[Pdns-users] Strange pdns-recursor crash

[John Miller]

On 9/3/12, Peter van Dijk <peter.van.dijk at netherlabs.nl> wrote:
> Hello John,
>
> On Sep 1, 2012, at 0:00 , John Miller wrote:
>
>> I've attached statistics logs from the day of the crash; we're seeing
>> somewhere between 100 and 300 qps on ns1 and 25 to 100 qps on ns2.  Any
>> thoughts on what would cause two nameservers to lock up simultaneously?
>> Anyone experienced anything similar?
>
>
> This does not sound familiar to us. Usually, unexplained crashes are related
> to out of memory situations, but your caches did not seem to be the same
> size at the point of crash. However, if you can keep an eye on memory use
> (perhaps with some graphs), that might provide some clues for the future.
>

Thanks for the reply, Peter.  I have sar stats for the times right
around the crash, but it didn't look like memory was the issue
here--the machines themselves didn't lock up, only the pdns-recursor
daemons.  Today's a holiday here in the US, so I'll post them when I'm
back at work tomorrow.

I'll also set up some identical servers in a sandbox environment, then
see if I can replay our traffic to them (or some other networking
voodoo) to see if I can repeat this.

To everyone on the list--I'm still a relative newbie to PowerDNS
Recursor; what parameters are you using in terms of max-cache-size,
max-packetcache-size, max-packetcache-ttl, ringbuffer-size etc.?  What
are you seeing in terms of queries/second?

John