04-22-2011 10:16 PM
HW :
Two Solaris 10 Update 9 cross connecting EMC Storage AX4-5
Setup:
Two nodes cluster running Veritas Storage Foundation Cluster File System/HA 5.1 SP1.
2 virtual disks are configured on EMC storage.
Problem:
1. vxdisk shows not only two EMC virtual disks (emcpower0 and emcpower1), but also the whole EMC storage device emc_clariion0_0.
2. Number of paths for emcpower0 or emcpower1 is 1 instead of 2.
Please note that the same configurations work OK on Solaris 9 + VCS 5.0 MP3.
Best Regards,
Gao
Solved! Go to Solution.
04-23-2011 12:02 AM
Ok, well looking at above outputs, I can not clearly say that emc_clariion0_0 is an invalid/stale entry or no .. reason being:
in the vxdmpadm listctlr all output above I see that c2 is connected to emc_clariion1 .. point to note is, status is shown as "connected" ... now, if we see the devices in vxdisk list output, they also appearing from "c2" ..
Also, if we carefully observe the path for c2 devices, they are coming from "qlc@0,1/fp" however all c1 disks are at "qlc@0/fp" (notice the bold characters)
so here is what I would suggest:
-- Get the storage checked , Lun masking/settings done correctly, make sure zoning is done correctly at SAN switches & that too connections reported to c2 controller are fine from storage end. Also storage settings are as per symantec recommended (see HCL for 5.1)
http://www.symantec.com/docs/TECH74012
-- Make sure OS part is correct, like luxadm -e port is reporting FC connected, cfgadm reports the controller configured, if not try configuring/reconfiguring
-- Lastly, if all above appears in order, try following below technote to cleanup the device tree, there are two choices in technote, one is reconfigure reboot while another is online method, give a go on online method first...
http://www.symantec.com/docs/TECH78463
Hope this helps
Gaurav
04-22-2011 11:05 PM
Hello,
I am not really concerned about only path being shown for emcpower devices because they are managed by powerpath, so yes, you should only see one path beneath emcpower device. If you want to see the alternate paths for that device, you should use
# powermt display dev=all (or dev=emcpower0a)
Now regarding other disks which are appearing as emc_clariion0_0, was this disk added later ? anything specific was done for this device ? Also, in the format menu, the disks are being shown like internal disks with no FC name or wwn in it , do you know why it is like this ? Also, is this a new setup or was working before ?
Was powerdevices configured later to DMP ? can u paste below outputs:
# vxdmpadm listenclosure all
# vxdmpadm listctlr all
# vxddladm listexclude all
# vxdmpadm listexclude all
Gaurav
04-22-2011 11:41 PM
Thanks Gaurav. I was wrong on problem 1, and "# powermt display dev=all" shows that two paths are available for either emcpower0 or emcpower1.
The disk appearing as emc_clariion0_0 was not added later. Actually no specific was done for EMC storage. And you are right the disks are shown like internal disks yet I don't know the reasons.
This is not a new setup. Previously T5220 + Solaris 9 + VCS 5.0 MP3 works OK for years.
Now we've changed to T3 + Solaris 10 + VCS 5.1, and this extra disk emc_clariion0_0 appears in VXVM (causing disk group failure later on).
I'm OT at home, so I have only logs recorded earlier as below. I've also done Google searching for 3 days with no luck. Hope that you can see what's wrong from below info. Many thanks!
bash-3.00# vxdmpadm listctlr all
CTLR-NAME ENCLR-TYPE STATE ENCLR-NAME
BRs,
Gao
04-23-2011 12:02 AM
Ok, well looking at above outputs, I can not clearly say that emc_clariion0_0 is an invalid/stale entry or no .. reason being:
in the vxdmpadm listctlr all output above I see that c2 is connected to emc_clariion1 .. point to note is, status is shown as "connected" ... now, if we see the devices in vxdisk list output, they also appearing from "c2" ..
Also, if we carefully observe the path for c2 devices, they are coming from "qlc@0,1/fp" however all c1 disks are at "qlc@0/fp" (notice the bold characters)
so here is what I would suggest:
-- Get the storage checked , Lun masking/settings done correctly, make sure zoning is done correctly at SAN switches & that too connections reported to c2 controller are fine from storage end. Also storage settings are as per symantec recommended (see HCL for 5.1)
http://www.symantec.com/docs/TECH74012
-- Make sure OS part is correct, like luxadm -e port is reporting FC connected, cfgadm reports the controller configured, if not try configuring/reconfiguring
-- Lastly, if all above appears in order, try following below technote to cleanup the device tree, there are two choices in technote, one is reconfigure reboot while another is online method, give a go on online method first...
http://www.symantec.com/docs/TECH78463
Hope this helps
Gaurav
04-23-2011 12:08 AM
I will go to my employer's site and try tomorrow. Thanks anyway.
BRs,
Gao