1753726 Members
4370 Online
108799 Solutions
New Discussion юеВ

importing and exporting

 
gany59
Regular Advisor

importing and exporting

In which case, we need to export and import the file system on the cluster servers..
4 REPLIES 4
VVS
Regular Advisor

Re: importing and exporting

We import VGs in below cases,
1. Add a new volume group to a package.
2. Add logical volume to an existing volume group of a package.
3. Add disk to a volume group of a package.


Regards,
VVS
Work is life, you know, and without it, there's nothing but fear and insecurity.
Modris Bremze
Esteemed Contributor

Re: importing and exporting

The volume group (VG) must be exported and re-imported when adding a new disk device to the cluster VG. After adding the disk on primary node, vgexport the VG in preview mode, creating a map file on the primary node, and then vgexport and vgimport the VG using the created map file on all other nodes. Check the MC/SG manual and forums as well if still in doubt.
rariasn
Honored Contributor

Re: importing and exporting

Hi,

/usr/sbin/vgexport -m mapfile -s -p -v vg_name


Using the vgexport command generates a mapfile that can be copied to other systems that are part of a high availability cluster and the vgimport command (see vgimport(1M)) can be used to recreate the volume group.

See also vgchange(1M). The mapfile contains a description of the volume group and its associated logical volume(s) (if any). The
logical volume information found on the physical volumes is not modified. Note that with this option, the volume group is not removed from the system. (See the second example below).

The volume group specified in the mapfile can be shared with the importing systems. The volume group is not removed from the exporting system.

$ man vgexport

rgs,
Thierry D
Valued Contributor

Re: importing and exporting

Hi all,

I have some doubts after reading the different posts.

You all say when adding a new disk to a vg you need to export and import the vg configuration map to the other node of the cluster....

is it really necessary ? In the mapfile there is no information regarding the disk being used (strings mapfile), only information about VGs and LVs ....

I would think if we add a new disk to a VG in a cluster we would just need to do a pvscan on the other node, we would then scan the LVM header of the disk (PVRA, which holds specific info on the VG it belongs to, ...) and recreate the lvmtab ...


Any suggestion about how I see things?

Thierry