View Full Version : [data write: Broken pipe] error

August 14th, 2008, 07:31 AM
I've set up amanda on my centos 5.1 x86_64 box here at the office. We have a single LTO3 tape drive. The file system I am backing up is rather large 1.2TB ext3 file system. There are a few files in it that are over 2GB in size but I would think that is not a problem on a 64 bit system. I am not using the amanda package that came with centos, I compiled amanda 2.6.0p1 myself.

Here is the error I am seeing.

aais1.company.com /u01/smbshares lev 2 FAILED [data write: Broken pipe]
aais1.company.com /u01/smbshares lev 2 was successfully retried

Total Full Incr.
-------- -------- --------
Estimate Time (hrs:min) 0:01
Run Time (hrs:min) 0:36
Dump Time (hrs:min) 0:19 0:00 0:19
Output Size (meg) 13392.1 0.0 13392.1
Original Size (meg) 20238.8 0.0 20238.8
Avg Compressed Size (%) 66.2 -- 66.2 (level:#disks ...)
Filesystems Dumped 2 0 2 (1:1 2:1)
Avg Dump Rate (k/s) 12118.1 -- 12118.1

Tape Time (hrs:min) 0:19 0:00 0:19
Tape Size (meg) 13392.1 0.0 13392.1
Tape Used (%) 3.4 0.0 3.4 (level:#disks ...)
Filesystems Taped 2 0 2 (1:1 2:1)
(level:#chunks ...)
Chunks Taped 2 0 2 (1:1 2:1)
Avg Tp Write Rate (k/s) 12183.7 -- 12183.7

Label Time Size % Nb Nc
DailySet1-008 0:19 13392M 3.4 2 2


/-- aais1.company.com /u01/smbshares lev 2 FAILED [data write: Broken pipe]
sendbackup: start [aais1.company.com:/u01/smbshares level 2]
sendbackup: info BACKUP=/bin/gtar
sendbackup: info RECOVER_CMD=/bin/gzip -dc |/bin/gtar -xpGf - ...
sendbackup: info COMPRESS_SUFFIX=.gz
sendbackup: info end

Thanks in advance for any help you can provide.

August 14th, 2008, 07:39 AM
can you attach the dumper and chunker debug logs?

August 14th, 2008, 08:26 AM
Apparently only one attachment per post here is the amdump log

August 14th, 2008, 08:29 AM
The other log file. Again thanks for taking a look at this for me.

August 14th, 2008, 09:27 AM
Those are actual dump and trace logs, respectively. The debug logs are probably in /tmp/amanda or /var/log/amanda and are named dumper.[0-9]*.log and chunker.[0-9]*.lo

August 14th, 2008, 09:49 AM
Ok let me try again with the correct log files this time.

August 14th, 2008, 09:50 AM
and the dumper files...

August 14th, 2008, 10:20 AM
1218689967.653534: chunker: putresult: 13 RQ-MORE-DISK
1218689967.653841: chunker: getcmd: CONTINUE 01-00003 /var/amanda-holdingdisk/20080814004501/aais1.company.com._u01_smbshares.2 1048576 493120
1218690015.819202: chunker: putresult: 13 RQ-MORE-DISK
1218690015.819425: chunker: getcmd: ABORT
1218690015.819438: chunker: putresult: 14 ABORT-FINISHED

OK, so the dump is running out of holding-disk space, and the dumper is aborting it and re-trying, probably directly to the taper. Your holding-disk space should be at least large enough to hold your biggest dump, and preferably a factor of INPARALLEL larger than that.

August 14th, 2008, 10:41 AM
Hm I'll admit my current holding disk size is qutie small 10gb, however I'm not sure I can make it as big as the file systems I plan to back up. I'll try increasing it and see how it goes. Thanks for the recommendation.