ITEM: AG5773L

HACMP question regarding reintegration.

Question:

Env:  2 RS6000 Model390 at AIX 3.2.5 in an HACMP 2.1 Mutual Take-over
cluster with 7134 as the shared dasd.  The two nodes share two SCSI2
DE F/W buses.   Each bus has 6 SCSI2 DE F/W 7134 disks on it.  The
SCSI IDs for the disks are 0-5 and the external SCSI IDs for the SCSI2
DE F/W adapters on NodeA are set to 6 and the SCSI IDs for the SCSI2
DE F/W adapters on NodeB are set to 7.  Thus there aren't any
conflicting SCSI IDs on the two shared busses.

Desc:  The nodes are not reintegrating properly because the PVIDs of
the shared PVs are not in the ODM.  During the fail-over testing of
the cluster, the customer removed all of the shared hdisks and then
ran configuration manager to rebuild them into the odm.  All of the
disks were identified and configured into the ODM and made Available
but the PVIDs couldn't be read because the reserve bit was set by the
node that was already up.  (Note:  the reserve bit is set by varying
on a volume group)

Act: The problem was resolved by performing the following steps:

 1. remove all of the shared disks from both nodes.
 2. run configuration manager on NodeB to build correct hdisk
    entries into the ODM with the PVIDs.
 3. vary on the shared VGs onto NodeB to set reserve bit on shared
    hdisks
 4. vary off the shared VGs onto NodeB to reset reserve bits on
    shared hdisks.
 5. run configuration manager on NodeA to build correct hdisk
    entries into the ODM with the PVIDs.

Once the above steps were completed, I had the customer bring up HACMP
on both nodes and perform fail-over tests from NodeA to NodeB,
reintegrate NodeA, fail-over NodeB to NodeA, and reintegrate NodeB.

All works as designed.

Closing Item!



Support Line: HACMP question regarding reintegration. ITEM: AG5773L

Dated: March 1995 Category: N/A

This HTML file was generated 99/06/24~13:30:27

Comments or suggestions?
Contact us