FSCK of corrupted ext3 filesystem

Matt Bernstein mb/ext3 at dcs.qmul.ac.uk
Fri Aug 26 12:16:34 UTC 2005


On Aug 25 Stephen C. Tweedie wrote:

> Which version of e2fsprogs?  There were some serious algorithmic
> inefficiencies in some really old versions that could explain this sort
> of thing.

e2fsprogs-1.37-4 from Fedora Core 4 x86_64.

On Aug 25 Theodore Ts'o wrote:

> What sort of messages was e2fsck printing beforehand?

Nothing I found overly unusual. A few errors in pass 1; it reached the 
"checking directory structure" pass (pass 2 IIRC) and then nothing.

I guess at this point I should confess that the filesystem had (as now 
I've trashed it again with mkfs) of the order of hundreds of millions of 
directory entries, and tens of millions of inodes.

It is used to rsync across our staff fileserver (currently around 500G 
used). We "cp -rpl" the previous night before rsync-ing the current night.
It's fast and makes for cheap restores at the expense of a dense fs.

The same volume is our amanda holding space--so I'm afraid (as I'm about 
to take a holiday) I had to trash it so that I could get some vaguely 
recent backups before what could have been a gap of 3-4 weeks.

Sorry not to be helpful

Matt




More information about the Ext3-users mailing list