vgimport query

[Adam Noble]

Hi all,

We have just carried out a bcv split and re-mounted the bcv volumes onto an alternate server. This is all good and it has worked but I thought it would fail. This is because the mapfiles which we have used to import the data were taken when the bcv's were mounted on a completely different host with different pv paths. I thought I would have had to use the -f option when exporting and importing to specify correct pv paths however this was not required. Why is this how does it know which disks to use to import the volumes on the new host. Hope this makes sense its Friday and late.

[Pete Randall · Accepted solution]

Hi Adam,

Did you happen to use the -s option?

The map files only contain the lvnames and are not even required if you used default (lvol1, lvol2, etc) names. The -s option scans the disks and reads the LVM information written in the lvm header files to determine what disks should belong in the VG.


Pete

[Rita C Workman]

I agree with you Adam, I would think it should have failed.
It would seem it worked by using the -s option, as Pete mentions, since obviously these devices could be shared between boxes....it didn't choke (this time). This is not a practice done here.

And I wouldn't trust that you will get the same results in every case. Using the -f tends to be the best and 'solid' way to ensure things come up as they should.

Just my 2cents,
Rgrds,
Rita