[PATCH 0/5] Fix ntpd to not poll frequently

Miroslav Lichvar mlichvar at redhat.com
Thu Sep 25 13:46:48 UTC 2014


On Wed, Sep 24, 2014 at 10:30:22PM +0200, Denys Vlasenko wrote:
> On Wednesday 24 September 2014 19:24, Isaac Dunham wrote:
> > > I am not sure any of these conditions warrant increasing poll interval.
> > > 
> > > Can you explain why you think it should be done?
> > 
> > #3 at least would seem to be a fairly good justification to me.
> 
> Yes, fixing this "RATE" TODO item would be reasonable:
> 
> // TODO: stratum 0 responses may have commands in 32-bit m_refid field:
> // "DENY", "RSTR" - peer does not like us at all
> // "RATE" - peer is overloaded, reduce polling freq
>                 bb_error_msg("reply from %s: peer is unsynced", p->p_dotted);
>                 goto pick_normal_interval;
>         }

For completeness it would be nice to have that implemented, but it's
not that useful. Most public NTP servers have rate limiting disabled
completely, becase there are clients broken so horribly that trying to
limit their polling rate actually increases the NTP traffic.

They send a new request seconds after missing a reply or send a new
request immediately when a KoD reply (which looks like an invalid
reply) is received. An example of this was the systemd SNTP client
(timesyncd).

I'm not very surprised when some people leave the pool.ntp.org
project. It's sad when you know most of your resources are wasted for
few broken NTP implementations and there is nothing you can do about
it.

-- 
Miroslav Lichvar


More information about the busybox mailing list