freebsd stable - Jul 2008 - softdepflush bad block error has led to negative blocks in free inode and handle_workitem_freeblocks: block count

Hi,
The problem started when i installed a kodicom 4400 card and started to run
zoneminder.
Prior to that no problems with my machine, which now runs
FreeBSD panix.internal.net 7.0-RELEASE-p3 FreeBSD 7.0-RELEASE-p3 #3: Mon Jul 14
16:35:37 EEST 2008
doroot@panix.internal.net:/usr/obj/usr/src/sys/GENERIC  i386
This hardware change happened in Sunday Jul 13.
The next day (Jul 14) morning "periodic daily" cron job at 03:01 gave:
/var/log/messages.1.bz2:Jul 14 03:01:04 panix kernel: pid 48 (softdepflush), uid
0 inumber 2662656 on /usr: bad block
/var/log/messages.1.bz2:Jul 14 03:01:04 panix kernel: pid 48 (softdepflush), uid
0 inumber 2662656 on /usr: bad block
/var/log/messages.1.bz2:Jul 14 03:01:04 panix kernel: pid 48 (softdepflush), uid
0 inumber 2662656 on /usr: bad block
/var/log/messages.1.bz2:Jul 14 03:01:04 panix kernel: pid 48 (softdepflush), uid
0 inumber 2662656 on /usr: bad block
... (15 times)
The funny think is that "df -h" showed a huge negative capacity.
Yesterday (Mon Jul 14) i had a crash when i tried to run (by hand) pkg_info .
Today (Mon Jul 15) the morning "periodic daily" cron job resulted in a
crash as well in when running find.

I speculated that it was one of those cases that bad memory, or overheated
memory could cause such problems
and i removed the most suspicious sim. After that i didnt get any crashes when
trying to run pkg_info or
periodic daily,weekly,monthly, but i get the following whenever i run periodic
weekly:
panix kernel: free inode /usr/2662656 had -3549356 blocks (negative)
and after a while
panix kernel: handle_workitem_freeblocks: block count

I suspect that even if i have a healthy system as far as memory is concerned (i
hope),
the problem with the 2662656 inode is still there.

Any thoughts are very welcome.

-- 
Achilleas Mantzios

???? Tuesday 15 July 2008 19:58:12 ?/? Achilleas Mantzios
??????:
> Hi,
> The problem started when i installed a kodicom 4400 card and started to run
zoneminder.
> Prior to that no problems with my machine, which now runs
> FreeBSD panix.internal.net 7.0-RELEASE-p3 FreeBSD 7.0-RELEASE-p3 #3: Mon
Jul 14 16:35:37 EEST 2008
> doroot@panix.internal.net:/usr/obj/usr/src/sys/GENERIC  i386
> This hardware change happened in Sunday Jul 13.
> The next day (Jul 14) morning "periodic daily" cron job at 03:01
gave:
> /var/log/messages.1.bz2:Jul 14 03:01:04 panix kernel: pid 48
(softdepflush), uid 0 inumber 2662656 on /usr: bad block
> /var/log/messages.1.bz2:Jul 14 03:01:04 panix kernel: pid 48
(softdepflush), uid 0 inumber 2662656 on /usr: bad block
> /var/log/messages.1.bz2:Jul 14 03:01:04 panix kernel: pid 48
(softdepflush), uid 0 inumber 2662656 on /usr: bad block
> /var/log/messages.1.bz2:Jul 14 03:01:04 panix kernel: pid 48
(softdepflush), uid 0 inumber 2662656 on /usr: bad block
> ... (15 times)
> The funny think is that "df -h" showed a huge negative capacity.
> Yesterday (Mon Jul 14) i had a crash when i tried to run (by hand) pkg_info
.
> Today (Mon Jul 15) the morning "periodic daily" cron job resulted
in a crash as well in when running find.
> 
> I speculated that it was one of those cases that bad memory, or overheated
memory could cause such problems
> and i removed the most suspicious sim. After that i didnt get any crashes
when trying to run pkg_info or
> periodic daily,weekly,monthly, but i get the following whenever i run
periodic weekly:
> panix kernel: free inode /usr/2662656 had -3549356 blocks (negative)
> and after a while
> panix kernel: handle_workitem_freeblocks: block count
> 
> I suspect that even if i have a healthy system as far as memory is
concerned (i hope),
> the problem with the 2662656 inode is still there.
> 
> Any thoughts are very welcome.
> 
I cleared the inode 2662656 with fsdb and clri and rerun fsck, and this seems to
have eliminated the problem.

-- 
Achilleas Mantzios