1834814 Members
2589 Online
110070 Solutions
New Discussion

vgexport /import problem

 
M. Tariq Ayub
Regular Advisor

vgexport /import problem

Hi i am having problem in vg export / import. I have exported a file and rcp it to another node of cluster and import there. but when i try to activate it it says filesystem corrupted. run fsck. what should i do.
3 REPLIES 3
A. Clay Stephenson
Acclaimed Contributor

Re: vgexport /import problem

We seem to be confused on terminology. Activate refers to a volume group (e.g. vgchange -a y /dev/vg01) while filesystem refers to a mount operation. Probably what is happening is that you have correctly vgimport'ed and are now trying to mount. I suspect that what is happening is the mount command is failing because 1) the filesystem was not cleanly unmounted on the old system 2) you did not specify a filesystem type in your mount and it defaulted to hfs when the filesystem is actually vxfs.

You should copy the /etc/fstab entry for this filesystem from the old box to the new -- posibly changing the volumme group -- if you changed vg01 on the old, fog example, to vg02 on the new. In the worst case, you can always run fsck on the filesystem but be certain to specify the correct filesystem type.
If it ain't broke, I can fix that.
M. Tariq Ayub
Regular Advisor

Re: vgexport /import problem

Hi,

We are using mc service guard.
A. Clay Stephenson
Acclaimed Contributor

Re: vgexport /import problem

Well that changes a few things but not much. You will not have an entry in /etc/fstab but rather in your package scripts for the VG and the filesystems. If the package/node crashed then the filesystem was not cleanly unmounted so the same applies as above. The VG activation should almost certainly change from vgchange -a y to -a e (exclusive) assuming the the VG has been made cluster aware with a vgchange -C y.

Normally before trying to activate a package, I do all the vgexport/vgimports on all the nodes and activate each one manually and mount all the filesystem manually before doing a vgchange -C y -- that only has to be done once on any node you choose because that data is carried on the disks themselves.

I assume you did a vgexport /dev/vgxx -m vgxx.map -p -s -v /dev/vgxx and copied the mapfile over to the new box. You should then vgchange -a n /dev/vgxx on the old box and try to vgimport on the new using the -s option.
Of course, you must create a /dev/vgxx directory and group file on the new box. Make certain that the group file's minor device number is unique on that system.

If it ain't broke, I can fix that.