WRITE_DMA problem again

[Anthony Atkielski]

arax writes:

> I was just wondering if there's a solution to fix the WRITE_DMA problem. I've
> searched the archives but couldn't find a complete answer to this,
> like why it's happening and what to do about it. I'd appreciate any help.

I had the same problem (same logical device, in fact, and also on
FreeBSD 5.3) and it turned out to be a bad drive. I was able to verify
this by downloading the vendor's standalone test software, booting it on
the server, and running exhaustive tests against the drive. On the full
test, the software hung (whereas it worked fine on the other, identical
drive that I had configured on the system). Since this was a bootable
disk and FreeBSD wasn't even in the system, this ruled out FreeBSD. I
reluctantly replaced the drive (which was only 60 days old or so) with
an identical model, and I've had no problems since.

> It seems that this doesn't do any harm to my system, but I'm not sure
> because at this time the system is under no load at all. These errors just
> appear from time to time. I would like to fix this before I move the
> server into production.

The problem eventually corrupted some structure on my drive, but
apparently the system was able to recover most of it.  The drive worked
well enough to allow me to run full backups before replacing it.
Fortunately, there wasn't much on it.

If your disk vendor offers bootable test software, download that and
test the drive exhaustively offline.  If it fails or hangs, you have
your answer: replace the drive.  If it passes with flying colors, then
you'll have to look back to FreeBSD.

-- 
Anthony