[Pdns-users] stuck tcp sessions on recursor

[Charles Sprickman]

On Thu, 24 Feb 2011, Kenneth Marshall wrote:

> On Thu, Feb 24, 2011 at 01:20:13PM -0500, Charles Sprickman wrote:
>> Howdy,
>>
>> I'm seeing an issue with tcp queries on powerdns recursor 3.2.  We run two
>> instances of pdns recursor on an internal network and it's mainly hit by
>> lots of qmail delivery servers.  They are doing a ton of lookups, I think
>> we peak around 4000 queries/second.  With more people using dnssec, many
>> providers returning insanely huge lists of mxers, and qmail's propensity
>> for "any" queries, we're finding that our tcp queries are going way up.
>>
>> What we are seeing is that when "max-tcp-clients" is exceeded, it appears
>> that pdns is not properly closing sockets.  We then get a pile-up of
>> qmail-remotes that are "stuck" and deliveries stop until pdns is restarted.
>>  netstat shows the connections in state "CLOSED".
>>
>> We're running FreeBSD 8.1 on all the hosts involved.  This mailing list
>> post is interesting, and explains why these "CLOSED" sockets are in that
>> state:
>>
>> http://lists.freebsd.org/pipermail/freebsd-net/2008-June/018544.html
>>
>> Any pointers on how to troubleshoot this further?
>>
>> Thanks,
>>
>> Charles
>
> There is not much to troubleshoot. This is a known bug. Try running
> the latest release instead. They really do fix bugs. :)

Ah.  I read this a few weeks ago, but wasn't sure if it was what I was 
looking for:

"Simon Bedford, Brad Dameron and Laurient Papier discovered relatively 
high TCP/IP loads could cause TCP/IP service to shut down over time. 
Addressed in commits 1546, 1640, 1652, 1685, 1698. Additional information 
provided by Zwane Mwaikambo, Nicholas Miell and Jeff Roberson. Testing by 
Christian Hofstaedtler and Michael Renner"

I was scanning for "stuck in CLOSED" and similar stuff.  I'll try 3.3 and 
see how that goes, thanks!

Charles

> Cheers,
> Ken
>