On 03/29/2012 09:17 AM, Markus Lauterbach wrote:
Hi there,
during my daily backup I recieve the error-message for quite a few days.
These dumps were to tape AKREG1-08.
*** A TAPE ERROR OCCURRED: [runtapes volumes already written].
There are 506710M of dumps left in the holding disk.
Run amflush to flush them to tape.
Beside this error, I found the following lines in the corresponding
taper.log (taper.20120328210001.debug)
1332980895.993331: taper: warning: Got EIO on /dev/nst3, assuming end
of tape
What is before that line, what amanda was trying to do when it got EIO?
Is it when it wrote the tape label? or did it wrote a lot of data before
the error? Did it hit the end of tape?
1332980895.993377: taper: Device /dev/nst3 error = 'No space left on
device'
1332980895.993441: taper: Device /dev/nst3 setting status flag(s):
DEVICE_STATUS_VOLUME_ERROR
1332980903.652733: taper: device /dev/nst3 ran out of space
1332981061.192603: taper: putresult: 4 PARTIAL
1332981061.192955: taper: getcmd: FILE-WRITE 01-00094
/data1/amanda/daily1/20120328210001/server.domain.de._data1_BACKUP.0
server.domain.de /data1/BACKUP 0 20120328210001 0
1332981061.193110: taper: putresult: 26 REQUEST-NEW-TAPE
1332981061.193234: taper: getcmd: NO-NEW-TAPE "runtapes volumes
already written"
1332981063.305514: taper: putresult: 10 FAILED
1332984143.214650: taper: getcmd: QUIT
1332984143.214684: taper: putresult: 2 QUITTING
At first, I expected an dirty tape device
(http://wiki.zmanda.com/index.php/No_space_left_on_device). But after
cleaning the device, it even doesn't work poperly.
I'm using three LTO4 and one LTO5 device. The LTO4 drives are from the
same typ (TANDBERG LTO-4 HH, same firmware). The LTO4 tapes are
connected to two Adaptec 29320LPE PCIe Ultra320 SCSI adapter.
For all three LTO4 devices I'm using the same device-definition in
amanda.conf
define tapetype tandberg-lto4hh {
comment "Tandberg LTO-4 HH; compression disabled"
length 794001824 kbytes
filemark 0 kbytes
speed 50554 kps
blocksize 32 kbytes
}
But only this one device is causing problems. Then I expected a
hardwareerror. So I downloaded the Tandberg test software. I was able
to write and read a hugh amount of random data several times on the
device.
Selected Target: 3 - TANDBERG LTO-4 HH W619 SCSI
>>> Tape IO Test Start
Param: WR, 64, Random Data
Checking block size info...Completed
Setting block size info...Completed
Get current compression status - Enabled
Disabling compression...Completed
Rewinding...Completed
Resetting IO log parameters...Completed
Writing logical beginning of tape...Completed
Writing 750.000 MB data to the tape...Completed
Retrieving log status parameters...Completed
Write Transfer Rate = 4527 MB/min
Rewinding...Completed
Locating the first data...Completed
Reading 750.000 MB data from the tape...Completed
Retrieving log status parameters...Completed
Read Transfer Rate = 4529 MB/min
Rewinding...Completed
Erasing Tape...Completed
>>> Tape IO Test Passed
Writing directly to the tape using 'tar' is also posible and doesn't
cause any error. Just amanda keeps on telling me, that the device
reports 'EIO' and won't write the backupdata to tape. Running amflush
leads to the same problem.
On the serverside (RHEL 6.2), I use the following repository based
versions:
- amanda-server-2.6.1p2-7.el6.x86_64
- amanda-client-2.6.1p2-7.el6.x86_64
- amanda-2.6.1p2-7.el6.x86_64
Can anyone give me a short hint, where I should take a look to?
Thanks in advance
Markus