Hp McDATA 4Gb SAN Switch for p-Class BladeSystem release notes

McDATA 4Gb SAN Switch for HP p-Class BladeSystem release notes

*AA-Rw1ZH-TE*

Part number: AA-RW1ZH-TE

eventh edition: March 2 008

Description

These release notes contain late breaking and supplemental information for the McDATA 4Gb SAN Switch. Be

sure to read these release notes prior to installing the M cDATA 4Gb SAN Switch. This document is intended for customers who purchased the McDATA 4Gb SAN Switch. This information is periodically updated and

available on the HP website: h

ttp://h18006.www1.hp.com/storage/saninfrastructure.html.

From the web BladeSystem, then select Manuals.

site, select Retired Products, then navigate to the McDATA 4Gb SAN Switch for HP p-Class

Update recommendation

Routine

Firmware version

Firmware 6.4.0.13.00 is the latest (as of this date) ﬁrmware released with the McDATA 4Gb SAN Switch. Th

saninfras

From the website, select Retired Products, then navigate to the McDATA 4Gb SAN Switch for HP p-Class BladeSystem,then select Download drivers and software.

elatest ﬁrmware is available on the HP website: h

tructure.html.

Product models

McDATA 4Gb SAN Switch, part number A 8001A

Devices supported

Supported conﬁgurations

ttp://h18006.www1.hp.com/storage/

For supported conﬁgurations, see the following resources:

• The BladeSystem support matrix, available on the following HP website:

ttp://h18006.www1.hp.com/products/storageworks/bladesystemmatrix/index.html

• The HP StorageWorks SAN design reference guide, available on the HP website:

ttp://www.hp.com/go/sandesignguide.

Cable requirements

measuring up to 500 meters (1-Gb/s), or 300 meters (2-Gb/s), or 150 meters (4-Gb/s),

For cab use transceiver modules in a switch.

Verify that connectors interfacing with the M cDATA 4Gb SAN Switch use LC Duplex connectors with a PC depen

les

/125 multi-mode FC cables. Multi-mode optical cables are connected to short-wave o ptical

ﬁnish. In addition, the connector at the opposite end of the cable must be of either LC or SC type,

ding on the requirements of the connected device.

Operating systems

For a list of supported operating systems, see the HP StorageWorks SAN design reference guide, available on the HP website: h

ttp://www.hp.com/go/SANdesignguide.

McDATA 4Gb SAN Switch for HP p-Class BladeSystem release notes

Languages

English

Fixes

This section describes the ﬁxed i ssues related to the McDATA 4Gb SAN Switch.

Fixes in switch

ﬁrmware 6.4.0.13.00

Paths to devices reconnected to Mi10K director sometimes did not recover

When the McDATA 4Gb SAN Switch was connected to an Mi10K director, and the connection of a device to the Mi10K was interrupted and then reconnected, paths that existed between that device and a device connected to the McDATA 4Gb SAN Switch sometimes would not be recovered. This was caused by an error in the McDATA 4Gb SAN Switch with processing the Address Format bits of the

SW_RSCN (Inter-Switch Registered State Change Notiﬁcation) from the Mi10K. This could occur only if the fabric was in Open Fabric/Standard interoperability mode, and did not occur in McDATA Fabric/Homogeneous interoperability mode.

Certain erroneous responses caused the switch SNMP services to become unresponsive

When the switch enc ountered erroneous responses to the ﬁbre channel commands, GIELN (Get Interconnect Element Logical Name) or GMAL (Get Interconnect Element Management Address List), the switch some become unresponsive. When this occurred, the switch reported this as a Firmware error (Blink

2).

times did not properly handle the errors which caused the internal switch SNMP services to

Fixes in switch ﬁrmware 6.4.0.11.00

Devices di

dnot always logbackintoswitchfollowing ResetorHardReset

When a Reset or Hard Reset is executed in the switch, c onnections to devices attached to switch ports are normally interrupted as they are logged out. When the switch is in McDATA Fabric interoperability mode,

or H a rd Reset was executed in the switch, there were instances where devices connected to

and a Rese the

switch did not succeed in log ging back in. This was due to an erroneous port ID in the Domain RSCN that was issued, where the domain ID offset (0x60) was not applied correctly. The workaround for this wastoset switch portsofﬂine and then back online, and then devices would log back into the switch.

Fixes in switch ﬁrmware 6.4.0.10.00

Nameser

Port fc address error when upgrading ﬁrmware from 5.2.x.xx.xx to 6.4.x.xx.xx

ver error when booting from SAN to storage attached to switch with Domain ID = 1

An error another switch with Domain ID = 1. This was only encountered when processing Fibre Channel comma nd GetAll Next (GA_NXT),which allows adevicetoretrieve informationfor thenextportinthe nameserver database that is included in the zoning for the requesting device. The GA_NXT command is only used by a The er the attached to the switch with Domain ID = 1 would not be discovered.

When upgrading ﬁrmware from 5.2.x.xx.xx to 6.4.x.xx.xx on a switch conﬁgured with Domain ID =

was introduced in 6.4.0.xx.xx ﬁrmware that caused a nameserver error for devices attached to

Host Bus Adaptor (HBA) in a server which boots from a SAN, to discover the path to its bo ot LUN.

ror caused only the ﬁrst accessible port of the switch with Domain ID = 1 to be reported, where

subsequent GA_NXT command would skip to the next available Domain ID. Other paths to storage

1, the port fc addresses for some ports were changed, and could result in loss of connectivity to

BladeServers. The issue was due to an error where the port offset (4) and domain ID offset (0x60) were not applied correctly to the port address when ﬁrmware was upgraded.

Fixes in switch

ﬁrmware 6.4.0.08.00

Nameserver became unresponsive when speciﬁc Fibre Channel commands were issued

The nameserver became unresponsive while processing a GIP_NN (Get IP address by Nodename) of a device on a remote switch if the device had not registered an IP address in the nameserver, and thus had a

null value. The switch was unable to properly process the null IP value. This issue could also occur

in response to GIPP_ID and GIPP_PN Fibre Channel commands.

Implicit hard zoning issue caused CPU overload

With implicit hard zoning, h igh trafﬁc of bad frames (such as attempting to violate zoning) sent to a switch caused the switch CPU to overload, and fail to process ISL trafﬁc in a timely fashion, resulting in failed ISL connections.

Changes to port properties were slow to take effect

The port properties dialog appeared unresponsive (could take many minutes) when "select to propagate changes to entire column" was checked.

Port went ofﬂineafter acabledisconnect

There were port required a port reset to enable the connection to be reestablished.

occasionswhere aswitchportwentofﬂine after a cable to a device was disconnected. The

Changes to port status and port type conﬁguration were not always activated

Port status and p ort type selections in the Port Property display in Element Manager, which appears in the selection window, were not always activated unless ﬁrst selected from the drop-down menu.

Multiple changes of large zone sets caused th e nameserver to become unreponsive

Due to a memory leak, the nameserver became unresponsive when a large zone set was activated and

ivated multiple times.

deact

Fixes in switch ﬁrmware 6.4.0.06.00

Switch hang on boot due to internal timing issue with BladeSystem Management Module

If an HP p-Class BladeSystem is running BladeSystem Management Module ﬁrmware version 2.30 or later, the switch would frequently hang during the boot process. This was caused by an interrupt loo p that occurred du ring switch initialization.

Switch reboot due to very high Ethernet port activity

Due to an Ethernet driver issue, if a switch was subjected to very high Ethernet activity, the switch would occasionally reboot.

Internal ﬁrmware error due to uncommon SW_RSCN issue

The swi SW_RSC do

tch n am eser ver would crash (reported as an Internal Firmware Error) when receiving multiple

Ns for the same device within 200 msec for which the port states speciﬁed in the SW_RSCNs

not match. This only occurred when the switch was in McDATA Fabric interoperability mode.

McDATA 4Gb SAN Switch for HP p-Class BladeSystem release notes

+ 9 hidden pages