Sun Traffic Manager not managing 2 QLC controllers
We have a V440 solaris 9, latest patch cluster and storedge traffic manager 4.4 with patches. 2 Sun QLC single channel controllers each see the same SAN disks via brocade switches.
Why wont Traffic Manager create pseudo controller details in /dev/rdsk & /dev/dsk? I have checked fp.conf, scsi_vhci.conf, qlc.conf for mpxio_disable="no" etc
reboot -- -r messages show multipathing enabled, auto failback enabled.
Help! Suggestions for further checks and gotchas
stmsboot -L shows the correct associations so somewhere STMS is not enabled but I cannot figure it out.
The local scsi boot disks are not dual pathed, they are mirrored. Running stmsboot -e errors immediately refering to the root mirror.
[760 byte] By [
ianjames] at [2007-11-25 23:00:29]
# 1
are these devices seen in format?
If not, I'd check if they are configured using cfgadm:
cfgadm -al -o show_FCP_dev
will tell you if the hba paths see your luns, or not. If they happen to show up as "unconfigured", configure them using:
cfgadm -c configure c#
THEN check format.
If you see them in format, unconfigure them:
cfgadm -c unconfigure c#::WWPN
clean up the device paths:
devfsadm -Cv
then configure them. format should now see the scsi_vhci paths.
If not, I'd be interested to see the scsi_vhci.conf, fp.conf, and qlc.conf files, as well as the format and cfgadm -al -o show_FCP_dev outputs.
# 2
Everything looks ok in that respect -
cfgadm -al -o show_FCP_dev
Ap_Id Type ReceptacleOccupantCondition
c3fc-fabricconnectedconfiguredunknown
c3::50060e8003286f10,0 disk connectedconfiguredunknown
c3::50060e8003286f10,1 disk connectedconfiguredunknown
c3::50060e8003286f10,2 disk connectedconfiguredunknown
c3::50060e8003286f10,3 disk connectedconfiguredunknown
c3::50060e8003286f10,96disk connectedconfiguredunknown
c3::50060e8003286f10,97disk connectedconfiguredunknown
c3::50060e8003286f10,98disk connectedconfiguredunknown
c3::50060e8003286f10,240disk connectedconfiguredunknown
c3::50060e8003286f10,241disk connectedconfiguredunknown
c4fc-fabricconnectedconfiguredunknown
c4::50060e8003286f00,0 disk connectedconfiguredunknown
c4::50060e8003286f00,1 disk connectedconfiguredunknown
c4::50060e8003286f00,2 disk connectedconfiguredunknown
c4::50060e8003286f00,3 disk connectedconfiguredunknown
c4::50060e8003286f00,96disk connectedconfiguredunknown
c4::50060e8003286f00,97disk connectedconfiguredunknown
c4::50060e8003286f00,98disk connectedconfiguredunknown
c4::50060e8003286f00,240disk connectedconfiguredunknown
c4::50060e8003286f00,241disk connectedconfiguredunknown
# 3
I have unconfigured each of the SAN devices showing in format, removed the devices entries, /dev/dsk and /dev/rdsk. After a reboot, only the local disks show in format. Configuring the both SAN fcps, cfgadm shows the controllers as connected, configured "failed" and each disk as "failing", this persists across a reboot. The disks are accessible via format on c3 or c4, can be labelled and formated but still no traffic manager.
# 4
connected, configured, failed or failing, indicates that one ore more devices could not be accessed to complete the configure process
during boot.
Keep in mind, that if you turn on STMS, and you have only one path, it will still create the pseudo device path for fibre-channel devices.
So work with only the one-path model for now.
Try:
cfgadm -c unconfigure c3::50060e8003286f10
cfgadm -c unconfigure c4::50060e8003286f00
devfsadm -Cv
make sure that the the above WWPNs do NOT appear in the /etc/cfg/fp/fabric_WWN_map
init 6
Now there is one caviate that I forgot to ask you about earlier, there are certain array makes that require an entry in the scsi_vhci.conf. If your array is not a Sun Branded array, Hitachi,
or LSI-Logic array, it may require the "symmetric-option". Take a moment to evaluate this before performing the "init 6" above.
Once the system is init 6'd, take a look at
cfgadm -o show_FCP_dev -al
Do you see all the luns, and are they in the unconfigured, unknown state?
If so, configure only one of the paths. If possible have a "tail -f /var/adm/messages" running in another terminal window. This will show you the messages that occur during the configure. If you get some WARNINGs, I'd suggest taking a look a the lun masking/security on the array. If no warning, check format, you should see your paths!!
Another question is if you turn STMS off, are all the paths able to configure successfully, and can they be seen in format with:
cNtWWPNdN
