Migration phase issues – HP 3PAR Operating System Software User Manual
Page 19
Cause 3
When any of the peer ports has the wrong source system device connected.
Solution 3
Rezone the switch fabric so that the destination system peer ports are zoned to the correct source
system device.
Migration Phase Issues
•
Verify the destination system has no active paths with the showhost CLI command.
•
Monitor the migration task with the showtask CLI command. Determine where the operation
failed.
•
If a failure occurs before the volume import stage, but after volumes have been admitted on
the destination system, it is possible to manually return the system to the pre-admit state. This
process is non-disruptive to the hosts provided the appropriate zoning and host multipathing
is performed. The host must have access to the volume through the source system.
nl
To return the system to its state before a volume admit failure:
1.
On the fabric and host — If the host has already been unzoned from the source system
then rezone it back to that system and confirm that I/O is occurring on the source system.
2.
On the fabric and host — Unzone the host from the destination system, verify that all
access to the volumes is now only through the source system.
3.
On the destination system — Remove the VLUNs on the destination system for the volumes
admitted to the hosts.
4.
On the destination system — Remove the admitted volumes from the destination system.
5.
On the destination system — Remove the host from the destination system.
6.
On the source system — Remove the VLUN exports to the destination system from the
source system.
7.
On the source system — Remove the host representing the destination system from the
source system.
•
If a failure occurs after volume import tasks have started the hosts access the volumes on the
source system is interrupted. A failed import returns the system to the point where the import
can be retried after the cause of failure is resolved. A full rollback to the system state before
the volume import failure is possible.
nl
To return the system to its state before a volume import failure:
1.
On the host — Cleanly stop access to the destination system from the host. The host will
lose access to the volumes being migrated. To prevent consistency issues any active
applications should be cleanly shutdown.
2.
On the destination system — Remove the VLUNs on the destination system for the volumes
admitted to the host.
3.
On the destination system — Remove the admitted volumes from the destination system.
4.
On the destination system — Remove the host from the destination system.
5.
On the source system — Remove the VLUN exports to the destination system from the
source system.
6.
On the source system — Remove the host representing the destination system from the
source system.
7.
On the source system — Issue the setvv -clrrsv CLI command for the volumes that
were being migrated.
8.
On the fabric and host — Rezone the host back to the source system.
9.
On the fabric and host — Restart the host and any applications that were shutdown at
the beginning of this process.
Migration Phase Issues
19