HP StorageWorks 4/8 - SAN Switch Release Note page 28

Cold panic occurs due to race condition when multiple requests are
being issued throughout the fabric. Restarts with no impact to
host/target traffic.
Enhancements needed to supportsave to provide additional
troubleshooting information.
Zoning operations will not propagate to backbone router from edge
fabric when performed while the fabric is building.
Buffer-limited ports were not persistent across reboots, such that they
do not stay buffer limited and other ports become buffer limited.
Credit recovery does not reallocate credits properly when QoS is ON,
and QoS E_Port does not recover credit via link reset, so traffic may
not run properly through the fabric.
Misbehaving third-party devices can cause soft fault entries per SN
failures. Need to minimize these entries.
Some NPIV-capable HBAs have problems talking to the switch due to
FDISC occurring after a successful ACC of NPIV establishment was
not supported. Leads to ABTS being sent to the wrong N_Port ID.
After removing a Long Distance EX_Port configuration and doing
portenable to Long Distance (non-EX_Port), the FCR-1063 message
is displayed followed by VERIFY being displayed on the console.
Channel receives bad LS_RJT reason code of 0x0B00 rather than ex-
pected 0x092c on attempt to register LIRR with Registration Function
Always Receive.
Kernel access of bad area panic from NSD task due to device going
away in a small timing window when switch is in process of doing
PLOGI. Both CPs could hit the same panic during this small window.
Switch panic due to detected termination of ficud, due to access of
uninitialized data.
Switch panic due to management server daemon (MSD) data structure
corruption during hafailover/hareboot.
After a specific hafailover sequence (disable FCR before
hafailover, then do hafailover, then another hafailover,
then enable FCR again), the switch encounters a frame drop on
EX_Port.
The command firmwareupgrade fails (does not commit) during a
very small window of time, when active CP did not wait for standby
CP to timeout before recovering it.
Performance monitor filter block is processed in the wrong block size,
leading to memory corruption, because the trace buffer filled up. Could
result in switch panic.
Port state mismatch seen on embedded switches between switchshow
and portcfgshow regarding PersistenDisable flag. Portcfg-
show shows persistent disable state while port is actually enabled.
28
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2
Fixed in Fabric OS 6.1.2