gmirror or ata problem

Jeremie Le Hen jeremie at le-hen.org
Tue Feb 13 17:27:42 UTC 2007


Oliver,

On Tue, Feb 13, 2007 at 05:07:56PM +0100, Oliver Fromme wrote:
> Jeremie Le Hen wrote:
>  > FWIW, I had experienced this before likely because my drives went
>  > too warm (I am using gmirror as well).  I've bought a couple of fan
>  > racks and the problem hasn't occured since then.
>
> Last week we got a real read error reported on the disk
> (with a useful log entry, finally), so we arranged a
> replacement disk.  It has been working flawlessly so far.
>
> So maybe it was really the disk causing the trouble.
>
> However, the strange thing is that _no_ ATA errors were
> logged at all previously.  If gmirror disconnects a drive
> because of ATA trouble (i.e. a read/write error), I would
> expect that the ATA driver should report that read/write
> error, shouldn't it?

I can't remember if I had such messages.  However I suspect the
following messages to be issued by the ata layer:

% Jan 29 19:20:26 pluto kernel: ad1: FAILURE - device detached
% Jan 29 19:20:26 pluto kernel: subdisk1: detached
% Jan 29 19:20:26 pluto kernel: ad1: detached

BTW, this is is confirmed by grep(1):
% jarjarbinks:tataz$ grep "device detached" /sys/dev/ata/*
% /sys/dev/ata/ata-all.c:             device_printf(request->dev, "FAILURE - device detached\n");

The following gmirror error messages is then an expected
consequence.

Regards,
-- 
Jeremie Le Hen
< jeremie at le-hen dot org >< ttz at chchile dot org >


More information about the freebsd-geom mailing list