On 2025-08-27 15:57:01 [+0200], Miroslav Lichvar wrote:
> On Tue, Aug 26, 2025 at 02:59:12PM +0200, Sebastian Andrzej Siewior wrote:
> > The benchmark is about > 1k packets/ second while in reality you have
> > less than 20 packets a second.
> 
> I don't want to argue about which use case is more important, but it's
> normal for NTP servers to receive requests at much higher rates than
> that. In some countries, public servers get hundreds of thousands of
> packets per second. A server in a local network may have clients
> polling 128 times per second each.

There might be a misunderstanding here. You can't receive 1k packets a
second and each one with a HW timestamp for PTP. This does not work.
SW timestamps more likely.

> Anyway, if anyone is still interested in finding out the cause of
> the regression, there is a thing I forgot to mention for the
> reproducer using ntpperf. chronyd needs to be configured with a larger
> clientloglimit (e.g. clientloglimit 100000000), otherwise it won't be
> able to respond to the large number of clients in interleaved mode
> with a HW TX timestamp. The chronyc serverstats report would show
> that. It should look like the outputs I posted here before.

How does this work with HW timestamps vs SW? I can't believe that 1k
packets are sent and all of them receive a HW timestamp.

Sebastian

Reply via email to