vgimport issue

[Fenny_1]

Hi All,
Am facing the below mentioned error while importing the VG on the failover node in MC/SG

tppora04:/#vgimport -v -m xpvgsiedata11.map /dev/xpvgsiedata11
Beginning the import process on Volume Group "/dev/xpvgsiedata11".
vgimport: Unable to read the physical volume.tppora04:/#
tppora04:/#
tppora04:/#
tppora04:/#vgimport -v -m xpvgsiedata11.map -f xpvgsiedata11.df /dev/xpvgsiedata11
Beginning the import process on Volume Group "/dev/xpvgsiedata11".
Physical Volume "/dev/dsk/c26t3d2" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d2" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t3d3" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d3" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t3d4" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d4" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t3d5" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d5" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t3d6" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d6" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t3d7" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d7" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t4d0" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t4d0" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t4d1" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t4d1" is not part of a Volume Group

Please help me in this regard.

[A. Clay Stephenson]

It seems to be telling you exactly what is wrong. None of the PV's in the infile can be found. I suspect that these pathnames correspond to the other node and do not match the pathnames on this nodes. Rather than using this form of vgexport/vgiport, I suggest that you use the "-s" version.

On the other node:
vgexport -m /tmp/xpvgsiedata11.map -v -s /dev/xpvgsiedata11

This will write the Volume Group ID to the mapfile.

You then copy the mapfile to the other node and do a vgimport.

vgimport -m /tmp/xpvgsiedata11.map -v -s /dev/xpvgsiedata11

The -s vgimport then searches all the disks for a matching VGID.


If this is a SAN, I assume that you have made all these LUN's accessible to both nodes; if you have created these LUN's since after the last boot then I assume that you have done an ioscan -fn and an insf so that the device nodes have been created on both nodes.

[Steven E. Protter]

Shalom,

insf -C disk on the secondary node?

Did you do a vgexport -p (preview) to get a map file on the first node?

Did you vgchange -a n the volume group on the first node?

SEP

[Shardha]

Dear

Find a complete procedure to create, export and finaly import the File system. You are missing -s in the export and import command. Make sure before doing this procedure you can see the physical disk on both nodes.

Shardha


SET UP VOLUME GROUP vg_billing
# ioscan
# insf â e (Note down the new pv names)
# pvcreate /dev/rdsk/c4t0d1 /dev/rdsk/c4t0d2 /dev/rdsk/c4t0d3
/dev/rdsk/c4t0d4

# mkdir /dev/vg_billing
# mknod /dev/vg_billing/group c 64 0x010000
# chmod 755 /dev/vg_billing
# chmod 640 /dev/vg_billing/group

# vgcreate â s 16 vg_billing /dev/rdsk/c4t0d1 /dev/rdsk/c4t0d2 /dev/rdsk/c4t0d3 /dev/rdsk/c4t0d4 /dev/rdsk/c6t0d1 /dev/rdsk/c6t0d2 /dev/rdsk/c6t0d3 /dev/rdsk/c6t0d4 /dev/rdsk/c8t0d1 /dev/rdsk/c8t0d2 /dev/rdsk/c8t0d3 /dev/rdsk/c8t0d4 /dev/rdsk/c10t0d1 /dev/rdsk/c10t0d2 /dev/rdsk/c10t0d3 /dev/rdsk/c10t0d4

create logical volumes on vg_billing

# lvcreate vg_billing
# lvcreate vg_billing
# lvcreate vg_billing
# lvcreate vg_billing

# lvextend â l 3199 /dev/vg_billing/lvol1 /dev/dsk/c4t0d1
# lvextend â l 3199 /dev/vg_billing/lvol2 /dev/dsk/c4t0d2
# lvextend â l 3199 /dev/vg_billing/lvol3 /dev/dsk/c4t0d3
# lvextend â l 3199 /dev/vg_billing/lvol4 /dev/dsk/c4t0d4

# newfs â F vxfs â o largefiles /dev/vg_billing/rlvol1
# newfs â F vxfs â o largefiles /dev/vg_billing/rlvol2
# newfs â F vxfs â o largefiles /dev/vg_billing/rlvol3
# newfs â F vxfs â o largefiles /dev/vg_billing/rlvol4

# mkdir /billing1
# mkdir /billing2
# mkdir /billing3
# mkdir /billing4

# mount /dev/vg_billing/lvol1 /billing1
# mount /dev/vg_billing/lvol1 /billing2
# mount /dev/vg_billing/lvol1 /billing3
# mount /dev/vg_billing/lvol1 /billing4

# umount /billing1
# umount /billing2
# umount /billing3
# umount /billing4


Export LVM configuration on 1st node.

#vgexport -v -s -p -m /etc/lvmconf/vg_billing.map vg_billing
#vgchange â a n vg_billing

On Node2 :-

# cd /etc/lvmconf
# rcp node1:/etc/lvmconf/vg_billing.map .
# mkdir /dev/vg_billing
# mknod /dev/vg_billing/group c 64 0x010000
# vgimport â v â s â m /etc/lvmconf/vg_billing.map vg_billing
# mkdir /billing1
# mkdir /billing2
# mkdir /billing3
# mkdir /billing4
# vgchange -a r vg_billing
# vgcfgbackup /dev/vg_billing
# vgchange -a n vg_billing

[Fenny_1]

Hi All,
I used these commands for VG export/import.

vgexport -p -v -s -m xpvgsiedata11.map /dev/xpvgsiedata11
vgchange -a n /dev/xpvgsiedata11

on the other node when i did vgimport now am facing this error.
vgimport -v -s -m xpvgsiedata11.map /dev/xpvgsiedata11
Beginning the import process on Volume Group "/dev/xpvgsiedata11".
Couldn't access the list of physical volumes for volume group "/dev/xpvgsiedata11".

i checked with ioscan and i have found that disks h/w paths are different. so i tried again with vgimport using the -f option and specifying all the PV's in the file
this time it gives the same error.


tppora04:/#vgimport -v -m xpvgsiedata11.map -f xpvgsiedata11.df /dev/xpvgsiedata11
Beginning the import process on Volume Group "/dev/xpvgsiedata11".
Physical Volume "/dev/dsk/c26t3d2" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d2" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t3d3" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d3" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t3d4" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d4" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t3d5" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d5" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t3d6" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d6" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t3d7" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t3d7" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t4d0" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t4d0" is not part of a Volume Group
Physical Volume "/dev/dsk/c26t4d1" is not part of a Volume Group
Physical Volume "/dev/dsk/c27t4d1" is not part of a Volume Group

Now please help me to resolve this.

[A. Clay Stephenson]

When the "-s" version of vgimport failed. It almost certainly means that the 2nd node cannot "see" the LUN's.

The VGID is written on every PV in the VG and that is what the "-s" version of vgimport is looking for.

Here's what I want you to do (and do this exactly):

On the 1st node:
vgimport -p -v -m /tmp/xpvgsiedata11.map /dev/xpvgsiedata11
cat /tmp/xpvgsiedata11.map # you should see the VGID; it is unique to a volume group

Next:
vgdisplay -v /dev/xpvgsiedata11 | grep 'PV Name' and note any PV listed. Now pick one of these PV's (it doesn't matter which one).
I'll pretend it is /dev/dsk/c27t4d1. That's the block device; we will use the character device, /dev/rdsk/c27t4d1 for our experiment to extract the VGID directly from the disk.

dd if=/dev/rdsk/c27t4d1 bs=512 count=1 skip=16 of=/tmp/f1

Now lets examine this file:
xd -Ad -tx /tmp/f1

Note the values displayed at offset 16, the next 8 bytes should match the VGID displayed in your map file above. If it does, you now know how to extract the VGID.

Repeat the dd and xd commands on the 2nd node using a disk that you THINK is part of the same VG. I will bet that you do not find a VGID.

[Sudalaimani]

Hi,

Some times, we may miss to check the little things, you verify the following before importing your VG on the other node.

On the other node (Fail over node), aslo due to the controller id your cxtxdx no. will be different on the fail over node., so its not an issue

1. Aftrer running ioscan -fnC disk- run #insf -e ( this will re attach any new disk or newly added luns to the sytem)

2. Make sure you are exporting the VG first on the Fail over node before importing

3. Make sure you have the same minor number for the VG (similar to the primary node)

4. If you have powerpath, verify the LUNS is visible here by checking the LUN no. with syminq output and related cxtxdx in #powermt display dev=all |grep cxtxdx

Hope this will give you some idea

regard

Mani

[A. Clay Stephenson]

Ooops, I'm stupid:

On the 1st node:
vgimport -p -v -m /tmp/xpvgsiedata11.map /dev/xpvgsiedata11
cat /tmp/xpvgsiedata11.map # you should see the VGID; it is unique to a volume group

should be:
On the 1st node:
vgexport -p -v -m /tmp/xpvgsiedata11.map /dev/xpvgsiedata11
cat /tmp/xpvgsiedata11.map # you should see the VGID; it is unique to a volume group

[Sudalaimani]

Hi

One more thing,

1. Make the VG cluster aware by the below command
#vgchange -a n /dev/xpvgsiedata11
#vgchange -c y /dev/xpvgsiedata11
#vgchange -a 2 /dev/xpvgsiedata11

2. Also use -s while generating the map file

#vgexport -p -v -s -m xpvgsiedata11.map /dev/xpvgsiedata11

Hope this will be bit help

regards

Mani

[Fenny_1]

Hi A Clay and All,
Thanks for everyone's detailed response. But let me tell you one thing.
This is a newly created VG and am trying to import it first time on the failover node.

A Clay,
I'll try all your recommended commands and let you know the result tomorrow.

Thanks alot everyone , i still need a lot of your help to resolve this issue asap.

[Fenny_1]

Hi Guys,
Am waiting for your feedback.