<div dir="ltr"><div><div><div><div>Thanks Bert,<br><br></div><span style="font-family:arial,helvetica,sans-serif">You are 100% correct! I used grepq as you suggested and indeed I have queries which took over 100ms plus the small sample size explains this.</span><span style="font-family:arial,helvetica,sans-serif"> Out of curiosity is there a more elegant way to extract queries using grepq for external requests? I did the following to exclude internal domains:</span><br><span style="font-family:arial,helvetica,sans-serif"></span></div><span style="font-family:arial,helvetica,sans-serif"><br></span></div><span style="font-family:arial,helvetica,sans-serif"> dnsdist -e "grepq('.')" | egrep -v '<internal domain>|<another internal domain>|arpa' </span><br><br><br></div><div>Nigel.<br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On 19 March 2018 at 11:39, bert hubert <span dir="ltr"><<a href="mailto:bert.hubert@powerdns.com" target="_blank">bert.hubert@powerdns.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On Mon, Mar 19, 2018 at 11:30:08AM +0000, Nigel H wrote:<br>
> I'm seeing a discrepancy when i run showServers() in dnsdist. Our internal<br>
> pool of dns servers appears correct and monitoring shows the odd spike in<br>
> latency but our external pool (opendns) seems to just keep increasing in<br>
> value. The documentation states its the latency for this servers in<br>
> milliseconds but when I test using dig I'm seeing normal resolution times<br>
> 1-8ms mostly but dnsdist reports latency like below (27ms).<br>
<br>
</span>Hi Nigel,<br>
<br>
The likely reason for the increase is that opendns has only received 169 and<br>
148 queries so far. This means the weighted average latency still includes<br>
a lot of 0.<br>
<br>
Eventually, you should see the opendns latency plateau.<br>
<br>
With 'grepq()' you should be able to find the queries being sent to opendns,<br>
and how long dnsdist thought they were taking.<br>
<br>
You could then perhaps retry those queries with 'dig' to see how long they<br>
take. 27ms is not unreasonable for mostly cache misses.<br>
<span class=""><br>
> Can someone confirm its meaning? Is it an average latency based an amount<br>
> of queries to this server or is it the highest latency seen? Also does<br>
> dnsdist incorporate the latency of the health check?<br>
<br>
</span>The last part, no. First part, it is average.<br>
<span class="HOEnZb"><font color="#888888"><br>
Bert<br>
</font></span></blockquote></div><br></div>