HP StorageWorks Multi-protocol Router XPath OS 7.4.1e Release Notes (AV-RW5LN-TE, August 2007)
Table 6 Fibre Channel routing services
Issue
Description
Unexpected LSAN Zone Update
messages or unexpected physical
device offline messages
These messages can occur during normal operation. However, if the MP Router
reports numerous occurrences that cannot be correlated to actual initiated events,
such as planned device reconfigurations or reboots, examine low-level port error
counts using the portstatsshow command.
Faulty cables and SFPs that cause interruptions in EX_Port connectivit y might also
cause unexpected fabric reconfigurations. In the event of unexpected fabric
reconfigurations, ensure th at there are no low-level errors occurring on ports and
focus diagnostic efforts on cables and SFPs.
Zoning restrictions for LSAN zones
and local fabric zones
Remote WWNs cannot be in both a Logical Storage Area Network (LSAN) zone
and a local fabric zone of fabrics that they do not physically reside in. Putting them
in both locations, may result in excessive boot times for host or servers in fabrics
connected to the MP Router.
IfaremotenodeWWNisinanLSANzone,thenitcannotalsobeinalocal
non-LSAN zone in a fabric that it does not physically reside in. For example, if
youhaveahostorstoragedevicedefined in a remote fabric zone and you wish
to add either of these devices to an LSAN zone, you need to ensure that these
devices are not listed in any local zones other t han the physical fabric they reside
in. LocalWWNsareallowedinbothanLSANzoneandthelocalzonesofthe
fabric where they are physically located.
Fabric ID oversubsc ribed message
The port
error Fabric ID oversubscribed displays if you configure multiple
EX-port connections to the same fabric using different Fabric IDs. While the Fabric
ID used
for each fabric connected to the MP Router is unique, all EX-ports connected
to the s
ame fabric from the MP Router must use the same Fabric ID.
HBA configuration In certain large routed fabrics with large numbers of EX_Ports (for example, ≥ 8)
connected to each fabric (usually done for bandwidth for large numbers of shared
initiators and targets across fabrics) and a large number of shared devices (for
example, ≥ 800), set the PLOGI timeout value for 1 Gbps JN I FC HBAs to 3000
ms. This will allow time for the routed fabric to answer PLOGI requests.
Meta-SAN traffic In a non-routed SAN, if there are multiple E_Port links between two switches and
fabric reconfigurations occur due to the loss of a single E_Port link, then traffic
around the fabric is generally not interrupted (or is interrupted for a very short time:
for example, less than one second) .
Due to a design decision in this release of the Router, trafficdisruptionforrouted
traffic always occurs for a short period (less than one second), even if EX_Port
links remain available. The manager of a non-routed SAN will not expect this
performance characteristic.
During testing, applications have been shown to tolerate the trafficdisruptioninthe
routed fabric and to recover based on an ap plication specific recovery timeline,
which is typically longer than when the fabric again permits traffictoflow . However,
thecommonlyusedI/Otesttool,Iometer,isknowntopanicorcrashundercertain
circumstances of traffic disruption; this has been witnessed to occur in large routed
fabric configurations.
E
dge fabric switch booting and
EX_Ports
C
ertain configuration values and versions of Fabric OS switches in edge fabrics
might cause EX_Ports on the Router to be disabled when the switches are rebooted.
EvidenceoftheunexpecteddisabledEX_PortcanbeseenintheRouter’sport
display:
# switchshow
:
7 id AN No_Light disabled EX_PORT
(Last error: Failed to init in time)
Workaround for switches running FOS 3.x /4.x:
• Issue the diagdisablepost command on edge switches
connected to Router EX_Ports.
• Upgrade edge switches to Fabric OS 3.x/4.x.
Multi-protocol Router XPath OS 7.4.1e
7