block size mismatch
July 28, 2014 02:09AM
At a customer I get this:

$ amcheck daily
Amanda Tape Server Host Check
-----------------------------
Holding disk /mnt/amhold/daily: 592780 MB disk space available, using
589708 MB
slot 1: Device /dev/nst0 use fixed block size of 512 and tapetype use 32768
all slots have been loaded
Taper scan algorithm did not find an acceptable volume.

Nothing changed in amanda.conf, seems related to upgrading to
amanda-3.3.5 a few days ago.

What can I do? Downgrade? ;-)

Thanks, S
block size mismatch
July 28, 2014 02:18AM
Am 28.07.2014 11:04, schrieb Stefan G. Weichinger:
[quote]
At a customer I get this:

$ amcheck daily
Amanda Tape Server Host Check
-----------------------------
Holding disk /mnt/amhold/daily: 592780 MB disk space available, using
589708 MB
slot 1: Device /dev/nst0 use fixed block size of 512 and tapetype use 32768
all slots have been loaded
Taper scan algorithm did not find an acceptable volume.

Nothing changed in amanda.conf, seems related to upgrading to
amanda-3.3.5 a few days ago.

What can I do? Downgrade? ;-)
[/quote]
... amanda-3.3.3 works with exactly this tape and config.
block size mismatch
July 28, 2014 08:09AM
We added code to check if the tape have the correct block size.

What is the blocksize of the tape?
mt -f /dev/nst0 status

What kind of tape is it?

Jean-Louis

On 07/28/2014 05:12 AM, Stefan G. Weichinger wrote:
[quote]Am 28.07.2014 11:04, schrieb Stefan G. Weichinger:
[quote]At a customer I get this:

$ amcheck daily
Amanda Tape Server Host Check
-----------------------------
Holding disk /mnt/amhold/daily: 592780 MB disk space available, using
589708 MB
slot 1: Device /dev/nst0 use fixed block size of 512 and tapetype use 32768
all slots have been loaded
Taper scan algorithm did not find an acceptable volume.

Nothing changed in amanda.conf, seems related to upgrading to
amanda-3.3.5 a few days ago.

What can I do? Downgrade? ;-)
[/quote]... amanda-3.3.3 works with exactly this tape and config.

[/quote]
block size mismatch
July 28, 2014 08:28AM
Am 28.07.2014 17:02, schrieb Jean-Louis Martineau:
[quote]mt -f /dev/nst0 status
[/quote]

SCSI 2 tape drive:
File number=0, block number=64, partition=0.
Tape block size 512 bytes. Density code 0x42 (LTO-2).
Soft error count since last status=0
General status bits on (1010000):
ONLINE IM_REP_EN

... LTO-2 ...

What does that mean?
Do I have to relabel tapes with amanda 3.3.5 or edit my tapetype?

I don't remember that issue with my own LTO-2 drives (where I use 3.3.5
and 3.3.6 now for quite a while).

Stefan
block size mismatch
July 28, 2014 08:58AM
On 07/28/2014 11:23 AM, Stefan G. Weichinger wrote:
[quote]Am 28.07.2014 17:02, schrieb Jean-Louis Martineau:
[quote]mt -f /dev/nst0 status
[/quote]
SCSI 2 tape drive:
File number=0, block number=64, partition=0.
Tape block size 512 bytes. Density code 0x42 (LTO-2).
Soft error count since last status=0
General status bits on (1010000):
ONLINE IM_REP_EN

... LTO-2 ...

What does that mean?
[/quote]The tape is in fixed 512 bytes block, this is bad.
it should be in variable block size.
[quote]Do I have to relabel tapes with amanda 3.3.5 or edit my tapetype?
[/quote]All tapes written with a blocksize smaller than 32K must be rewritten.
They probably needs to be erased before relabeling with a different
block size.

Jean-Louis
block size mismatch
July 28, 2014 09:00AM
Am 28.07.2014 17:52, schrieb Jean-Louis Martineau:

[quote][quote]What does that mean?
[/quote]The tape is in fixed 512 bytes block, this is bad.
it should be in variable block size.
[/quote]
I don't know how that comes ... never intentionally set anything here.

[quote][quote]Do I have to relabel tapes with amanda 3.3.5 or edit my tapetype?
[/quote]All tapes written with a blocksize smaller than 32K must be rewritten.
They probably needs to be erased before relabeling with a different
block size.
[/quote]
... phew. A bit of a task for a productive environment.

How to erase? dd something?
Could you provide a howto (maybe later on the wiki) ?

Thanks!
Stefan
block size mismatch
July 28, 2014 10:19AM
On Monday, July 28, 2014 05:55:40 PM you wrote:
[quote][quote]All tapes written with a blocksize smaller than 32K must be rewritten.
They probably needs to be erased before relabeling with a different
block size.
[/quote]
... phew. A bit of a task for a productive environment.

How to erase? dd something?
Could you provide a howto (maybe later on the wiki) ?

Thanks!
Stefan
[/quote]

I actually had to do this once, I used this script with an LTO-3 changer:

#!/bin/sh

# $1 = slot
# $2 = tape name

/usr/sbin/mtx -f /dev/sg3 load $1
/bin/mt -f /dev/st0l compression 0
/bin/mt -f /dev/st0l setblk 32768
/bin/mt -f /dev/nst0l compression 0
/bin/mt -f /dev/nst0l setblk 32768
/bin/dd if=/dev/zero bs=32768 count=200 of=/dev/st0l
/bin/mt -f /dev/st0l compression 0
/bin/mt -f /dev/st0l setblk 32768
/bin/mt -f /dev/nst0l compression 0
/bin/mt -f /dev/nst0l setblk 32768
/usr/sbin/mtx -f /dev/sg3 unload $1
/usr/sbin/amlabel -f DailySet1 $2 slot $1
/bin/mt -f /dev/st0l compression 0
/bin/mt -f /dev/st0l setblk 32768
/bin/mt -f /dev/nst0l compression 0
/bin/mt -f /dev/nst0l setblk 32768
/usr/sbin/mtx -f /dev/sg3 unload $1

It worked ... and apparently I had some problems with the different tape device
aliases along the way, although I don't remember the particulars. It probably
does more things than are strictly necessary.

I also seem to remember that one time a drive started dying, and one of the
symptoms was block size errors ... if you continue to have problems, it may be
a hardware issue.
block size mismatch
July 28, 2014 10:30AM
Am 28.07.2014 19:13, schrieb Alan Hodgson:
[quote]On Monday, July 28, 2014 05:55:40 PM you wrote:
[quote][quote]All tapes written with a blocksize smaller than 32K must be rewritten.
They probably needs to be erased before relabeling with a different
block size.
[/quote]
... phew. A bit of a task for a productive environment.

How to erase? dd something?
Could you provide a howto (maybe later on the wiki) ?

Thanks!
Stefan
[/quote]

I actually had to do this once, I used this script with an LTO-3 changer:

#!/bin/sh

# $1 = slot
# $2 = tape name

/usr/sbin/mtx -f /dev/sg3 load $1
/bin/mt -f /dev/st0l compression 0
/bin/mt -f /dev/st0l setblk 32768
/bin/mt -f /dev/nst0l compression 0
/bin/mt -f /dev/nst0l setblk 32768
/bin/dd if=/dev/zero bs=32768 count=200 of=/dev/st0l
/bin/mt -f /dev/st0l compression 0
/bin/mt -f /dev/st0l setblk 32768
/bin/mt -f /dev/nst0l compression 0
/bin/mt -f /dev/nst0l setblk 32768
/usr/sbin/mtx -f /dev/sg3 unload $1
/usr/sbin/amlabel -f DailySet1 $2 slot $1
/bin/mt -f /dev/st0l compression 0
/bin/mt -f /dev/st0l setblk 32768
/bin/mt -f /dev/nst0l compression 0
/bin/mt -f /dev/nst0l setblk 32768
/usr/sbin/mtx -f /dev/sg3 unload $1

It worked ... and apparently I had some problems with the different tape device
aliases along the way, although I don't remember the particulars. It probably
does more things than are strictly necessary.

I also seem to remember that one time a drive started dying, and one of the
symptoms was block size errors ... if you continue to have problems, it may be
a hardware issue.
[/quote]
Thanks for the script ... I don't have a changer there ... and I would
have to do the erasing before every single amdump ... because if I do it
for all the tapes in one session, there are no valid backups anymore ;-)

Maybe I just keep amanda-3.3.3 there for now ... I have to plan this
somehow.

For the hw-issues: I will soon do some test-restores ...

btw: as amanda is able to eject tapes now after amdump, it would also be
fine if we could tell it to disable compression before backups as well,
right? (sure, via boolean parameter in amanda.conf)

thx, Stefan
block size mismatch
July 28, 2014 10:32AM
On Monday 28 July 2014 11:52:28 Jean-Louis Martineau did opine
And Gene did reply:
[quote]On 07/28/2014 11:23 AM, Stefan G. Weichinger wrote:
[quote]Am 28.07.2014 17:02, schrieb Jean-Louis Martineau:
[/quote][/quote]
[quote][quote]
SCSI 2 tape drive:
File number=0, block number=64, partition=0.
Tape block size 512 bytes. Density code 0x42 (LTO-2).
Soft error count since last status=0

General status bits on (1010000):
ONLINE IM_REP_EN

... LTO-2 ...

What does that mean?
[/quote]
The tape is in fixed 512 bytes block, this is bad.
it should be in variable block size.

[quote]Do I have to relabel tapes with amanda 3.3.5 or edit my tapetype?
[/quote]
All tapes written with a blocksize smaller than 32K must be rewritten.
They probably needs to be erased before relabeling with a different
block size.

Jean-Louis
[/quote]
I've found on some of the smaller tape formats that I could afford and
therefore play with, that this can be accomplished by issuing a rewind,
and without doing anything else that would cause the drive to re-scan the
tape, immediately write the new label block, a 32k block. The sequence is
easy enough to do, so its at least worth a try. Do that, eject the tape,
reinsert it so the drive recognizes it again, and then check the output of

mt -f /dev/nst0 status

If it works, its many hours less wear on the drive heads than a full
format, something I have never actually done to any of several QIC's,
Travan, or DDS-2 tapes I've used. The above Just Worked(TM) with all of
them.

There may be an mt command to change that block size, in that case do it,
then write the header with amlabel before the drive re-reads the tape and
over-rides your choice.

This procedure will also enable turning off the drives compression so that
amanda can know to a much more accurate level, the exact capacity of the
tape.

Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page http://geneslinuxbox.net:6309/gene
US V Castleman, SCOTUS, Mar 2014 is grounds for Impeaching SCOTUS
Sorry, only registered users may post in this forum.

Click here to login