[dnsdist] question about latency stats

[Joel Carnat]

Oh, ok. I was rather referring to those ones https://www.dnsdist.org/statistics.html#latency0-1 but indeed, the Web page states it is for UDP only; whereas the Prometheus output doesn’t. There doesn’t seem to be those metrics for TCP based queries. And the histogram ones seem to be about average latency from last packets rather than number of packets in that latency range.

I guess this means I have to turn my graphs in another way.

Thanks!


> Le 11 avr. 2025 à 19:47, Otto Moerbeek <otto at drijf.net> a écrit :
> 
> On Fri, Apr 11, 2025 at 07:23:24PM +0200, Joel Carnat via dnsdist wrote:
> 
>> Hi,
>> 
>> I'm new to DNSdist, have configured it and pointed a bunch of servers
>> and workstation toward it. Now, I'm looking at statistics; using the
>> Carbon export to a VictoriaMetrics database.
>> 
>> There is a strange thing regarding latency values. The sum of those, is
>> not equal to the queries statistics:
>> dnsdist_queries 28699
>> dnsdist_latency0_1 2553
>> dnsdist_latency1_10 433
>> dnsdist_latency10_50 899
>> dnsdist_latency50_100 271
>> dnsdist_latency100_1000 758
>> dnsdist_latency_slow 67
>> 
>> Having a look at this[1] dashboard, I thought the sum of all latency
>> values would be equal to the queries one.
>> 
>> What did I misunderstood?
> 
> According to https://www.dnsdist.org/statistics.html#latency-avg100
> this histogram only covers UDP queries. There are a few other
> histograms that conver other types of queries.
> 
>    -Otto
>> 
>> Not sure it that matters, but queries include UDP, TCP, DoT and DoH
>> queries. And that number seem correct.
>> 
>> Thank you,
>> Joel C.
>> 
>> [1] https://grafana.com/grafana/dashboards/18044-dnsdist/
>> _______________________________________________
>> dnsdist mailing list
>> dnsdist at mailman.powerdns.com
>> https://mailman.powerdns.com/mailman/listinfo/dnsdist
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.powerdns.com/pipermail/dnsdist/attachments/20250411/d5e9a809/attachment.htm>