fsck fails/lost+found questions

[Mark D. Holbert]

Eric,

There was a great article in the latest Linux Magazine (August 2000) about
Journaling Filesystems.  It covers the current extfs2 filesystem and how it
works down to the inodes and what happens during a crash, i.e. how a crash
causes lost inodes to be placed in the lost+found directory.  Here is a
quote from the article...

"...all of the entries in the corrupted directory can be "lost," (which
means they get linked into the lost + found directory for each filesystem.
Blocks put into the lost + found directory are in use, but there is no way
to know where they are referenced from)."

The article then goes into how a journaling file system works and
descriptions about each that are going to be available in the near future
such as XFS - from Silicon graphics, JFS - from IBM, Ext3fs from Stephen
Tweedie and Red Hat, and others.  The article will be found at
www.linux-mag.com as a back issue in about 2 months as they won't put
current issues on the web.

Hope that helps.

Mark Holbert
mark@linuxsoftware.org

>Date: Fri, 4 Aug 2000 22:27:37 -0700 (MST)
>From: Eric Thelin <eric@thelin.org>
>To: plug-discuss@lists.PLUG.phoenix.az.us
>Subject: fsck fails/lost+found questions
>Reply-To: plug-discuss@lists.PLUG.phoenix.az.us
>
>I have a machine running a database that has crashed a number of times
>due to flaky power/UPS issues.  I know this is VERY bad and have since
>fixed it.  But that is not the point that I have a question about.
>Every time this machine crashed it would fail the fsck on startup to the
>point that I had to login and run fsck manually.  When running manually
>I get asked a lot of questions about deleting duplicate inodes and
>inodes with bad times (or something like that it has been a long time).
>Because I didn't know what else to do I just went with the default which
>was always Y.  After a while I went started adding a -y which just
>answers Y and doesn't even ask.  So now the big question that I have is
>what dammage has been gettin done and is there anything that can done
>about it?  Other than the obvious don't let the power fail or run a
>journaling filesystem.  I now have thousands of entries in my lost+found
>directory that I assume came from these filesystem scans as some kind of
>safety net.  But what could I do with them?  If there is no way for a
>mere mortal to use them is it safe to delete them?  They have seemingly
>random permissions, ownership, and timestamps so I am vary scared to do
>anything with them.  My guess is that they are essentially hard links to
>the block that was dammaged.  But I am certainly no fs hacker and I have
>never seen any documentation explaining lost+found or what to do with
>it.  Any help would be great and probably enlightening to many people on
>this list.
>
>Eric Thelin
>aztechbiz.com
>