EMC Time Finder & Cross-device link

[Ivan_86]

Hello Gurus.!

I need your Help.!

We make copies with Time Finder from one partiotion of SD with 11.23 to another similar.

Firts we made the copy, generated a map of the VG and transfer to another machine

We made the vgimport, fsck an vgchange and the VG startup without problems.

Then, we deactivated the VG, we wanted to make another copy, but in the moment to activate the VG i have this errors:

hibrida:/ #vgchange -a y vgbck
vgchange: Warning: Couldn't attach to the volume group physical volume "/dev/dsk/c41t0d7":
Cross-device link
vgchange: Cross-device link
vgchange: Cross-device link
vgchange: Cross-device link
vgchange: Cross-device link
vgchange: Cross-device link
vgchange: Cross-device link
vgchange: Warning: couldn't query physical volume "/dev/dsk/c41t0d7":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query physical volume "/dev/dsk/c41t1d0":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query physical volume "/dev/dsk/c41t1d1":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query physical volume "/dev/dsk/c41t1d2":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query physical volume "/dev/dsk/c41t1d3":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query physical volume "/dev/dsk/c41t1d4":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query physical volume "/dev/dsk/c41t1d5":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query all of the physical volumes.
vgchange: Couldn't activate volume group "vgbck":
Quorum not present, or some physical volume(s) are missing.

The question:

We have to repeat the vgimport.?
It's necessary to repeat the steps like in the first time?


Thanks a lot.!

[Duncan Edmonstone · Accepted solution]

Ivan,

I think we just need to understand what you did a little clearer.

This is my understanding - is it correct?

- Server A has volume group, lets call it vgA which contains EMC disks, lets call them DiskA and DiskB

- Using Timefinder you make a copy of DiskA and DiskB by creating BCVs (lets say DiskA-1 and DiskB-1) - those are prsented to Server B and you vgimport them into a volume group called say vgB

- Then you deactivated the volume group vgB containing DiskA-1 and DiskB-1.

Then I'm not clear on what you did next:

- did you resync DiskA-1 and DiskB-1 with DiskA and DiskB and then split again?

-or-

- did you make another copy of DiskA and DiskB (lets call them DiskA-2 and DiskB-2) and present those to Server B?

ALso is there any Serviceguard cluster involved here?

HTH

Duncan

[Ivan_86]

Ok, thanks Duncan the history is :

I have Server A with a VG04 and server B with a new Vg called VGBCK.

The steps are :

# symmir -g group_name establish
# symmir -g group_name split

In node source

# vgexport -p -m /tmp/vg04.map vg04
# scp -p /tmp/vg04.map node_target:/tmp

In node target

# mkdir /dev/vgbck
# mknod /dev/vgbcv/group c 64 nnnnnn

For all 7 disks of vgbck:

# vgchgid /dev/rdsk/cxtydz /dev/rdsk/... ...

# vgimport -m /tmp/vg04.map /dev/vgbck /dev/dsk/cxtydz ...

# vgchange -a y vgbcv

For all files systems of vgbcv:

# fsck -y -o full /dev/vgbck/lvnnn

# mount /dev/vgbcv/lvnnn mount_point


All this steps works without problem.

For resync the Disc we make the next steps:

In the Target (SistemB)

Umount all FS

# umount /lvnnnnn

Deactivate VGBCK

# vgchange -a n VGBCK

Start and split resync

#symmir -g group_name establish
#symmir -g group_name split

Finally, in the target i try to activate the VG again, and in this point i have the problem

# vgchange -a y VGBCK


Is necessary firts export VGBCK in the target and make establish en then import again ?

[Tim Nelson]

Each time you sync, split and then import on B you need to completely remove the vgBCK form B. Once you sync the vgid is set back and the old vgBCK is confused.

1) symmir
2) split
3) vgexport to get the map file from A
4) import on B using map file

Reverse:
1) unmount, deactivate and vgexport on B
2) symmir establish on A

Start over from begining.

[Duncan Edmonstone]

Why are you running vgchgid?

The only two cases where you do this are:

a) You want to mount the split copy on the same node as the original copy.

-or-

b) Your VG is part of a cluster and you want to clear the flag in the VGRA that identifies the VG as a cluster VG .

If neither of those are true then just take out the vgchgid stage and you will only need to do the vgimport the once.

HTH

Duncan

[Kevin Wright]

vgexport the VG, vgimport the VG after each sync/split.