The officially official Devuan Forum!

paculino

Member

From: North Carolina, USA

Registered: 2023-12-26

Posts: 15  

Healthy drive, bad fs (inodes, alternative superblocks, groups)

Hello, some days ago I made the major mistake of allowing code for calculating terms in a number sequence run without any limits...' after one hundred minutes of unresponsive everything, I did a hard shutdown (I have since ran the code with limits and see that it should have been done in that time without limits, but maybe not). Upon rebooting, my partions mounted but had some files missing, so I booted elsewhere and checl it out with fsck, then preened it. Here's where my impatience and hubris really ruined me... when that failed, I tried auto, and it still mounted, but then I tried, still on the original (I had no other drive), manual fsck, and I agreed to optimization once out of two prompts, which failed. Falsely thinking it wouldn't get worse, I tried it with -y, without even checking alternative superblocks. Two days later I cloned it onto my new drive.

I suspect that if I were to correctly get the inode and block, mke2fs -S would fix it or at least undo the optimize shrink, but the guess  it had failed so badly that the alternative superblocks were never found. I checked every alternative superblock on another clone, and the invalid checksum for the inode was now for inode 7 rather than 2, but it remains unmountable.

At every step, groups 384-447 overlapped with another fs, and this is nearly 800 GB (and cloned to 800 GB partitions), so I believe it is relatively early. Is there a way to skip that section and get a recovery of the directory afterwards? I do have an old backup and a more recent partial backup, so can more or less get the directories back and, assuming creation time metadata remains after recovering files without directories, get them populated with many of the same files as before, but this is not my first choice.

Short version:
Healthy drive, mismatched/missing (checksums) for inodes/altsuperblocks and groups 384-447 overlapping with other fs no matter what I do. Can I skip those (I assume either near oldest or newest) sections? I believe optimization shrinkage is to blame for unmountability, but the issues are deeper.