Subject: panic during rm causes fsck to fill lost+found
To: None <current-users@netbsd.org>
From: Tracy Di Marco White <gendalia@iastate.edu>
List: current-users
Date: 12/29/2000 23:04:15
I was just wandering through the archives, and found the mail below.
(http://mail-index.netbsd.org/port-alpha/2000/04/26/0000.html)
I have just had this happen to me.  It's a pain.  I was removing
source directories (src pkgsrc xsrc src-1.5), and the computer
panic'd.  It was a kmem related panic, but I got neither a crash
dump nor any other information of use.  The system is running
1.5.1_ALPHA on i386.  Also seems to be the same as kern/11773.

This is my rather large RAID.  Over the long weekend we're planning
to back it up, recreate the RAID with sectPerSU set to 128 so it'll
be a bit faster.  fsck & parity rebuilds take a while, though.

I'd think that lost+found should not be full with under 9000 files
in it, though.

----
Subject: re: SOFTDEP on *DEC 3000*?
From: matthew green <mrg@eterna.com.au>
List: port-alpha
Date: 04/26/2000 13:09:53 

FWIW, i've been running softdep on a 3000/400 with 2 tcds cards and about
13 disks attached.. all of the filesystems on these disks use softdep and
it has worked almost flawlessly for me.

the only flaw was due to an unrelated crash happening at the wrong moment,
or something.  i was running `rm -rf' on a $DESTDIR that had a full
NetBSD/sparc build in it at the time the crash happened.  when rebooting,
fsck found *thousands* of unreferenced files and proceeded to place them
in lost+found.  the problem is that there were too many files for
lost+found and eventually it gave up and had to mount the not-quite-clean
file system, remove/delete these files from lost+found, umount and then
re-fsck.  this second fsck completed (recoverying many more thousands of
files)... 

i didn't lose any data, it just was very annoying to recover from.  i've
not heard of anyone else experiencing this, either.
----

Tracy J. Di Marco White
Project Vincent Systems Manager
gendalia@iastate.edu