Hi All,
We have a number of RHEL5 hosts that we're wanting to connect to a
backend SAN (a Sun 6540). The hosts have two Emulex HBAs and they are
connected to the SAN via a pair of redundant switches.
This means there are *four* paths to each LUN - two active and two passive.
We are using the standard "lpfc" driver, configured with these options
in /etc/modprobe.conf:
options lpfc lpfc_nodev_tmo=28 lpfc_lun_queue_depth=16
lpfc_discovery_threads=32
The problem is that when the machines boot and the lightpulse driver
is started, large numbers of I/O errors are output on the console and
the bootup process subsequently takes a _long_ time.
Reading all physical volumes. This may take a while...
end_request: I/O error, dev sdr, sector 0
/dev/sdr: read failed after 0 of 4096 at 0: Input/output error
end_request: I/O error, dev sdr, sector 573312
/dev/sdr: read failed after 0 of 4096 at 293535744: Input/output error
end_request: I/O error, dev sdr, sector 573424
/dev/sdr: read failed after 0 of 4096 at 293593088: Input/output error
end_request: I/O error, dev sdr, sector 0
/dev/sdr: read failed after 0 of 4096 at 0: Input/output error
end_request: I/O error, dev sdr, sector 8
/dev/sdr: read failed after 0 of 4096 at 4096: Input/output error
end_request: I/O error, dev sdr, sector 0
/dev/sdr: read failed after 0 of 4096 at 0: Input/output error
end_request: I/O error, dev sdah, sector 0
/dev/sdah: read failed after 0 of 4096 at 0: Input/output error
end_request: I/O error, dev sdah, sector 573312
/dev/sdah: read failed after 0 of 4096 at 293535744: Input/output
error
end_request: I/O error, dev sdah, sector 573424
/dev/sdah: read failed after 0 of 4096 at 293593088: Input/output
error
end_request: I/O error, dev sdah, sector 0
/dev/sdah: read failed after 0 of 4096 at 0: Input/output error
end_request: I/O error, dev sdah, sector 8
/dev/sdah: read failed after 0 of 4096 at 4096: Input/output error
end_request: I/O error, dev sdah, sector 0
/dev/sdah: read failed after 0 of 4096 at 0: Input/output error
end_request: I/O error, dev sdc, sector 0
/dev/sdc: read failed after 0 of 4096 at 0: Input/output error
end_request: I/O error, dev sdc, sector 573312
/dev/sdc: read failed after 0 of 4096 at 293535744: Input/output error
end_request: I/O error, dev sdc, sector 573424
/dev/sdc: read failed after 0 of 4096 at 293593088: Input/output error
end_request: I/O error, dev sdc, sector 0
There are also some errors output to the console when the machine is shutdown:
Please stand by while rebooting the system...
md: stopping all md devices.
md: md0 switched to read-only mode.
md: md1 still in use.
Synchronizing SCSI cache for disk sdah:
Synchronizing SCSI cache for disk sdag:
FAILED
status = 1, message = 00, host = 0, driver = 08
<6>sd: Current: sense key: Illegal Request
<<vendor>> ASC=0x94 ASCQ=0x1ASC=0x94 ASCQ=0x1
Synchronizing SCSI cache for disk sdaf:
Synchronizing SCSI cache for disk sdae:
Synchronizing SCSI cache for disk sdad:
Synchronizing SCSI cache for disk sdac:
FAILED
status = 1, message = 00, host = 0, driver = 08
<6>sd: Current: sense key: Illegal Request
<<vendor>> ASC=0x94 ASCQ=0x1ASC=0x94 ASCQ=0x1
Synchronizing SCSI cache for disk sdab:
Synchronizing SCSI cache for disk sdaa:
FAILED
status = 1, message = 00, host = 0, driver = 08
<6>sd: Current: sense key: Illegal Request
<<vendor>> ASC=0x94 ASCQ=0x1ASC=0x94 ASCQ=0x1
[...]
We have been able to mitigate the long boot times using the modprobe.conf
parameters as outlined above, but the I/O errors are still coming up.
Do you have some recommendations as to how we can prevent these I/O errors
from happening? Is there some way to make the lpfc driver handle the passive
paths more elegantly?
(I should add that we have a perfectly functioning multipathd
configuration, so all's well once the system is booted. It's just the
slow boot times and excessive console errors I'm interested in finding
a solution for).
Regards,
Robert Sturrock
_______________________________________________
rhelv5-list mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/rhelv5-list
