HP StorageWorks Fabric OS 6.3.1c Release Notes (5697-0509 September 2010 - includes all 6.3.1x versions)
Fixed in Fabric OS 6.3.1b
After 1606 extension switch reboot, PLOGI aborts seen due to F_RJT
for the ACCEPT frame. Host cannot see the target if PLOGI comes in
before RSCN is propagated on the 1606 extension switch or multipro-
tocol extension blade VE ports due to backend port programming not
completing in time.
Fixed in Fabric OS 6.3.1b
In a rare corner case, the encryption switch or encryption blade does
not handle an error case correctly when key attribute retrieval from key
vault fails, leading to a panic.
Fixed in Fabric OS 6.3.1b
With two FCRs (routers or switches with IR functionality) in the backbone
and both connected to the same edge fabric through EX ports, and
edge fabric has MEOS in IM2 mode, switch can panic upon receiving
periodic GE_PT from MEOS switches due to the FCR not handling it
properly.
Fixed in Fabric OS 6.3.1b
Internal ports of of 8Gb c-Class switch need to be forced to arb/arb
for portcfgfillword (mode 1) to improve characteristics for future
mezzanine card.
Fixed in Fabric OS 6.3.1b
HP 1606 extension switch and multiprotocol extension blade do not
always auto-restart after detection of a critical error when running
Fabric OS 6.3.1a. A manual reboot may be required to recover the
1606 switch and a slot power off/on to recover the multiprotocol exten-
sion blade.
Fixed in Fabric OS 6.3.1b
Supportsave hangs after unplugging and then plugging in the network
cable on the encryption switch or blade due to a race condition that
causes the encryption switch cluster to update continuously and sup-
portsave not to finish.
Fixed in Fabric OS 6.3.1b
Workaround prior to upgrade: In
environments with devices that rely
on the area of the PID to match the
port number, users should bind PIDs
using the portaddress -bind
[slot_number/]port_number [16-
bit_address] command prior to
allowing devices to log in to the
switch.
On HP 8/8, 8/24, 8/40 and 8/80 switches shipped with Fabric OS
6.3.1a factory installed, the assignment of PIDs is non-deterministic.
The area field will be assigned based on the order that ports are recog-
nized and brought up by the system the first time (and only the first time)
it is booted. This can impact servers that need their area equal to the
port numbers to login to the fabric successfully. This does not impact
switches that upgrade to 6.3.1a in the field.
Fixed in Fabric OS 6.3.1b
Host that implements strict FCP/FCP-2 protocol frame fields checking
for EOF delimiter, fails to discover LUNs when more than 64 LUNs are
added to the encryption switch or encryption blade from the storage
array.
Fixed in Fabric OS 6.3.1b
Port based routing frames are out of order when FCP image pairs con-
currently initiate exchanges as originators over GE ports. Leads to very
poor performance over FCIP link between two 1606 extension switches
or multiprotocol extension blades over 1G GE port.
HP StorageWorks Fabric OS 6.3.1c Release Notes 89