You can use Live Partition Mobility to perform a stateful migration of a logical partition (LPAR) in a Cluster Server (VCS) environment. VCS supports LPAR live migration in two ways:
Requirements for high availability support for live migration through VCS commands:
The ProfileFile attribute must contain correct information. If it does not, the LPAR creation or deletion fails. VCS cannot guarantee the correctness of the ProfileFile attribute.
The ProfileFile for an LPAR resource must contain valid VIOS mappings. If it does not, and the LPAR resource fails, then VCS is not able to delete VIOS mappings. This leaves the LPAR configuration in an intermediate state.
The ProfileFile for an LPAR must be recreated for the specific physical server if it is live migrated to a physical server. The live migration might assign mapping information which is not the same as earlier ProfileFile.
If VCS encounters an error while creating or deleting an LPAR configuration or VIOS mappings, then online or offline of LPAR resource stops immediately and does not recover from the intermediate state. Administrative intervention is required when an LPAR configuration or VIOS mappings creation or deletion fails.
Some limitations for LPM apply when VCS is configured to manage high availability of LPARs.
See Limitations and unsupported LPAR features.
For more information, refer to the Cluster Server Administrator's Guide.
LPAR migration outside of VCS control
To avoid false failovers for LPAR migration outside of VCS control
If not, increase the peer inactive timeout to an appropriate value on all the nodes in the cluster before beginning the migration.
For example, to set the LLT peerinact timeout to 90 seconds, use the following command:
# lltconfig -T peerinact:9000
The value of the peerinact command is in .01 seconds.
# lltconfig -T query
Current LLT timer values (.01 sec units): heartbeat = 50 heartbeatlo = 100 peertrouble = 200 peerinact = 9000 oos = 10 retrans = 10 service = 100 arp = 30000 arpreq = 3000 Current LLT flow control values (in packets): lowwater = 40
To make LLT peerinact value persistent across reboots:
set-timer peerinact:9000
After appending the above line, /etc/llttab file should appear similar to the following:
# cat /etc/llttab set-node host1 set-cluster 1234 link en2 en-00:15:17:48:b5:80 - ether - - link en3 en-00:15:17:48:b5:81 - ether - - set-timer peerinact:9000
Some limitations for Live Partition Mobility (LPM) apply when VCS is configured to manage high availability of LPARs.
See Limitations and unsupported LPAR features.
For more information on VCS commands, see the Cluster Server Administrator's Guide.
For attributes related to migration, see the Cluster Server Bundled Agents Reference Guide.
To migrate the managed LPAR without ProfileFile support
For VCS to manage the LPAR, the profile is required on the managed physical system of the management VCS that is part of the system list of the LPAR resource.
If migration succeeds, the profile on source is removed. Restore and rename the LPAR profile from the backup that was taken in step 1. Remove the renamed LPAR profile on the destination.
If migration fails, remove the backup profile on the source. On the destination, rename the renamed LPAR profile to original LPAR profile.