If you move them to CORRUPTED (either inside the namespace or outside) then Lustre shouldn't do anything further to them. That would also prevent the blocks from being reallocated.
The other option would be to run "e2fsck -l" and feed it the bad block numbers and have those stored in the ext4 bad blocks inode. That should prevent ext4 from ever reallocating those blocks again, but note that the badblocks feature is rarely used these days as most devices had internal bad block remapping. Cheers, Andreas On Dec 21, 2016, at 08:03, Tommi T <[email protected]> wrote: > Hi, > > We have 33 files/inodes which we'd like to move out MDT ROOT/ directory due > to FS corruption issue. If I mount MDT as as ldiskfs and create some > directory (e.g. CORRUPTED) and move corrupted files from ROOT/path/file to > CORRUPTED, will lustre ignore those files? > > Another option is just mark those inodes as a used with debugfs so that > e2fsck could "fix" the fs. > > Corruption is not lustre fault, something writes EFI partition label on top > of lustre fs. It's exactly at $(fs size) - 2 TB location so atm. we suspect > Qlogic FC card bios settings. > > TIA, > Tommi > _______________________________________________ > lustre-discuss mailing list > [email protected] > http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org
_______________________________________________ lustre-discuss mailing list [email protected] http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org
