HP 3PAR Cluster Extension Software Administrator Guide (5697-0925, May 2011)

NOTE: Host persona setting varies based on the Windows operating system installed on the
cluster nodes, For more information, see the HP 3PAR Implementation Guide.
fail_wrt_on_err and no_fail_wrt_on_err settings
The fail_wrt_on_err and no_fail_wrt_on_err policies apply only to synchronous Remote
Copy. The fail_wrt_on_err group policy ensures that the two copies are always in
synchronization by refusing to write changes to a Remote Copy volume group on one server when
the links between the members of the Remote Copy pair are broken for any reason. This prevents
the primary and secondary volume groups from becoming out of synchronization with each other.
This policy can be set for disaster recovery applications where the Remote Copy must be up-to-date
at all times, and where an application outage for a brief period can be tolerated (for example if
the links or one member of the Remote Copy pair fails).
The no_fail_wrt_on_err group policy (default policy) allows writes to a primary Remote Copy
volume group even though the link between the systems in the Remote Copy pair is broken.
Continuing to write changes despite a broken link leads to a loss of synchronization between
primary and secondary Remote Copy volume groups. The no_fail_wrt_on_err policy is used
when you would rather keep the applications that are writing the data running, even when the
Remote Copy has failed and become out of sync. After the error that caused the broken link is
resolved, you must manually restart the Remote Copy operation between the members of the Remote
Copy pair to bring the Remote Copy in sync again.
NOTE: HP 3PAR Cluster Extension supports only the Remote Copy volume groups with
no_fail_wrt_on_err policy. For information on setting the Remote Copy volume group policies,
see the “Remote Copy Commands” section in HP 3PAR Remote Copy User’s Guide .
Remote Copy and 3PAR Virtual Domains
Remote Copy checks HP 3PAR Virtual Domains (domains) on the remote system to ensure that the
virtual volume is mirrored to the same domain name as the local domain name. The domain needs
to be called a correct domain name. For more information on domains, see the HP 3PAR Concepts
Guide and for instructions on setting up domains, see the HP 3PAR CLI Administrator’s Manual.
NOTE: When configuring the HP 3PAR storage system password file for Cluster Extension usage,
user in the password file should have access to all the domains of Remote Copy virtual volumes
managed by Cluster Extension.
Promote issue
When the data copy is going on between the primary and the secondary Remote Copy volume
groups, if the Remote Copy link is broken for some reason, the Remote Copy volume groups go
to the stopped state and the snapshots of the secondary volumes start getting promoted to the
base volumes. This is the design of 3PAR Remote Copy. This activity may take some time to complete.
At this time, if the start or restore operation is attempted on the Remote Copy volume groups,
then this operation may fail with the error Promote operation is going on. At this time, if
the local replication role is secondary, the remote replication role is primary, and the remote copy
link is up, Cluster Extension executes stop, reverse, and start operations for the Remote
Copy volume group. If the stop and reverse operation succeed, then the secondary volumes
become read-write and Cluster Extension resource comes online even though the start operation
may fail. If the start operation fails, replication I/O does not start from the new primary volumes
to the new secondary volumes. But Cluster Extension will continuously attempt to start the group
during the monitoring interval of the Cluster Extension resource. At this time, if the local replication
role is secondary-rev, the remote replication role is primary-rev, and the remote copy link
is up, the restore operation fails due to promoting of the snapshot, and the Cluster Extension
resource does not come online.
58 Troubleshooting