Cost-Effective High-Availability Solutions with HP Instant Capacity on HP-UX

The following two sub-sections discuss how these ideas relate to managing usage rights—based on
whether some or all virtual partitions are down.
Some virtual partitions down (migrate rights with vparmodify and icapmodify)
If a subset of virtual partitions within the nPartition are down and you want to distribute those usage
rights to other partitions in the group, you must use techniques other than rights seizure. To release
core usage rights from a down virtual partition, use vparmodify and icapmodify commands.
You can use this technique as long as the static attribute is not set for the virtual partitions that are
down, and as long as an active virtual partition has management privilege. The steps are:
3. From another virtual partition in the same nPartition, remove unused capacity from each down
virtual partition using the following command:
vparmodify -d cpu::<n> -p <down_vpar>
where, <n> is the number of cores to remove from the specified down virtual partition.
4. Release the usage rights from the nPartition using the following command on an active virtual
partition in the same nPartition:
icapmodify -d <n>
where, <n> is the sum of all the cores removed from virtual partitions in step 1.
5. Activate cores in any other partition of the GiCAP group using the released core usage rights by
running the following command on that partition:
icapmodify -a <n>
Using icapmodify ensures that you maintain compliance with the Instant Capacity contract.
All virtual partitions down (seize rights; restore rights before booting virtual partitions)
If all of the virtual partitions for an nPartition are down, you can seize rights from that nPartition.
In this case, you can seize usage rights by specifying the host name for any of the virtual partitions in
the nPartition as the target of the rights seizure command (icapmanage -x). (Issuing a second
seizure command targeting another virtual partition in the same nPartition is rejected because the first
seizure command seizes the maximum rights possible from the nPartition.)
When usage rights are seized from an nPartition hosting failed virtual partitions, once the failures
have been corrected, you must restore usage rights to the virtual partitions before attempting to reboot
them. In the case of a failure that was a partial outage (at least one host in another nPartition still
running), the rights seizure would have been committed immediately, resulting in an immediate
reduction of the number of usage rights for the failed virtual partitions. In the case of a failure that was
a complete server outage, the first connection to the Group Manager (typically during the boot
process) commits the rights seizure, resulting in a similar reduction of the number of usage rights. In
either case, the virtual partitions will likely not be bootable because the minimum number of core
usage rights may not be sufficient depending on the assignments in the vPars partition database.
To avoid this problem, the failback process for virtual partitions must include a restore operation. Before
rebooting any failed virtual partition, you must revert any seized usage rights in a two-step process:
1. Deactivations elsewhere in the group (typically, deactivating the cores that were activated on the
failover node).
2. Execution of a restore command for the partition designated by the rights seizure operation.
The restore command ensures that the usage rights are transferred back to the previously
failed nPartition.
24