atom feed70 messages in org.freebsd.freebsd-scsiSCSI tape data loss
FromSent OnAttachments
Kern SibbaldJun 1, 2003 10:54 am 
Dan LangilleJun 1, 2003 11:32 am 
Justin T. GibbsJun 1, 2003 1:08 pm 
Kern SibbaldJun 1, 2003 2:44 pm 
Justin T. GibbsJun 1, 2003 3:39 pm 
Matthew JacobJun 1, 2003 5:00 pm 
Matthew JacobJun 1, 2003 5:13 pm 
Dan LangilleJun 1, 2003 6:58 pm 
Matthew JacobJun 1, 2003 7:03 pm 
Kern SibbaldJun 2, 2003 1:28 am 
Kern SibbaldJun 2, 2003 1:29 am 
Kern SibbaldJun 2, 2003 1:57 am 
Kern SibbaldJun 2, 2003 3:45 am 
Dan LangilleJun 2, 2003 4:28 am 
Matthew JacobJun 2, 2003 8:05 am 
Justin T. GibbsJun 2, 2003 8:10 am 
Dan LangilleJun 2, 2003 8:14 am 
Matthew JacobJun 2, 2003 8:21 am 
Kern SibbaldJun 2, 2003 8:27 am 
Dan LangilleJun 2, 2003 9:46 am 
Dan LangilleJun 2, 2003 11:05 am 
Matthew JacobJun 2, 2003 11:11 am 
Justin T. GibbsJun 2, 2003 11:49 am 
Dan LangilleJun 2, 2003 12:06 pm 
Justin T. GibbsJun 2, 2003 12:10 pm 
Matthew JacobJun 2, 2003 1:14 pm 
Dan LangilleJun 2, 2003 2:16 pm 
Matthew JacobJun 2, 2003 2:24 pm 
Kern SibbaldJun 2, 2003 2:46 pm 
Matthew JacobJun 2, 2003 2:55 pm 
Kern SibbaldJun 2, 2003 3:31 pm 
Carl ReisingerJun 2, 2003 3:44 pm 
Matthew JacobJun 2, 2003 3:44 pm 
Dan LangilleJun 2, 2003 6:37 pm 
Kern SibbaldJun 3, 2003 12:28 am 
Kern SibbaldJun 3, 2003 6:07 am 
Carl ReisingerJun 3, 2003 6:19 am 
Kern SibbaldJun 3, 2003 6:37 am 
Carl ReisingerJun 3, 2003 7:01 am 
Matthew JacobJun 3, 2003 7:34 am 
Justin T. GibbsJun 3, 2003 7:51 am 
Kern SibbaldJun 3, 2003 8:05 am 
Kern SibbaldJun 3, 2003 8:11 am 
Matthew JacobJun 3, 2003 9:03 am 
Dan LangilleJun 3, 2003 9:10 am 
Justin T. GibbsJun 3, 2003 9:24 am 
Kern SibbaldJun 3, 2003 9:40 am 
Justin T. GibbsJun 3, 2003 10:03 am 
Kern SibbaldJun 3, 2003 10:19 am 
Kern SibbaldJun 3, 2003 10:34 am 
Matthew JacobJun 3, 2003 11:00 am 
Matthew JacobJun 3, 2003 11:16 am 
Matthew JacobJun 3, 2003 11:39 am 
Justin T. GibbsJun 3, 2003 12:12 pm 
Dan LangilleJun 3, 2003 12:43 pm 
Matthew JacobJun 3, 2003 12:46 pm 
Kern SibbaldJun 3, 2003 1:05 pm 
PostMaster GeneralJun 3, 2003 2:21 pm 
Kern SibbaldJun 4, 2003 12:20 am 
Matthew JacobJun 4, 2003 7:51 am 
Kern SibbaldJun 4, 2003 9:51 am 
Kern SibbaldJun 6, 2003 7:38 am 
Dan LangilleJun 6, 2003 8:59 am 
Matthew JacobJun 6, 2003 11:50 am 
Dan LangilleJun 20, 2003 6:17 pm 
Dan LangilleJul 1, 2003 5:07 pm 
Matthew JacobJul 1, 2003 11:11 pm 
Michael L. SquiresAug 25, 2003 4:16 am 
Dan LangilleAug 25, 2003 9:13 am 
Michael L. SquiresAug 27, 2003 5:27 am 
Subject:SCSI tape data loss
From:Kern Sibbald (ke@sibbald.com)
Date:Jun 2, 2003 2:46:34 pm
List:org.freebsd.freebsd-scsi

On Mon, 2003-06-02 at 22:14, Matthew Jacob wrote:

Probably. Actually, it was 63k.

Most of Bacula writes are 64512 bytes, and all the data that was lost consisted of blocks of 64512 bytes.

But I sorta doubt that this was the issue.

A buddy of mine at Mirapoint did just remind me that physio can silently break up xfers that are even less than 64k if the buffer isn't page aligned- I'd forgotten about that. But I'm not sure that this is what is occurring.

The buffers are 64 bit aligned but not page aligned.

I need to think about this some more, but it may be that the actions that are being taken after EOM detection may be overwriting data. But don't take that to the bank at all.

Dan and I have been working on this for some time, so I'm sure there is data loss and that it is related to the EOM.

I suspect that the problem is something very simple such as the drive buffering data then hitting the physical EOM and of course any buffered data goes down the bit bucket.

-matt

On Mon, 2 Jun 2003, Justin T. Gibbs wrote:

And we have finish:

./tpt -v -b 5120 -r 10000000000 -n 10 -f /dev/nrsa0

Shouldn't the test be run with the 64k record size that Bacula uses?

-- Justin