HP StorageWorks SAN Virtualization Services Platform 2.1 release notes (5697-8060, March 2009)

Solaris host is running I/O to a virtual disk and the command is applied on that virtual disk, the Solaris
host reactivates the virtual disk and the virtual disk does not become active on the other DPM.
Rescan is required after deleting or unpresenting virtual disks
Hosts may not automatically detect that a virtual disk has been deleted or unpresented. A rescan
should be performed on the host after any virtual disks to that host have been deleted or unpresented.
This is required before any new virtual disks, especially any with the same LUN as a
deleted/unpresented virtual disk, are presented to the host.
Period of high back-end load after a DPM reboot when synchronous mirror virtual disks are
used
In a configuration with synchronous mirror virtual disks, there may be a period of higher than normal
load on the back-end storage arrays after a DPM reboot. This will vary depending on the capacity
of the synchronous mirror virtual disks and with the load and write patterns on those virtual disks
before the DPM reboot. To reduce performance impact on production activity, HP recommends that
DPM reboots (for example, during a DPM upgrade) be scheduled for off-hours or during a maintenance
window.
I/O errors/timeouts when performing failover on large numbers of synchronous mirror virtual
disks
If there is a heavy load of I/O on a large number of synchronous mirror virtual disks and an event
occurs that requires a large number of virtual disks to be failed over at the same time (for example,
a reboot of a DPM), it is possible that I/O errors or timeouts may be returned to the hosts. HP
recommends that operations requiring a DPM reboot (for example, a DPM firmware upgrade),
particularly in a large environment, be scheduled for off-hours or during a maintenance window, or
manually fail over each synchronous mirror group, one at a time.
DPM ports in state 'Signal' rather than 'Link Established' when using McData switches
There is an interoperability issue with some McData switches that causes the DPM ports to not properly
establish a link with the switch. To work-around this issue, the port speed of the DPM port and the
McData switch port should be manually set to the desired speed instead of relying on automatic speed
negotiation.
DPM show status CLI command does not work
The show status CLI command prints an error ‘stack trace.’ This command should not be used.
DPM show rport CLI command does not display matching WWPNs when the WWPN
provided as an argument contains hexadecimal digits
The show rport command incorrectly converts any arguments to all capital characters. This causes
the message No matching wwpns found even when that WWPN has been discovered by the
system.
14