UNSOLVED
Power Path cant see al paths to come LUNs
I hope some one can help.
We recently had an issue with a Linux cluster it seems that some paths were disable on the clariion side.
once that was corrected, most paths to LUNs were restore except for three LUNs they only show two path to SPB but no paths to SPA:
Pseudo name=emcpoweraj
CLARiiON ID=APM00061101309 [GHPMD1_2]
Logical device ID=60060160DA7118001A4DEC683EE7DA11 [LUN 28]
state=alive; policy=CLAROpt; priority=0; queued-IOs=0
Owner: default=SP A, current=SP B
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
0 QLogic Fibre Channel 2300 sdai SP B2 active alive 0 1
1 QLogic Fibre Channel 2300 sdbr SP B1 active alive 0 1
We ran powermt config - restore - check, all with no luck. the last option here is to reboot the servers but we want to avoid that downtime, does any body have an idea?
Thanks.
We recently had an issue with a Linux cluster it seems that some paths were disable on the clariion side.
once that was corrected, most paths to LUNs were restore except for three LUNs they only show two path to SPB but no paths to SPA:
Pseudo name=emcpoweraj
CLARiiON ID=APM00061101309 [GHPMD1_2]
Logical device ID=60060160DA7118001A4DEC683EE7DA11 [LUN 28]
state=alive; policy=CLAROpt; priority=0; queued-IOs=0
Owner: default=SP A, current=SP B
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
0 QLogic Fibre Channel 2300 sdai SP B2 active alive 0 1
1 QLogic Fibre Channel 2300 sdbr SP B1 active alive 0 1
We ran powermt config - restore - check, all with no luck. the last option here is to reboot the servers but we want to avoid that downtime, does any body have an idea?
Thanks.
1302
0
Posted June 24th, 2008 10:00
I'll have to stand up for PowerPath here... can the OS see the newly presented paths to these LUNs? what can you see in /proc/scsi/scsi?
I'm guessing that the OS cannot 'see' these new paths, therefore PP cannot know about them... What we need to focus on it to force the OS to rescan for these paths, I'm no Linux guru, but the older kernels we could unload and reload the HBA module. In the newer kernels the HBA driver is built into the kernel so that isn't an option. I suggest digging out the Emulex/QLogic manual looking for a utility/script that will perform a SCSI rescan down that HBA.
Conor1302
0
Posted June 24th, 2008 10:00
Wouldn't that be an offline activity as well though? You might not have to reboot for it, but if you "unload the HBA module" you are effectively cutting off access to the storage. If you do this without unmounting the file systems I can't see the OS being happy about it.1302
0
Posted June 24th, 2008 11:00
thanks guys,
yeah I know Linux is buggy from time to time...
loading and unloading the module will result in temporary access loss to storage, but yet that is not the problem since there are several LUNs from the same array and the problem persist on two of them, so for some reason PP doesnt add the missing path for those LUNs.
I think they are going to have to reboot the server.1302
0
Posted June 24th, 2008 13:00
well those are existing LUNs, so there was no need to rescan.1302
0
Posted June 24th, 2008 13:00
if it is just effecting only these two LUNs then i would suspect if the devices are presented properly.
Which utility did u use to scan the new LUNs?1302
0
Posted June 25th, 2008 02:00
It would be worth trying a rescan before rebooting. If the Clariion problem had caused loss of access to a couple of paths the OS may not know about them (especially if its been rebooted) - so they effectively may be "new" devices.
Get a copy of "inq" for Linux, or "cat /proc/scsi/scsi" to determine if the OS can even see the missing devices. If not, PowerPath can't be expected to see them.1302
0
Posted June 25th, 2008 03:00
if the box came up and those paths were missing during the reboot, in my experience the only thing that will bring them back is reboot. I tried all kinds of scans (qlogic/emulex rescan scripts, echoing proc ..etc), no dice. I even had a ticket open with EMC (fruitless affair). I would bounce the box and it should be fine.1302
0
Posted June 25th, 2008 08:00
Thanks guys,
The sysadmin decided to reboot the box on the next maintenance window, in the mean time I'll tell him to do a rescan just in case it works.221
0
Posted October 7th, 2011 02:00
it really wroks guy...
Tks
Allen Ward
6 Operator
•
2124 Posts
1302
0
Posted June 24th, 2008 09:00
Seriously though, I think it looks like you are going to have to reboot. Occasionally PowerPath needs that kick in the pants to get everything back to normal.
...unless someone else has an alternative?