[chrony-users] A question about the output of 'chronyc sources'.

[Mattias Nilsson]

I am curious about what would make chrony (3.2) reject packets (if this is the correct terminology).
In our setup, which is used in a system with two computers and a GPS receiver in a car, the output of 'chronyc sources' on the computer connected to the GPS receiver, includes the following:

MS Name/IP address Stratum Poll Reach LastRx Last sample
===============================================================================
[...]
^? ltl-m-1 0 6 377 - +0ns[ +0ns] +/- 0ns

'ltl-m-1' is the other computer, which has an NTP client having the first computer as its server. The reason for the server entry here is to be able to monitor the offset of the client.

Now, the state is '?', Reach is 377, and LastRx is empty (and the offset is 0).

So, my question is: what could cause 'chronyc sources' to output a line like the one that we are seeing, without an offset?

(We are using chrony as provided by openSUSE, so there may be distribution patches.)

[Miroslav Lichvar]

Post by Mattias Nilsson
MS Name/IP address Stratum Poll Reach LastRx Last sample
===============================================================================
[...]
^? ltl-m-1 0 6 377 - +0ns[ +0ns] +/- 0ns
'ltl-m-1' is the other computer, which has an NTP client having the first computer as its server. The reason for the server entry here is to be able to monitor the offset of the client.

The measurements are probably ignored to avoid a synchronization loop
(two computers synchronized to each other). If this is the case, in
the chronyc ntpdata output the last NTP test will be 0.

There is no option to disable that test. You could, however, configure
one host with an extra IP address and specify it with the
bindacqaddress directive, so the client will use a different source
address than what the other host is synchronized to.

[Mattias Nilsson]

Post by Miroslav Lichvar
There is no option to disable that test. You could, however, configure
one host with an extra IP address and specify it with the
bindacqaddress directive, so the client will use a different source
address than what the other host is synchronized to.

I will try that. Thanks for the quick reply.
(And sorry for any bad formatting of my emails - Outlook is mandatory here at work.)

Mattias

[Mattias Nilsson]

I just want to report that using the bindacqaddress directive solved our issue.

Thanks again,
Mattias