Amanda-Users

Re: Amrestore error

2005-02-20 13:10:50
Subject: Re: Amrestore error
From: Gene Heskett <gene.heskett AT verizon DOT net>
To: amanda-users AT amanda DOT org
Date: Sun, 20 Feb 2005 13:01:46 -0500
On Sunday 20 February 2005 12:13, Spicer, Kevin (MBLEA it) wrote:
>Hello,
>
>I'm having a strange problem with amrestore.  I get an I/O error in
> the middle of a restore and amrestore exits, but st0 device remains
> busy and only a reboot can clear it.  There is an error on the
> console as follows...
>
>st0: Error 70002 (sugg. bt 0x0, driver bt 0x0, host bt 0x7)
>
>... My first thoughts were hardware errors, but I've checked out
>everything I can (short of buying a new library!) and everything
> seems fine.  Perhaps more revealingly I have no issues performing
> backups with amanda or extracting data from the tape using dd.
>
>Heres the amidxtaped.debug file (I originally experienced the
> problem while using amrecover)
>
>###### START amidxtaped.debug [extract] ########
>
>amidxtaped: time 6.803: Ready to execv amrestore with:
>path = /usr/local/sbin/amrestore
>argv[0] = "amrestore"
>argv[1] = "-p"
>argv[2] = "-h"
>argv[3] = "-l"
>argv[4] = "UnixDaily-FHH545"
>argv[5] = "-f"
>argv[6] = "72"
>argv[7] = "/dev/nst0l"
>argv[8] = "^myhost.mydomain.com$"
>argv[9] = "^/export/home$"
>argv[10] = "20050219"
>amrestore:  72: restoring
> myhost.mydomain.com._export_home.20050219.0 amrestore: read error:
> Input/output error
>amidxtaped: time 38.951: amrestore terminated normally with status:
> 2 amidxtaped: time 38.951: rewinding tape ...
>
>gzip: stdin: unexpected end of file
>
>####### END ######
>
>Running the amrestore command manually (redirecting output to a
> file) gives the same errors.  The command was...
>
>amrestore -p -h -l UnixDaily-FHH545 -f 72 /dev/nst0l
>^myhost.mydomain.com$ ^/export/home$ 20050219 > 
> 1-amrestore-fromtape
>
>The resulting output file is indeed smaller than expected.
>
>Then...
>mt -f /dev/nst0l rewind
>my -f /dev/nst0l fsf 72
>dd if=/dev/nst0l of=2-dd-fromtape bs=4194304               [My tape
>block size is 4096k]
>
>...results in the full file being copied from disk with no errors. 
> I can then extract the dump from the file with amrestore...
>
>amrestore -p -b 4096k -h ./2-dd-fromtape > 3-amrestore-fromdd
>
>Which is fine.
>
>In case it was relevent I also tried (by trial and error) to see if
> the truncated version from the original amrestore fell on a tape
> block boundary - it did (22 blocks including the header).   I found
> this by truncating the dd'd image then running amrestore on it.
>
>dd if=2-dd-fromtape of=4-dd-truncated-to-22-blocks bs=4194304
> count=22 amrestore -p -b 4096k -h ./4-dd-truncated-to-22-blocks >
>5-amrestore-fromddtruncated
>
>The file sizes from the above can be seen below...
>
>-rw-r--r--    1 amanda   disk     131203072 Feb 20 15:54
>1-amrestore-fromtape
>-rw-r--r--    1 root     root     343932928 Feb 20 16:04
> 2-dd-fromtape -rw-r--r--    1 root     root     498728960 Feb 20
> 16:11
>3-amrestore-fromdd
>-rw-r--r--    1 root     root     92274688 Feb 20 16:15
>4-dd-truncated-to-22-blocks
>-rw-r--r--    1 root     root     131203072 Feb 20 16:15
>5-amrestore-fromddtruncated
>
>All of which seems to point to some problem with the way amrestore
>interacts with the tape device.  Has anyone seem this before?
>
>For completeness:
>Amanda 2.4.4p4
>Tao Linux release 1 (Mooch Update 4) [ Rebuild of Red Hat Enterprise
> 3 ] Compaq Proliant 1600R
>Overland Powerloader with Quantum SDLT320 drive
>Adaptec 29160 Scsi card
>
>My tapetype definition...
>define tapetype Quantum-SDLT320 {
>    comment "Quantum SDLT 320 hardware compression off/ blocksize
> 4m" length 159080 mbytes
>    filemark 4096 kbytes
>    speed 15762 kps
>    blocksize 4096 kbytes
>}
>[The reason for the blocksize change was to improve speed, it now
>achieves speeds approaching the manufacturers quoted ones]

I'm not sure whats happening at the 92 megabyte mark, but I would make 
an observation re the block size you are using.  To me, that seems 
rather large.  To do a checksum over a 4 megabyte block might 
possibly be running into a math problem because much of that code is 
legacy, and probably never expected anybody to use more than 128k, 
and I personally haven't explored the range above 32k.

OTOH, not having walked thru that code, I could be full of it.  But 
thats my intuitive feel for this.

>BMRB International
>http://www.bmrb.co.uk
>+44 (0)20 8566 5000

[snip totally irrevelant legalese]

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.33% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com attorneys please note, additions to this message
by Gene Heskett are:
Copyright 2005 by Maurice Eugene Heskett, all rights reserved.

<Prev in Thread] Current Thread [Next in Thread>