Dell Emulex Family of Adapters - Manual (Page 286)

OneCommand™ Manager Application P010066-01A Rev. A
14. Troubleshooting
Emulex Driver for Linux and OneCommand Manager Application Situations
286
Emulex Driver for Linux and OneCommand Manager
Application Situations
Table 14-2 Emulex Driver for Linux and OneCommand Manager Application Situations
Situation Resolution
NIC Link fails to come up. For Emulex UCNAs and CFAs in NIC mode, you may need to
properly configure the network interface using system
administration utilities.
The OneCommand Manager application
software package does install. An error
message states that: “inserv Service Elxlpfc
has to be enabled for service ElxDiscSrvinserv:
exiting now/sbin/ inserv failed exit code 1.” Reinstall the driver with the lpfc-install script.
If a SAN configuration has 256 targets mapped
by the LPFC driver, any additional added
targets do not get a target ID mapping by the
driver and cause target discovery to fail.
Removing targets or reinitializing the link does
not solve the problem. Unload and reload the driver to reset available target IDs. Ensure
that the SAN configuration is correct prior to reloading the driver.
This clears the driver’s consistent binding table and free target
IDs for new target nodes.
In some cases, after loading an OEM supplied
combined firmware/OpenBoot image you are
not able to enable BootBIOS from the lputil
Boot BIOS Maintenance menu. 1) Download the current OpenBoot only image for your adapter
from the Emulex website.
2) Load the current OpenBoot only image following steps listed in
Updating BootBIOS section of this manual.
3) Run lputil, return to the Boot BIOS Maintenance menu.
4) Enable BootBIOS.
rmmod fails to unload LPFC driver module due
to ERROR: Module LPFC is in use.
This message can appear when you attempt to
remove the driver and there is a Logical
Volume Group dependent on the driver. Make the Logical Volume Group unavailable.
Type
lvchange -a n xxxxxxx
where xxxxxx is the Volume Group Name.
Slow targets or extended link faults on the
storage side may result in storage being
marked off-line by the mid-layer and
remaining off-line (not recovered) when the
link faults are corrected. The 8.2 version of the driver should eliminate this problem.
However, if you experience off-line device issues, increase the
SCSI command timeout to a value greater than or equal to sixty
seconds. Emulex also provides a script which addresses this issue
(for 2.6 kernels). To access the lun_change_state.sh script, go to
http://www.emulex.com/files/downloads/linux/tools.html, then
click the link to the appropriate driver, and click the Linux tools
link.
Under certain conditions of an I/O load, some
targets cannot retire an I/O issued by a Linux
initiator within the default timeout of 30
seconds given by the SCSI midlayer. If the
situation is not corrected, the
initiator-to-target condition deteriorates into
abort/recovery storms leading to I/O failures
in the block layer. These types of failures are
preceded by a SCSI IO error of hex 6000000. Emulex provides a script which addresses this issue. To access the
set_target_timeout.sh script, go to
http://www.emulex.com/files/downloads/linux/tools.html, then
click the link to the appropriate driver, and click the Linux tools
link.