Ext3-FS error in FC2 and 3

[John Trump]

Not sure if this email is actually making it ot the list my apologies
if this shows up several times.

I have two dell 2650's, 3 disk raid 5. Both servers on a frequent
basis gives the following error, only method to get the server back is
to power cycle it. Any suggestions as to what the cause of the error
is:
Ext3-FS error (device:sda2) in start_transaction journal has stopped
According to the raid card the raid5 volume is optimal.

[Stephen C. Tweedie]

Hi,

On Tue, 2005-04-26 at 15:53, John Trump wrote:
> I have two dell 2650's, 3 disk raid 5. Both servers on a frequent
> basis gives the following error, only method to get the server back is
> to power cycle it. Any suggestions as to what the cause of the error
> is:
> Ext3-FS error (device:sda2) in start_transaction journal has stopped
> According to the raid card the raid5 volume is optimal.

That means that there was some IO error sufficiently serious that the
filesystem took the journal offline, and turned the filesystem readonly,
to avoid further damage.

When that happens, the kernel will log a message telling you why it
happened.  It's that initial error that we need to see in order to know
why this is happening.

Cheers,
 Stephen

[Ralf Corsepius]

On Wed, 2005-05-04 at 16:20 +0100, Stephen C. Tweedie wrote:
> Hi,
>
> On Tue, 2005-04-26 at 15:53, John Trump wrote:
> > I have two dell 2650's, 3 disk raid 5. Both servers on a frequent
> > basis gives the following error, only method to get the server back is
> > to power cycle it. Any suggestions as to what the cause of the error
> > is:
> > Ext3-FS error (device:sda2) in start_transaction journal has stopped
> > According to the raid card the raid5 volume is optimal.
>
> That means that there was some IO error sufficiently serious that the
> filesystem took the journal offline, and turned the filesystem readonly,
> to avoid further damage.

Something similar had happened to me today, with on an FC3 system with a
single ide disk. A console message: "Ext3-FS error ..." appeared, 
afterwards the system kept running with all partitions mounted read-
only.

In /var/log/messages I can find this:

May  4 14:22:16 columbo kernel: hda: dma_timer_expiry: dma status ==
0x21
May  4 14:22:26 columbo kernel: hda: DMA timeout error
May  4 14:22:26 columbo kernel: hda: dma timeout error: status=0xd0
{ Busy }
May  4 14:22:26 columbo kernel:
May  4 14:22:26 columbo kernel: ide: failed opcode was: unknown
May  4 14:22:26 columbo kernel: hda: DMA disabled
May  4 14:22:59 columbo kernel: ide0: reset timed-out, status=0x80
> When that happens, the kernel will log a message telling you why it
> happened.  It's that initial error that we need to see in order to know
> why this is happening.selinux-policy-targeted-1.17.30-2.96

All this above happened during an "apt-get upgrade" while the system was
busy with a very intensive compilation job.

I suspect (Caution: wild guess!) be the upgrade of
selinux-policy-targeted
to be the culprit, because I already have experienced similar issues
with selinux updates on other systems, before.

Ralf