The 'Illegel seek' is not a problem.
The sendsize is working correctly, the problem is with sendbackup

Could you post the sendbackup and amandad debug files?

Jean-Louis

Chris Cameron wrote:

I have Amanda 2.5.1p3 running on Solaris 8 for the server, and the client is the same version running on OpenBSD 4.0.

Using GNUTAR, the dump times out. The error on the client machine is:

sendsize[3860]: time 0.058: waiting for any estimate child: 1 running

sendsize[28668]: time 0.061: calculating for amname /dev/raid0g, dirname /home, spindle -1 sendsize[28668]: time 0.063: getting size via gnutar for /dev/raid0g level 0 sendsize[28668]: time 0.094: spawning /usr/local/libexec/runtar in pipeline sendsize[28668]: argument list: runtar test /usr/local/bin/gtar --create --file /dev/null --directory /home --one-file-system --listed-incremental /var/amanda/gnutar-lists/carp0_dev_raid0g_0.new --sparse --ignore-failed-read --totals . sendsize[28668]: time 0.337: Total bytes written: 15718400 (15MiB, 124MiB/s)

sendsize[28668]: time 0.342: .....
sendsize[28668]: estimate time for /dev/raid0g level 0: 0.248
sendsize[28668]: estimate size for /dev/raid0g level 0: 15350 KB
sendsize[28668]: time 0.342: waiting for runtar "/dev/raid0g" child
sendsize[28668]: time 0.347: after runtar /dev/raid0g wait
gnutar_calc_estimates: warning - seek failed: Illegal seek

sendsize[28668]: time 0.365: done with amname /dev/raid0g dirname /home spindle -1

sendsize[3860]: time 0.370: child 28668 terminated normally

The tar part seems to work, as I can also run it just at a command line without error. However the message:

warning - seek failed: Illegal seek

Doesn't seem good.


In the report email:

  carp0  /dev/raid0g     lev 0  FAILED [data timeout]

carp0 /dev/raid0g lev 0 FAILED [too many dumper retry: "[request failed: timeout waiting for ACK]"] carp0 /dev/raid0g lev 0 FAILED [cannot read header: got 0 instead of 32768]

When amdump gives up, it leaves the following processes on the client machine:

operator 30460  0.0  0.2   968  2200 ??  Is     9:09AM    0:00.13 amandad

operator 10245 0.0 0.2 816 2240 ?? I 9:10AM 0:00.08 /usr/local/libexec/sendbackup amandad bsd operator 15664 0.0 0.1 824 1528 ?? I 9:10AM 0:00.04 /usr/bin/gzip --fast operator 31389 0.0 0.1 816 1392 ?? I 9:10AM 0:00.02 /usr/local/libexec/sendbackup amandad bsd operator 22560 0.0 0.1 768 576 ?? I 9:10AM 0:00.01 sh -c sh operator 28437 0.0 0.1 432 1192 ?? I 9:10AM 0:00.05 sed -e s/^\\.// operator 18449 0.0 0.1 792 648 ?? I 9:10AM 0:00.03 /usr/local/bin/gtar -tf -

Watching tcpdump I don't see any obvious retrying going on. I have a whack of other Solaris 8 machines that don't have any problem with tar (1.16)

Anyone know what I've done wrong here?


Thanks,
Chris