Re: [Pdns-users] Efficient query logging

2018-04-04 Thread Remi Gacogne
Hi, On 04/04/2018 03:52 PM, Brian Candler wrote: > One question.  Protobuf messages are in general not self-delimiting[^1], > but I couldn't find any description of how PDNS delimits protobuf > messages.  Are they wrapped?  Or does tag 1 mark the start of each > message (which I note is the only r

Re: [Pdns-users] Efficient query logging

2018-04-04 Thread Brian Candler
On 03/04/2018 22:39, Brian Candler wrote: Aha, "powerdns recursor protobuf export" was the search term I needed, thanks! https://doc.powerdns.com/recursor/lua-config/protobuf.html One question.  Protobuf messages are in general not self-delimiting[^1], but I couldn't find any description of ho

Re: [Pdns-users] Efficient query logging

2018-04-03 Thread Brian Candler
On 03/04/2018 22:28, Remi Gacogne wrote: Yes, in my humble opinion using our protobuf export feature is the way to go, at least that's the kind of goal we had in mind when we designed it. Note that it's not really Lua-related in the recursor, it only happens to be configurable via the Lua configu

Re: [Pdns-users] Efficient query logging

2018-04-03 Thread Remi Gacogne
Hi Brian, On 04/02/2018 12:06 PM, Brian Candler wrote: > I therefore wonder what approaches other people have taken to this > problem.  Is it possible to do this efficiently within pdns itself, e.g. > using LUA [^1]?  Should I put dnsdist in front [^2]? Yes, in my humble opinion using our protob

Re: [Pdns-users] Efficient query logging

2018-04-02 Thread Chris Stradtman
It's not powerdns specific, but we have been using packetbeat for that sort of work. Chris Stradtman On Mon, Apr 2, 2018 at 6:06 AM, Brian Candler wrote: > I'm investigating how to monitor DNS queries as a source of security > information for breach detection. In the case of client machines, w

[Pdns-users] Efficient query logging

2018-04-02 Thread Brian Candler
I'm investigating how to monitor DNS queries as a source of security information for breach detection.  In the case of client machines, we can check the queries against a blacklist of known C&C or malware domains; in the case of servers, we know they should only be making outbound connections t