last 543 kenel -> filesystem errors

[Stephen C. Tweedie]

Hi,

On Mon, 2004-09-06 at 21:48, Leonard den Ottolander wrote:

> On Sun, 2004-09-05 at 21:33, Andy Green wrote:
> > Okay... bad news.  I fixed /boot as shown in the parent and went out for the 
> > day.  Now I look again and...

> > /dev/hdc1              99M  -16Z  105M 101% /boot
> > [root at fastcat ~]# uname -r
> > 2.6.8-1.538
> 
> > It's deja vu all over again then.... 543 ext3 is pooping on the fs somehow... 

> ? The above suggests the issue is already existent in 538. So which is
> it?

It's 538.  538 corrupted these counts *on disk*, so just upgrading the
kernel won't fix them --- you need a fsck.

An inconsistency in the block counts is not serious enough to cause ext3
to abort the fs, so it won't set the flag that forces an automatic fsck
on the next boot.  So once 538 causes that problem, even once you
upgrade to a newer kernel the problem will persist until you force a
full fsck.  The easiest way to force a full fsck is 

	touch /.autofsck
	/sbin/reboot

--Stephen