> [ ... ] the nilfs2 filesystems were affected nonetheless as > the machine suffered several hard resets. [ ... ] Most likely then some corrupted data was written to the filesystem, because of incomplete barrier implementation, or because of attempts to write damaged sectors or some sectors are too damaged to read. > [ ... ] (sdd1): bad btree node (ino=6, blocknr=1500850679): level = 0, flags = 0x0, nchildren = 0 > [ ... ] (device sdd1): nilfs_bmap_lookup_at_level: broken bmap (inode number=6) Looks like some metadata blocks were zeroed, perhaps because some sectors became unreadable. > [ ... ] (sdd1): error -5 reading inode: ino=34057 $ errno 5 EIO 5 Input/output error Indeed that's an IO error: the disk is damaged and some sectors are unreadable. > [ ... ] How can I fix The first thing is to talk with the system administrator of the system and ask them to check that the disks have no bad blocks, and to check SMART logs for signs of bad block substitution (almost certainly there will be some); after that you can attempt to mount a previous checkpoint, going back in the checkpoint list until un "undamaged" checkpointn is found. Howevever NILFS2 already does a checksum verification on mounting, so it won't mount damaged checkpoints, and the errors above are not checksum errors. It looks like that because of IO errors or bad sectors (still present apparently) the metadata and perhaps also data in the filesystem was corrupted. NILFS2 like most other filesystems does not have built-in per-sector redundancy like RAID, only checkpoint redundancy, which is not as comprehensive. -- To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
