Queries after removing a server from the pool


#1

For those who have removed a server from the pool, how long has it been until NTP queries to your server stopped?

One of the servers I recently removed (in the AU zone) is still getting approx 40 queries per second almost a month after being removed from the pool.

I haven’t looked at it closely or graphed out how long the tail is, however clearly there are a lot of NTP clients who hang on for a long time :slight_smile: :dark_sunglasses:


#2

How long was it in the pool and what was the “net speed” setting? The clients you still have would be ntp daemons that don’t lookup the DNS name past the initial lookup.


#3

In the pool for approx 1 year, “Net Speed” set to 50Mbps.

Yeah I gather it’s from NTP daemons that aren’t doing a DNS lookup past the initial one.

It seems there are a lot of long running NTP daemons :slight_smile:


#4

Yeah, that’s definitely more than I’d have expected. Maybe if Australia is ~10x underserved vs some of the large European or North American zones that’d translate to just a handful of requests per second elsewhere by now.

If you (or someone else) have “spare IPs” and time to do the tests and analytics then it’d be interesting to add a list of IPs to various zones at a particular net speed, monitor usage while it’s in the zone and then after say 3 months disable the IPs in the system and see how long it takes to shed most of the clients. (We can help “rig” the server configuration in the pool to pretend to be in the appropriate zones for the test).


#5

Yes I’m happy to donate some IP’s / VMs / time to do some testing and analytics.

I’ll get some resources sorted over the next few weeks or so and get back to you!