Hello,

* David Trusty (dwtrusty AT hotmail DOT com) [20030105 13:25] thus spake:
> I have more information on this problem.
> 
> I see these messages in the 'amdump' file:
>    dumper: logfile.c@202: memory allocation failed (28 bytes requested)
>    dumper: debug.c@345: memory allocation failed (22 bytes requested)

I see this once in a while. I'm running Amanda-2.4.3b4-20020829 on 
a SGI Origin3800 IRIX6 with a STK-L40 and 4 LTOs Ultrium and using
chg-zd-mtx with mtx version 1.2.16rel.

INFO taper tape 000037 kb 104124640 fm 9 writing file: No space left
on device
START taper datestamp 20021225 label 000038 tape 1
INFO taper retrying bullcalf:/data/ai.0 on new tape: [writing file: No
space left on device]
FATAL dumper token.c@201: memory allocation failed (26 bytes requested)
WARNING driver dumper0 pid 11096267 is messed up, ignoring it.
INFO driver bullcalf /data/ai 0 [dump to tape failed, will try again]
SUCCESS dumper bullcalf /data/ai 20021225 0 [sec 7302.267 kb 93604920
kps 12818.6 orig-kb 93604920]
SUCCESS taper bullcalf /data/ai 20021225 0 [sec 7302.416 kb 93604960
kps 12818.4
 {wr: writers 2925155 rdwait 1644.424 wrwait 5524.192 filemark 1.813}]
WARNING driver dumper0 exited with signal 6

Question: what is your runtapes in amanda.conf?
Mine is set to 10. This happens only when a amanda
runs out of tape and start over with a new tape.
I will try to investigate this a little further.

regards,
jf

> 
> The 'debug.c' message is repeated many times.
> 
> I ran a periodic 'top' during last night's backup, and I found
> that the dumper process got into a loop and quickly used up all
> the memory.  Here is a report just before it died:
> 
>   PID USER     PRI  NI  SIZE  RSS SHARE STAT %CPU %MEM   TIME COMMAND
> 23204 amanda    25   0 2579M 2.5G   808 R    92.7 74.5   8:35 dumper
> 
> The start of the looping appears to have coincided with a tape
> change.  Note that I use chg-zd-mtx and ammt.
> 
> Any ideas?
> 
> Thanks!
> 
> David
> 

-- 
Seeing things just rest a while
As the tide sweeps out another mile