DM Multipath

[John Guster]

#mutilpath -l shows
mpath4 (xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx)
[size=30 GB][features="0"][hwhandler="0"] \_ round-robin 0 [active]
\_ #:#:#:# - 4:12 [failed]
\_ 3:0:1:7 sde 4:40 [active]

# multipath -v2
remove: mpath4 (wwid changed)
mpath4: map in use
how to clear this failed path? Storage guy says a reboot is needed.

[Ivan Ferreira]

Please see:

http://forums11.itrc.hp.com/service/forums/questionanswer.do?threadId=1326622

¿How is that the wwid changed for that device?

[John Guster]

Storage guy removed that path but still allow this LUN being used by the system. is wwid supposed be changed when # of paths is reduced?

[Ivan Ferreira]

No, I'm not sure of the problem, but it may report wwid changed because the path is not there anymore, or another LUN was detected on that place.

Anyway, you shold keep at least two paths to the LUN.

[John Guster]

there is another path. looks like a reboot is needed. Do you think so?

[Steven McCoy]

Path issues typically clear up for me after a reboot. What does your /etc/multipath.conf file look like?

[Ivan Ferreira]

Maybe you need to rescan your disks. Have you rescanned your disks after the LUN maintenance?

[John Guster]

not sure what is rescan mean? can you provide more detail?

[Steven McCoy]

echo "- - -" > /sys/class/scsi_host/host0/scan
echo "- - -" > /sys/class/scsi_host/host1/scan

Where 'host0' is the first fiber card, 'host1' is the second, etc.

Make sure to have a 'tail -f' going on /var/log/messages when you run the command so you can spot possible issues.

If a redetect doesn't work, I'd suggest a reboot, unfortunately.

Hope this helps.

[John Guster]

it didn't do the trick.
does it need
echo "1">/sys/class/fc_host/host2/issue_lip
before echo "---"> /sys/class/scsi_host/host0/scan?