Hi,

forgot not all of you are software developers, so added some explanation
of how to extract a useable backtrace from a running process with gdb.

Am 15.10.2008 8:27 Uhr, Felix Buenemann schrieb:
> You could try running a debug build of spamdyke (unstripped binary,
> remote the strip command from Makefile) and attach gdb or strace to a
> process (id) that is eating up all cpu. That should help Sam to find
> what the problem is.

Running gdb to trace the problem would look like this:
# gdb /path/to/spamdyke <process_id_of_cpumonster>
(gdb) break (to interrupt execution, if it doesn't do so automatically)
(gdb) bt
(gdb) disass $eip-64 $eip+64
(gdb) info all-registers
(gdb) kill (or 'c' to keep the process running)
(gdb) q
Then copy the output and mail it (to Sam only, shpuld be not so
interesting for the list ;-)

-- Felix

> Note that debug builds have a larger memory footprint, so you might have
> to reduce the maximum number of spamdyke processes you can run (only
> matters if you're already maxing out your mem).
> 
> -- Felix
> 
> Am 14.10.2008 18:50 Uhr, Erald Troja schrieb:
>> Hello,
>>
>> I second your findings.
>>
>> We reverted to 4.0.4 right away.
>> Did not report it as we were unable
>> to find a good explanation for it.
>>
>> The spamdyke processes were just lingering each consuming
>> between 70% to 100% of CPU.

>> ------------------------
>> Erald Troja
>>
>>
>> Paulo Henrique wrote:
>>> Hi,
>>> since the spamdyke upgraded to 4.0.5, I noted that my servers working
>>> with a high load,  the average of 0.65 and they were left to 3.5,
>>> someone noticed this problem? What may be happening?
>>>
>>> tks


_______________________________________________
spamdyke-users mailing list
[email protected]
http://www.spamdyke.org/mailman/listinfo/spamdyke-users

Reply via email to