRKCLU select option 7 to ensure that NODEA is the only active member of device domain TEMPDEVDMN.

7. RMVDEVDMNE CLUSTER(TEMPCLU) DEVDMN(TEMPDEVDMN) NODE(NODEA)

8. DLTCLU CLUSTER(TEMPCLU)

9. At this point, the system should be considered clean and is ready to be added into an existing cluster. For peace of mind an additional step could be to gather Advanced Analysis macro DSMINFO -ACR and then ensure that there is only an entry that has "ACR Entry: 1".

If there is any entries of "ACR Entry: 33" or any number higher, then the system is not yet "clean"

*NOTE - The ONLY time a system can have an ACR Entry of 33 or higher and be considered acceptable is if the system does legitimately have an IASP configured and is going to be the very first system added into the cluster device domain.