Re: Frequency of requests to a source whether it is available again

On Mon, Jul 20, 2026 at 03:11:03PM +0000, Kaelin Philippe SBB CFF FFS wrote:
> After that we closed all NTS and NTP ports and from 15:00 UTC July 17 chrony recognised this relatively quickly. The query whether a source is available again increases exponentially (see log).
> We want the fastest possible response time, even after an outage of 2 days. I reopened the ports today 20 July and waited for over an hour, but without success.
> Is there an option to shorten or deactivate this so that he queries the sources at regular intervals (e.g. 5 minutes)?

No, the maximum NTS-KE retry interval is hardcoded to 2**19 seconds
(about 6 days). The idea was that NTS-KE is orders of magnitude more
expensive for the server to handle than NTP, so clients should try
hard to not overload it, assuming there could be millions of them
doing the same.

To get around that you could have a script running "chronyc reset
sources" periodically when the sources are not reachable.

The NTS-KE polling interval could be configurable, e.g. add minntspoll
and maxntspoll options. There could also be an exception for the case
when not even the TCP connection can be established, e.g. due to
the server not running or at worst SYN-ACK getting lost on the way
back to client, which for the server is quite cheap, so I think
the client could try that much more frequently.

-- 
Miroslav Lichvar


-- 
To unsubscribe email chrony-users-request-kWFZVVI9zxvPqho9SqqRMmD2FQJk+8+b@public.gmane.org 
with "unsubscribe" in the subject.
For help email chrony-users-request-kWFZVVI9zxvPqho9SqqRMmD2FQJk+8+b@public.gmane.org 
with "help" in the subject.
Trouble?  Email [email protected]