Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 19 Dec 97 19:29:00 -0000
From:      mikebw@bilow.bilow.uu.ids.net (Mike Bilow)
To:        aic7xxx@freebsd.org
Subject:   sd0: MEDIUM ERROR
Message-ID:  <49ac4c04@bilow.bilow.uu.ids.net>

next in thread | raw e-mail | index | archive | help

Penisoara Adrian wrote in a message to Mike Bilow:

 PA> I've recently ran again into an older problem; from time to 
 PA> time (like 4-5 months) I get a weird error message like this:

 PA> -- excerpt from /var/log/messages --
 PA> Dec 18 22:57:01 ady /kernel: sd0: MEDIUM ERROR info:0x36e823
 PA> csi:a,f2,2,62 asc:11,0 Unrecovered read error sks:80,70
 PA> Dec 18 22:57:04 ady /kernel: , retries:4
 PA> ------------------------------------

In my experience, the kernel means what it says when you get MEDIUM ERROR.

 PA> Sould I assume that my current harddisk (Quantum VP32170S 
 PA> 'plugged into' a AHA2940AU) is good just to hold (at most) the 
 PA> squid-swap from now on ? I sincerely wouldn't trust in him 
 PA> now...

I'm not sure what a "squid-swap" is, but I assume it's an odd translation from
Romanian?  In any case, I had a bunch of problems along the lines you are
having, although slightly more frequently.  I found that many hard drive
manufacturers set their internal mode pages to default to having automatic
reallocation on write disabled.  I can't think of any legitimate reason for
doing that, since AWRE is one of the major features of SCSI.

In the case of the old Seagate I had this problem with, I first ran the Ext2
filesystem check utility with the option of test reading all sectors.  This
truncated a couple of files, but at least I knew which ones they were and was
guaranteed that the filesystem was in a well-defined and stable state.  Then I
ran the "verify media" option built into the Adaptec ROM BIOS, which remapped 9
areas into the drive's "grown defect" table.  Finally, under Linux, I used the
Scsiinfo utility to manually enable AWRE in the appropriate mode page, and then
also to write that information in the drive's NVRAM.  That drive has been fine
ever since.  You can also periodically use Scsiinfo to check and see if any
areas have been added to the grown defect list.

 PA> Could this be related with some sudden reboots I experienced 
 PA> lately (no, my UPS didn't scream about power problems) ?

Sure.  If the drive was writing when some sort of mysterious power event
occurred, it could blow up a sector or two.  To some extent, this is considered
normal aging, especially with older technology SCSI drives.
 
-- Mike





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?49ac4c04>