Is the volume a Clariion snap of a existing volume seen by the same server?

If it is, I'm not sure how the LVM will react because the snap will have the same UUID as the already imported source volume. That may be why its saying it can't find a valid physical volume.

I don't know how you handle that situation in Linux. On other platforms you have to somehow change the UUID, vgid, or equivalent. It can be quite tricky to import array snaps back onto the same host.

If that's not the case here, forgive my intrusion.

1)

[root@lnxjyXXX root]# pvdisplay /dev/emcpowerq
pvdisplay -- ERROR "pv_read(): pv_create_name_from_kdev_t" no VALID physical volume "/dev/emcpowerq"

2) yes i'm pressenting snap devices on the same server than the source, but its working for 90% of the devices. And even if i have problems with pseudo name, as you can see, the deferent san path devices for this pseudo are working well. So i think it's only a problem related to powerpath itself and the way it's declaring pseudo devices on the system.

what is if you pvdisplay on /dev/emcpowerq

Hmmm... I'm not so sure.

I did a bit of research and you do need to change the UUID to get an array snapshot to import reliably into the Linux LVM. See below:


http://sources.redhat.com/lvm2/wiki/FrequentlyAskedQuestions

(Item 5 - Is there a way to activate/mount a copy of a volume group taken with a disk-array based point-in-time copy on the same host? The problem I have occurs because the disk-array copy contains all information on the disk, including PVID/VGID's, which is confusing for Linux's LVM.)

Which points to:
https://www.redhat.com/archives/linux-lvm/2006-September/msg00026.html
https://www.redhat.com/archives/linux-lvm/2006-October/msg00045.html

See also the following EMC knowledgebase article:
emc120281 - How to set up a Linux host to use emcpower devices in LVM

Hope that's of some help!

Best Regards,
Marc

how can i access "emc120281 - How to set up a Linux host to use emcpower devices in LVM" ?

Visit

http://powerlink.emc.com

Support > Knowledgebase Search > Support Solutions Search

Change the "Search For" dropdown from "Content" to "ID", and paste emc120281 into the box. Click on the "Search" button.

found it. I'm aware of the problem about the unique vgid but i'm not sure that it's my problem because i can pvdisplay one of the path to pv under the pseudo device name that is not responding to pvdisplay... the id is supposed to be the same between the pseudo and its different paths.

for example, here is another snap device from the same session that is fully working well:

[root@lnxjyXXX ># powermt display dev=emcpowerm
Pseudo name=emcpowerm
CLARiiON ID=CK200053500xxx [Backup Refermag]
Logical device ID=6006016041A316001C31EE978584DA11 [snap406]
state=alive; policy=CLAROpt; priority=0; queued-IOs=0
Owner: default=SP B, current=SP A
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
1 lpfc sdaw SP B0 active alive 0 0
2 lpfc sdes SP A0 active alive 0 0
2 lpfc sdfy SP B1 active alive 0 0
1 lpfc sdq SP A1 active alive 0 0


[root@lnxjyXXX ># pvdisplay /dev/emcpowerm1
--- Physical volume ---
PV Name /dev/emcpowerm1
VG Name vgoracle
PV Size 50 GB [104856192 secs] / NOT usable 32.19 MB [LVM: 134 KB]
PV# 11
PV Status available
Allocatable yes (but full)
Cur LV 1
PE Size (KByte) 32768
Total PE 1598
Free PE 0
Allocated PE 1598
PV UUID ujhDeU-5COJ-hwUs-qOSd-61ZW-9eY3-mpdBiG


[root@lnxjyXXX ># pvdisplay /dev/sdaw1
--- Physical volume ---
PV Name /dev/sdaw1
VG Name vgoracle
PV Size 50 GB [104856192 secs] / NOT usable 32.19 MB [LVM: 134 KB]
PV# 11
PV Status available
Allocatable yes (but full)
Cur LV 1
PE Size (KByte) 32768
Total PE 1598
Free PE 0
Allocated PE 1598
PV UUID ujhDeU-5COJ-hwUs-qOSd-61ZW-9eY3-mpdBiG

You're right, the ID is supposed to be the same for the powerpath pseudo (parent) device & all the childen - but in the case of a snap presented back to the same host, you'll have two different devices with the same PVID because the snap is an exact copy including all the LVM metadata.

If you reboot a host with duplicate UUIDs there's a chance that the LVM will pick up the snap device before the source - and import the snap instead. This can lead to data loss in the future as new writes will go to the snap instead of the source.

Anyway, I take your point about pvdisplay working for some and not others. Try the following which might give some clues (maybe post the output if its not too long):

# lvmdiskscan -v
# powercf -q (not sure if this exists in Linux PowerPath)
# powermt config
# powermt display
# powermt check
# ls -l /dev/emcpowerq*
# pvdisplay -vv /dev/emcpowerq1 (that two v's)
# pvdisplay -vv | grep xKf1sO-33a7-SX0H-xS1p-7rNp-YpPL-ZJXgCA
This should find all the paths with the above ID, as taken from your earlier post.
# pvscan -u
# dmesg

Check the messages/syslog file for any LVM related errors.

M

Message was edited by:
MarcT - added "pvscan -u"

exact Mr. T,

the pvuuid are the same for the source and the snap. And when i check all the pseudo devices pairs (source/snap): when the pseudo work for the source, it doesn't work for the snap, and when it works for the snap it doesn't work for the source. Anyway for a snap it's strange that pvdisplay work for one the path and not for the pseudo (with the same pvuuid).

I've not not found a simple method to change pvuuid. Does someone ever do this on Linux ?