HP StorageWorks Fabric OS 5.3.0b Release Notes (AA-RWEYF-TE, October 2007)

Zoning considerations
with FC4–48 Blade
B-Series switches running pre-5.2.0 firmware do not accep t zone databases with
Domain, Port Index, references to ports higher than 255. Zoning in Fabric OS
5.2.0 and later, therefore, do not allow a Domain, Port Index entry for ports
above 255 to be created unless all switches in the fabric are at
Fabric OS 5.2.0 or later versions. Once such an entry has been put into the zone
database, an attempt to connect a pre-Fabric OS 5.2.0 switch to the fabric results
in that switch being segmented with a Zone Mismatch segmentation reason.
Any zoning that is already in use can continue to be used. However, in Fabrics
containing pre-Fabric OS 5.2.0 switches, if new devices are added on the upper
16 ports (ports 32-47) of the FC4-48 port Blade, then WWN zoning h as to
be used for those devices. If there is an existing Domain Port Index zone that
connects to ports 32–47, HP recommends that you transform just that zone to
WWN zoning. This would minimize the likelihood of rebuilding the entire fabric.
Fabrics containing only Fabric OS 5.2.0 an d later switches have no zoning
limitation.
Update to Brocade
4Gb SAN Switch for
HP c-Class BladeSystem
installation instructions,
part number 5697–5678
On page 2, please update the following step:
nl
7. At the command line, type:
nl
c bay x
nl
Where x is t
he I/O bay where the switch is installed.
nl
to
nl
7. At the command line, type:
nl
connect in
terconnect x
nl
Where x is the I/O bay where the switch is installed.
iSCSI
• Enterpris
e storage array targets may not show up consistently in the disk
management window of an iSCSI initiator when the same LUNs are mapped
to two dif
ferent virtual targets (VTs). When the initiator logs in to the VTs, the
targets keep disappearing from the disk management window. If the session
to one of
the VTs is disconnected, the remaining VT appears and stabilizes in
the disk
management window.
• Under c
ertain conditions, hosts on an IP network may not be able to issue
apingc
ommand to iSCSI gateway ports in another subnet. You can work
around this problem by issuing a ping command from the iSCSI gateway
port.
The hosts will then be able to successfully issue ping commands to the
iSCSI
gateway port.
• An IP
network disconn ection lasting fivesecondsormoremaycauseCOPA
failure on the disconnected PC, which in turn may cause a loss of c onnection
to an enterprise storage array
• If traffic is run from hosts to certain tar gets with severe impairment conditions
in t
he IP network for hours at time, throughput to the targets will drop, and may
take up to 10 minutes to recover after the impairment condition is removed .
Note that this problem is highly intermittent, and is unlikely to be seen in a
cu
stomer environment. We believe that this issue is the result of host/target
interaction, and is not the result of action on the iSCSI gateway.
• A Microsoft windows PC host was unable to discover enterprise storage array
L
UNs. The Microsoft iSCSI initiator is able to discover targets if the software is
uninstalled and reinstalled without the MPIO option.
Broadcast Zones
In Fabric OS 5.3.0 and later, a zone with the name “broadcast” (case-sensitive)
is a special zone for setting up recipients of broadcast packets. In Fabric OS
versions earlier than 5.3.0, a zone named “broadcast” does not h ave special
significance.Therefore,youmustmakesurethatifabroadcastzoneisconfigured,
thentheactiveandstandbyCPsusethesameFabricOSversion.Otherwisean
HA failover might change the zone configuration.
HP StorageWorks Fabric OS 5.3.0b release notes
15