Re: 2.5.47 / unusual ext3 fs errors

[Andrew Morton]

CaT wrote:
> 
> Under 2.5.x I seem to be getting a lot of fs errors on fsck, mainly
> dealing with bad inode counts in groups. Just now though, I had /var
> remounted read-only due to the following:
> 
> t_transaction: Journal has aborted
> EXT3-fs error (device ide0(3,9)) in start_transaction: Journal has aborted
> EXT3-fs error (device ide0(3,9)) in start_transaction: Journal has aborted
> EXT3-fs error (device ide0(3,9)) in start_transaction: Journal has aborted
> ...
> EXT3-fs error (device ide0(3,9)) in start_transaction: Journal has aborted
> EXT3-fs error (device ide0(3,9)) in start_transaction: Journal has aborted
> EXT3-fs error (device ide0(3,9)) in start_transaction: Journal has aborted
> 
> And again, on reboot into single user mode and a full fsck bad inode
> count errors were present. There were no errors detected whilst testing
> the disk with -c.

There was a problem in 2.5.46 which mucked up these counters.  On disk.

2.5.47 fixed that bug, and now you have fixed the on-disk mess which 2.5.46
created, all should be well.