I mean for the tiny servers that are only listed once every minute or so, maybe tad higher value.
And high values for BIG servers.
That way we drop out of the DNS-cache pretty quickly.
Just an idea…no idea of it actually works.
As I do believe DNS-servers can override it to avoid low TTL’s.
If you want to see the “aa” flag (and the default 130 second TTL) you need to query the authoritative servers of that zone. For example “dig @a.ntpns.org be.pool.ntp.org”. Other than that, the caching you have seen is how DNS works and there’s nothing unusual in how the pool does DNS in this regard. DNS is a textbook example of how things scale to Internet scales, and caching nameservers is an important part of it.
Yes you are right. Did not know that. Thanks.
I’ve added my IPv6 BE server with a 2.0 GBit netspeed again.
Any improvements in the traffic load for IPv6?
I have stopped serving IPv6 from my home for Belgium.
But my other larger servers do that, stratum 2.
I did ask all my servers to be added to Belgium, sadly no response yet.
As I probably set the speed a lot higher when they all are assinged to BE as well.
However, my big server has been limited to 6mbit:
To prevent spikes.
I wonder how much load your IPv6 server is getting now.
I think you better check your own server for load, maybe show graphs?
I have no graphs running on that server. Sorry.
Had the IPv4 side of the server at 1.5GBit until roughly noon yesterday, I believe, to see whether the rate would keep increasing, and for potential comparison with @stevesommars’ server kept at the same rate through Saturday at least. Unfortunately some hiccups on the provider side, as visible in the graph, so not sure how reliable the data is.
The graph for the IPv6 side of the server, which was at 3Gbit throughout. Diurnal pattern visible, and the various hiccups on the provider side.
That is what I see happening here.
IPv4 is fairly consitant in requests. But IPv6 spikes.
I not know why, I presume a lot of GSM traffic, where typical networktraffic is via Fiber/DSL/Cable and being consistant IPv4 by default.
But you set it very high, then you get the max requests. My tiny homeserver is set low, but gets spikes far beyond the setting. Even killing local networkspeed due to this.
I put the MikroTik back, as the Fritzbox wasn’t able to handle it. Wanted to test both, but the slowdowns happened again. NAT-table-overload, no other way to name it.
I wonder. If you just serve The Netherlands, nothing special, just set it the same. Will the same setting give the same requests?
No offense, there are other countries too, but NL is over-serviced. So to see what such speedsettings would do, would reveil a lot (in my opnion).
It would give much data to improve the pool of starving of servers and other countries over-served to share the load.
Ergo, what happens if you do the same to NL servers, if you can, without serving BE too.
Server is a chrony instance which is running on router itself. Don’t have a monitoring service running to collect statistics or create these nice graphs.
chrony serverstats gives over a 3.5 hr time period:
NTP daemon TX timestamps : 3095430
NTP kernel RX timestamps : 3095430
This translates to 246 NTPreq/s
Seems rather low to me…
Could it be that rates are still ramping up?
Current client distribution stats indicate that server IP is part of 2.7% of BE DNS requests but only contributing 5.5% of the bandwidth in BE zone.
For chrony, instead of those “RX/TX timestamps” fields, the “NTP packets received” and “NTP packets dropped” give a better view of the actual amount of traffic.
This is 7 mins worth of data:
NTP packets received : 104768
NTP packets dropped : 184
Command packets received : 354
Command packets dropped : 0
Client log records dropped : 0
NTS-KE connections accepted: 0
NTS-KE connections dropped : 0
Authenticated NTP packets : 0
Interleaved NTP packets : 0
NTP timestamps held : 0
NTP timestamp span : 0
NTP daemon RX timestamps : 0
NTP daemon TX timestamps : 104584
NTP kernel RX timestamps : 104584
NTP kernel TX timestamps : 0
NTP hardware RX timestamps : 0
NTP hardware TX timestamps : 0
That is not a lot.
Look here:
1400req/sec, if you set high speed, it should give high req’s. Like Magic is showing.
Maybe your router can’t handle it? Just guessing.
BTW, I went back to the DrayTek as MikroTik slows down too and RouterOS is driving me nuts 
