[Pdns-users] PDNS Crash

[Anthony Eden]

As previously described I am currently using PowerDNS for the root name 
servers for the .mp domain.  I have to backends, first bind and then 
pipe.  The pipe is to a perl script which is essentially delegating the 
entire domain to other name servers.

PDNS recently crashed with the following log:

Nov 25 05:38:06 mp1 pdns[8664]: 5002 questions waiting for database 
attention. Limit is 5000, respawning
Nov 25 05:38:06 mp1 free(): invalid pointer 0x83e1650!
Nov 25 05:38:06 mp1 pdns[8664]: Got a signal 11, attempting to print trace:

Any idea what is causing this and what I can do to fix it?  This is the 
second time this has occurred on our primary server in the last week. 
There was no trace in the log after the "attempting to print trace:" line.

I suppose I could increase the max-queue-length setting but I worry that 
this will just mask the real problem.  Is it possible that the Perl 
backend is just unable to handle the load or is crashing itself?  I have 
the TTL for the resulting delegation record from the perl script at only 
3600 so perhaps I should up that as well to reduce load?

Also, I noticed that the pipe backend supports a pipe-timeout 
configuration parameter...what is the default value for this if it isn't 
specified?  Perhaps I need to set that to a specific value?

Sincerely,
Anthony Eden