Followup On Perl Dumping Core

Tim Daneliuk tundra at
Sat Jun 19 14:29:33 UTC 2010

On 6/18/2010 2:45 PM, Tim Daneliuk wrote:
> On 6/18/2010 2:33 PM, Greg Larkin wrote:
>> Tim Daneliuk wrote:
>>> On 6/18/2010 2:09 PM, Matthew Seaman wrote:
>>>> On 18/06/2010 19:56:26, Tim Daneliuk wrote:
>>>>> I should mention that I don't think it is actually "dumping core".
>>>>> It's just reporting the problem in /var/log/messages...
>>>> You'll only get a core file if the current working directory of the
>>>> process is writable by the process.  Normally.  There are various
>>>> sysctls you can use to affect core-dumping:
>>>> kern.corefile: process corefile name format string
>>>> kern.coredump: Enable/Disable coredumps
>>>> kern.sugid_coredump: Enable coredumping set user/group ID processes
>>>> See core(5). It is possible to set kern.corefile to an absolute path --
>>>> eg /tmp/%N.core -- to always record corefiles in a writable directory.
>>>> Also, look at setrlimit values for the maximum size core file permitted.
>>>> 	Cheers,
>>>> 	Matthew
>>> Well ... I've figured out what's causing it, but I still don't know why.  This
>>> is caused when '/usr/local/etc/rc.d/mailscanner restart' is issued from a script
>>> I run to reset the mail system.  However, it does not happen every time ... go
>>> figure.
>> Hi Tim,
>> I apologize if you mentioned this before, but are you using Spamassassin
>> with mailscanner?  This message describes a problem that sounds very
>> similar to yours, and there's a solution included:
>> Hope that helps,
>> Greg
> Aha!  Der plot thickens.  I am indeed running SA. I've just clobbered
> /root/.spamassassin/*   This may well be the issue ...
> Many thanks.

It would seem that this was the problem.  Clearing out the old contents
of .spamassassin made the perl SEGVs stop...

Tim Daneliuk     tundra at
PGP Key:

More information about the freebsd-questions mailing list