On 01/19/2012 07:56 AM, Muhammad Panji wrote:

Dear All, I have one VG with one LV inside consist of four disk (and four PV) somehow the UUID is changed. I try to restore with the last known good configuration and use

pvcreate --uuid xxx --restorefile xxx

but I think when I first time do it I did use wrong UUID for two device. but haven't only using command above and haven't used vgcfgrestore. After carefully read the configuration, using hdparm to know device serial number I think I have assign the right UUID for each device. I succesfully use vgcfgrestore, but I cannot mount the device (unknown filesystem, bad superblock, etc). My questions are, is my first attempt to assigning (wrong) UUID make all the LVM corrupt? I see that all LE and PE is still there, the VG and LV size is right but still I cannot mount the volume. Is this just a bad filesystem or the data really gone? Thank you. Regards,

Hi,

Why theses 4 PV lost UUID? Have some hardware change ?

It's not common this happen, just lost LVM Meta Data on PV, something wrong happened.

The problem is reconstruct meta information about Mapping of blocks ( PE ) to LV, if your LV used blocks mapped across the four disk, I recommend you running check file system.. e2fsck to repair.

Well if e2fsck don't solve this problem, it time find out the last backup. :(

best regards, --aslan