On Fri, 2002-01-04 at 00:35, I wrote:
> I will verify this theory tonight with an acid test: An Amanda run
> (amdump/amverify) on a freshly erased (and labeled) tape _should_
> succeed. I'll tell you in the morning. :-)
Hi,
here are my results. Unfortunately, they're not too good:
amverify adr
Fri Jan 4 05:41:07 MET 2002
Loading current slot...
Using device /dev/nst0
Volume adr000, Date 20020104
Checked host1._mnt_disk1.20020104.0
Checked host2.__samba_disk.20020104.0
Checked host3._mnt_disk0.20020104.0
Checked host4._mnt_disk1.20020104.0
Checked host3._mnt_disk1.20020104.0
Checked host4._mnt_disk0.20020104.0
Checked host5._boot.20020104.0
** Error detected ()
amrestore: WARNING: not at start of tape, file numbers will be offset
amrestore: 0: reached end of information
** No header
0+0 in
0+0 out
I took from the amdump logfile some times for the inter-file gaps. The
number in front of the "FILE-WRITE" is the time (from the amanda log) of
the end of the write (taper return) and the start of the next dump.
FILE-WRITE 02-00008
DONE 02-00008 adr000 1
0 FILE-WRITE 02-00019
DONE 02-00019 adr000 2
0 FILE-WRITE 02-00021
DONE 02-00021 adr000 3
0 FILE-WRITE 02-00023
DONE 02-00023 adr000 4
0 FILE-WRITE 02-00024
DONE 02-00024 adr000 5
0 FILE-WRITE 02-00025
DONE 02-00025 adr000 6
0 FILE-WRITE 02-00026
DONE 02-00026 adr000 7 In this file, the first errors occur.
As you see, all dump come directly from the dumping disk which has
enough files to keep the tape going. No 120 second gap anywhere. There
is however, such a gap (even a really huge gap of something like 90
minutes, because one of the machines to back up wasn't online and I run
with a timeout of 90 minutes for the estimates) between writing the new
amanda tape label and the first dump. But verifying the first few
backups on this tape is ok.
An interesting datapoint may be, that _all_ of the first six backups,
the disks are basically empty:
DUMP SUMMARY:
DUMPER STATS TAPER STATS
HOSTNAME DISK L ORIG-KB OUT-KB COMP% MMM:SS KB/s MMM:SS KB/s
------------------------ --------------------------------- ----------
host1 /mnt/disk1 0 64 64 100.0 0:00 0.0 0:36 1.8
host2 samba/disk 0 10 64 -- 0:07 1.4 0:09 7.1
host3 /mnt/disk0 0 64 64 100.0 0:00 0.0 0:09 7.2
host4 /mnt/disk1 0 64 64 100.0 0:00 0.0 0:09 7.1
host3 /mnt/disk1 0 64 64 100.0 0:00 0.0 0:09 7.0
host4 /mnt/disk0 0 64 64 100.0 0:00 0.0 0:09 7.1
(32K of Header and 32K of data, which is simply an emtpy tar block)
Still, everything points at a firmware bug. I will try to hack the
amanda software tonight so that the tape label is not written when the
estimates are started but when the real backups start. I don't know how
complicated this will be (amanda hackers?) but I'll give it a shot. I
will try to contact Onstream on this matter, too.
Regards
Henning
--
Dipl.-Inf. (Univ.) Henning P. Schmiedehausen -- Geschaeftsfuehrer
INTERMETA - Gesellschaft fuer Mehrwertdienste mbH [EMAIL PROTECTED]
Am Schwabachgrund 22 Fon.: 09131 / 50654-0 [EMAIL PROTECTED]
D-91054 Buckenhof Fax.: 09131 / 50654-20
