Unsolved
This post is more than 5 years old
2 Posts
0
22180
July 15th, 2012 21:00
Solaris - Powerpath new pseudo names disappear upon reboot
Hi All,
I've got a problem, I've got a solaris 10 box that's causing me some grief. Historically the box had 20 LUNs which powerpath was quite happy to manage. Recently we added another 20-25 LUNS and powerpath was quite happy to manage these new LUNS.. However the server hadn't been restarted until now, upon restart the server still see's the original 20 luns however the new 20-25 LUNS would not mount.
Upon investigation we found that the new 20-25 LUNS were still visable by the host ie power display dev=all could see all of the LUNS but the pseudoname had vanished off the 20-25 recently added LUNS.
We can get the pseudoname by clearing the emcp.conf and powermt.custom and running the powermt config, powermt save commands but upon restart I am back in the same problem where my pseudonames for my new LUNS have vanished.
Can anyone provide any advice as to what I am missing
Cheers
Paul
in the below output LUN 78 (emcpower14a) was still visible/accessible/could be mounted by the host however LUN 101 (pseudoname disappeared after restart) was not mountable.
name=emcpower14a CLARiiON ID=CKM00100900897 [TTH-WFM1]
Logical device ID=6006016039302900E4A270901F1BE111 [LUN 78 - WFM old lun 61]
state=alive; policy=CLAROpt; priority=0; queued-IOs=0;
Owner: default=SP A, current=SP A Array failover mode: 1
==============================================================================
--------------- Host --------------- - Stor - -- I/O Path -- -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
3074 pci@1f,700000/pci@0/SUNW,qlc@2/fp@0,0 c2t5006016044604C67d1s0 SP A0 active alive 0 0
3074 pci@1f,700000/pci@0/SUNW,qlc@2/fp@0,0 c2t5006016844604C67d1s0 SP B0 active alive 0 0
3072 pci@1f,700000/pci@0,2/SUNW,qlc@1/fp@0,0 c3t5006016144604C67d1s0 SP A1 active alive 0 0
3072 pci@1f,700000/pci@0,2/SUNW,qlc@1/fp@0,0 c3t5006016944604C67d1s0 SP B1 active alive 0 0
CLARiiON ID=CKM00100900897 [TTH-WFM1]
Logical device ID=6006016039302900FAFF148DBA1BE111 [LUN 101 - WFM new]
state=alive; policy=CLAROpt; priority=0; queued-IOs=0;
Owner: default=SP B, current=SP A Array failover mode: 1
==============================================================================
--------------- Host --------------- - Stor - -- I/O Path -- -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
3074 pci@1f,700000/pci@0/SUNW,qlc@2/fp@0,0 c2t5006016044604C67d41s0 SP A0 active alive 0 0
3074 pci@1f,700000/pci@0/SUNW,qlc@2/fp@0,0 c2t5006016844604C67d41s0 SP B0 active alive 0 0
3072 pci@1f,700000/pci@0,2/SUNW,qlc@1/fp@0,0 c3t5006016144604C67d41s0 SP A1 active alive 0 0
3072 pci@1f,700000/pci@0,2/SUNW,qlc@1/fp@0,0 c3t5006016944604C67d41s0 SP B1 active alive 0 0
jespinoza1
52 Posts
0
July 16th, 2012 06:00
Hi! I found this Support Case that might help:
On the Solaris 10 x86 host, after upgrading PowerPath from version 5.1.1 to 5.3.0, emcpower pseudo name is getting changed on each reboot.
Workaround: To resolve this issue, perform the following steps:
Reboot the host again, and check if the pseudo naming changes.
Although emcp printed a message that it determined MPxIO was enabled, it was disabled.
The reason PowerPath thought host had MPxIO enabled was due to /kernel/drv/fp.conf file having duplicate entries of "mpxio-disable=yes". The duplicate entries caused PP install script installUtils.sh to incorrectly detect that mpxio was enabled for Solaris i386 platform.
If that doesn't work, I recommend you open a Support ticket with EMC.
paulgroth
2 Posts
0
July 16th, 2012 20:00
Hi There,
Thanks for the response. I had a look at that support case before my original post here but don't think it was valid in my situation. The pseudonames aren't actually changing, they are disappearing for half of my LUNS and I am on a SPARC box as opposed to x86 host. I've checked and am confident that MPXIO isn't enabled.. The only thing in common is that I am using powerpath 5.3.0.
I suspect you're right, the best option is going to be a support case with EMC but its complicated because of vendor engagement and escalation channel at the moment..
Alex_Ye
109 Posts
0
July 20th, 2012 06:00
It would be interesting to see how the boot messages look like, and whether the devices files under /dev and /devices do exist when you can't see them in "powermt display dev=all"?
Alex_Ye
109 Posts
0
August 14th, 2012 02:00
Hi Paul,
I'm not sure if you have fixed this issue. If not, it may be worth trying again to clean up the powerpath config files. In addition to /etc/powermt.custom, you can also try by renaming /etc/emcp_devicesDB.dat and /etc/emcp_devciesDB.idx, and then reboot the host.
I've never seen a solaris host showing this behavior repeatedly even after removing powermt.custom and reboot, but I guess you never know...
mon0201
2 Posts
0
September 11th, 2012 03:00
Hi, i have the same problem
installed powerpath 5.5 P 01 on Sparc 64 on 3 servers (2 M5000 and 1 T-3)
i am using solaris 11 latest SRU 10.5, even on 9.5 problem existed
after reboot the pseudo device names are reset to their original name
and emcpadm rename needs to be used to rename all pseudo devices again
powermt save does not seem to help
pseudoDevNames are correct in /etc/powermt_custom.xml after emcpadm
upon reboot, the pseudoDevNames are reset in /etc/powermt_custom.xml,
although both the original and correct names of the pseudo devices appear in /var/adm/messages
something is resetting the names of the pseudo devices dureing reboot
i opened a case at emc, but nothing so far
i will let you know whenever EMC finds something
chris van de vijver
mon0201
2 Posts
0
September 11th, 2012 06:00
It turns out that a emcpower service called powershift removes the pseudo devices at boot.
This is mentionned in the log file /var/svc/log/system-emcpower-powershift:default.log
the contents is:
....
Setting up paths
Could not validate the entry:
---------------------------------------
emcpower730: user ID = 15b000016d0
---------------------------------------
removing emcpower730
...
all renamed pseudo devices are removed by powershift and the original pseudo device names 0, 1, 2, 3, 4, ... are left over.
There is no problem to rename them again after boot this is is not a desirable situation
My temporary work around is to rename them again, make an export of the mappings and import the mappings automatically after each reboot...
SamiEMC
72 Posts
0
January 18th, 2013 10:00
facing same problem
below is the primus which should resolve this
The following is a Primus(R) eServer solution:
ID: emc298786
Domain: EMC1
Solution Class: 3.X Compatibility
Fact OS: Solaris SPARC
Fact EMC SW: PowerPath
Fact EMC SW: PowerPath for Solaris 5.5 (build 589)
Symptom Message during boot:
Jul 10 17:41:10 v490 PowerPath: [ID 702911 daemon.notice] MPxIO is enabled for CLARiiON storage arrays, switching off PowerPath path management for CLARiiON.
Jul 10 17:41:14 v490 PowerPath: [ID 702911 daemon.notice] MPxIO is enabled for VNX storage arrays, switching off PowerPath path management for VNX.
Symptom Solaris 10 SPARC U9 - EMC PowerPath 5.5, emcpower pseudo name getting changed on reboot.
Change On the Solaris 10 SPARC host, after installing PowerPath 5.5, emcpower pseudo name is getting changed on each reboot.
Fix Workaround:
To resolve this issue, perform the following steps:
Note: If the "powerstartup" service enabled then the pseudo name will be changed upon host reboot.
Fix The issue will be fixed on Powerpath 5.5 P02.