lspath output interpretation

On my VIo I see the following for my disks:

$ lspath | grep hdisk6
Enabled hdisk6 fscsi0 200600a0b82193f7,4000000000000
Enabled hdisk6 fscsi0 200700a0b82193f7,4000000000000
Enabled hdisk6 fscsi2 200600a0b82193f8,4000000000000
Failed hdisk6 fscsi2 200700a0b82193f8,4000000000000
 
$ lspath | grep hdisk7
Enabled hdisk7  fscsi0 200600a0b82193f7,5000000000000
Enabled hdisk7  fscsi0 200700a0b82193f7,5000000000000
Failed  hdisk7  fscsi2 200600a0b82193f8,5000000000000
Failed  hdisk7  fscsi2 200700a0b82193f8,5000000000000
 
$ lspath | grep hdisk23
Enabled hdisk23 fscsi0 200600a0b82193f7,14000000000000
Enabled hdisk23 fscsi0 200700a0b82193f7,14000000000000
Enabled hdisk23 fscsi2 200600a0b82193f8,14000000000000
Enabled hdisk23 fscsi2 200700a0b82193f8,14000000000000

Is see that some disks have a failed path on fscsi2. Other disks are just running fine with a path over fscsi2.

May I conclude that the hardware is working and that ther is a problem with the zoning of the SAN? Or could HW still be a problemen? All LUNS are provided from the same SAN storage box meaning that if hdisk23 is available on all paths, the cables between the SAN box and the server are correct.

--Peter

you get each disk over 2 fc adapters to the vio server, and each fc adapter is zoned to two storage controllers

depending on the san design, this may be due to a temporary problem with the fc adapter, or a san switch

first I would try to get the paths online manually

smitty mpio -> MPIO Path Management -> Enable Paths -> Enable Selected or All Paths 

and choose all paths, then run the lspath command again
depending on the storage subsystem you may have to work with the device drivers (sddpcm) for example

if this happens more often, you may want to open an ibm call, but first talk to your storage/san admins

1 Like

I tried smitty mpio but I still see a number of failed paths.

Have you always had each of your fibercards zoned to two different storage controllers? When we had dual fibercards and we had a one-to-one mapping between our fibercards and the storage controllers. Then we used MPIO to put the 2 paths (instead of 4) together. I didn't even know doing it the way you are was even possible.

which storage subsystem and drivers do you use?

$ ioslevel
2.1.1.10-FP21
$ oem_setup_env
# oslevel -s
6100-03-01-0921

Storage : DS4300

---------- Post updated at 04:05 PM ---------- Previous update was at 11:07 AM ----------

I got it fixed by removing the failed paths first. Then recreate them.

rmpath -dev hdisk.. -pdev fscsi..
mkpath -dev hdisk.. -pdev fscsi..

No reason found why the smit entry for enabling all paths failed.

are you using rdac drivers?

run

fget_config -Av

in oem_setup_env

it's been a while when I was working with rdac, but afaik each lun is presented by one storage controller, and in case the adapter zoned to this controller fails, the storage moves the lun to the other controller, which leads to a few seconds of time where the lun is not accessible

but our zoning was different, each adapter just saw one controller, maybe this has changed in the meantime

fget_config -Av does not return any output. I'm running AIX6 and my understanding is that the MPIO drivers are part of the base OS in AIX 6.

fget_config -Av does not show any output.

Isn't lspath the successor of fget_config in AIX 6?