Date: Wed, 30 May 2001 09:35:26 -0700 (PDT) From: Matt Dillon <dillon@earth.backplane.com> To: Mike Silbersack <silby@silby.com> Cc: Brian Reichert <reichert@numachi.com>, Peter Wemm <peter@wemm.org>, John Polstra <jdp@polstra.com>, <hackers@FreeBSD.ORG> Subject: Re: cvsup.freebsd.org I/O error Message-ID: <200105301635.f4UGZQn40890@earth.backplane.com> References: <20010530111251.R71465-100000@achilles.silby.com>
next in thread | previous in thread | raw e-mail | index | archive | help
I went through some of those messages myself. I didn't read any IBM
technical reps, only marketing dweebs who told people that it was
the controller, which is bullshit. PC motherboard controllers do not
cause hard read errors on disks.
In my case it was a hard-read error on a data block in a file in
/usr/ports. After removing the file I was able to cpdup the filesystem
(along with all the other filesystems on that disk) to a SCSI seacrate.
So there was only that one hard error. Of course, I didn't wait around
for others to show up :-)
Now I *did* see some unrelated controller/drive issues. A few days
earlier I was dd'ing the drive to /dev/null as a test and about half
way through I started getting DMA errors on the console and it backed off
to PIO mode (30MB/sec -> 6MB/sec). This is completely unrelated to
the hard error and never occured under normal operation, only when I
was doing the dd test.
-Matt
:If you do a search for "IBM DTLA failure rate" on google or deja (which
:would also be google, I suppose), you'll find a bunch of threads on the
:issue. The posts are divided into two types:
:
:1. IBM fanboys claiming that there's a conspiracy to tarnish the DTLA's
:name.
:
:2. Posts by people who say they're on their second replacement drive, and
:starting to see failure again.
:
:So, while it's not conclusive, it appears that there's something going on.
:
:Note that the people claiming failures say that they start as sectors
:going bad. Therefore, it seems entirely possible that those running their
:DTLAs with FAT32 won't notice a problem until they have the drive nearly
:full. Those with inode based FSes would probably see strange things a bit
:sooner. Unfortunately, I don't think you can run badblocks or anything
:similar right now. Maybe someone can suggest another test method.
:
:Mike "Silby" Silbersack
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200105301635.f4UGZQn40890>