On Mon, Feb 11, 2019 at 9:42 AM Emmanuel Florac <eflorac@xxxxxxxxxxxxxx> wrote: > > Le Mon, 11 Feb 2019 17:23:59 +0100 > Emmanuel Florac <eflorac@xxxxxxxxxxxxxx> écrivait: > > > (I know, I could simply restore the vgcfgbackup if it existed... > > please let's suppose this command doesn't exist). > > > > Hum, actually I have a backup of the vg configuration but... it looks > like actually 7 or 8 kB were erased from the disk. So no dice, I must > recreate the PV first but wouldn't it wipe everything? You're right to be concerned. First you need to make sure the raid is assembled correctly, and functioning correctly. Is there a battery backed cache? There should be raid logs that can be extract to show at least what the hardware raid thinks happened; and if it did a cache flush upon power being restored, the log should indicate that as well. If it's possible the cache was full and the contents lost, that's also useful information. Anyway, get the raid sorted out first. Only then can you move on to LVM checking and recovery options, from the LVM list: linux-lvm@xxxxxxxxxx And only once LVM is healthy and working correctly can you move on to XFS checks and if advised, repair. Just be really deliberate, don't go stomping around with repairs and writes to disk that aren't deliberate, or you risk user induced data loss. Basically, I advise doing only checks and posting the results, and only doing repairs recommended on the applicable list by developers or expert users. All the lists will want information not provided in your first post, example: http://xfs.org/index.php/XFS_FAQ#Q:_What_information_should_I_include_when_reporting_a_problem.3F -- Chris Murphy
