[Dnsmasq-discuss] Eviction statistics is always zero

Simon Kelley simon at thekelleys.org.uk
Mon Apr 12 16:48:14 UTC 2021


It's not absolute, since they may have been caught at the end of cache
lifetime upstream, but the Time To Live values of one second and five
seconds in those records suggests it's going to be pretty hard for them
to hang around in the dnsmasq cache long enough to be evicted.


Simon.


On 12/04/2021 15:03, Leonardo da Mata wrote:
> Hello, thanks for the answer.
> 
> The queries I sent are for AWS rds resolution names, I repeated the
> experiments with a sample of simple urls and cache size of 2 and could
> see the eviction metric.
> 
> I think there might be something on the AWS rds that might prevent it
> from being cached (Some data is redacted)
> 
>   dig prod-authdb.cluster-ro-h2dic5akfled.us-west-2.rds.amazonaws.com
> <http://prod-authdb.cluster-ro-h2dic5akfled.us-west-2.rds.amazonaws.com>
> @172.0.0.1 <http://172.0.0.1>
> 
> ; <<>> DiG 9.11.4-P2-RedHat-9.11.4-26.P2.amzn2.4 <<>>
> prod-authdb.cluster-ro-h2dic5akfled.us-west-2.rds.amazonaws.com
> <http://prod-authdb.cluster-ro-h2dic5akfled.us-west-2.rds.amazonaws.com>
> @172.0.0.1 <http://172.0.0.1>
> ;; global options: +cmd
> ;; Got answer:
> ;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 60600
> ;; flags: qr rd ra; QUERY: 1, ANSWER: 2, AUTHORITY: 0, ADDITIONAL: 1
> 
> ;; OPT PSEUDOSECTION:
> ; EDNS: version: 0, flags:; udp: 4096
> ;; QUESTION SECTION:
> ;prod-authdb.cluster-ro-h2dic5akfled.us-west-2.rds.amazonaws.com
> <http://prod-authdb.cluster-ro-h2dic5akfled.us-west-2.rds.amazonaws.com>. IN
>    A
> 
> ;; ANSWER SECTION:
> prod-authdb.cluster-ro-h2dic5akfled.us-west-2.rds.amazonaws.com
> <http://prod-authdb.cluster-ro-h2dic5akfled.us-west-2.rds.amazonaws.com>. 1
> IN CNAME     prod-authdb-1.h2dic5akfled.us-west-2.rds.amazonaws.com
> <http://prod-authdb-1.h2dic5akfled.us-west-2.rds.amazonaws.com>.
> prod-authdb-1.h2dic5akfled.us-west-2.rds.amazonaws.com
> <http://prod-authdb-1.h2dic5akfled.us-west-2.rds.amazonaws.com>. 5 IN A
> 10.0.4.23
> 
> ;; Query time: 49 msec
> ;; SERVER: 172.0.0.1#53(172.0.0.1)
> ;; WHEN: Mon Apr 12 12:55:08 UTC 2021
> ;; MSG SIZE  rcvd: 157
> 
> 
> Is CNAME not cached? 
> 
> On Fri, Apr 9, 2021 at 10:46 PM Geert Stappers <stappers at stappers.nl
> <mailto:stappers at stappers.nl>> wrote:
> 
>     On Fri, Apr 09, 2021 at 10:39:41AM +0100, Leonardo da Mata wrote:
>     > On Thu, Apr 8, 2021 at 10:50 PM Simon Kelley wrote:
>     > > On 08/04/2021 10:37, Leonardo da Mata wrote:
>     > > > Hello, recently I've done some experiments with dnsmasq
>     reaching 4k qps
>     > > > and, even with a very small cache size of 100 entries, I can't
>     see the
>     > > > evictions stats to increase. It always stays at zero.
>     > > >
>     > > > I'm wondering if anyone can see this data on their side and if
>     there
>     > > > might be any problem with the statistics count or something
>     that might
>     > > > interfere with this data.
>     > > >
>     > >
>     > > Thu Apr  8 21:33:08 2021 daemon.info <http://daemon.info>
>     dnsmasq[23124]: cache size 600,
>     > > 58/15779 cache insertions re-used unexpired cache entries.
>     > >
>     > >
>     > > What are you using as your test query set?
>     > >
>     > >
>     >
>     > I'm using a sample of queries from my services, but I'm introducing an
>     > artificial qps with flame using this sample.
>     >
> 
>     FWIW "this sample" did not reach my end of the Internet.
> 
> 
>     Please recieve
>     > > What are you using as your test query set?
>     as
>     } } Send your test query set, so we can reproduce the observation.
> 
> 
> 
>     Groeten
>     Geert Stappers
>     -- 
>     Silence is hard to parse
> 
> 
> 
> -- 
> Leonardo da Mata
> <http://goog_881972804/>
> leonardo.mata at samsara.com <mailto:leonardo.mata at samsara.com>
> 
> _______________________________________________
> Dnsmasq-discuss mailing list
> Dnsmasq-discuss at lists.thekelleys.org.uk
> https://lists.thekelleys.org.uk/cgi-bin/mailman/listinfo/dnsmasq-discuss
> 




More information about the Dnsmasq-discuss mailing list