[SOLVED] Re: ntpd never achieves synchronized, even with PPS source; ntpdig KOD

[Robin H. Johnson]

On Tue, Dec 18, 2018 at 02:16:16PM -0800, Hal Murray wrote:
> > --enable-lockclock wasn't set during compile, but this DID lead me to making
> > a useful discovery: refclock_local was enabled as a compile-time option,
> > probably from when I was ensuring all Gentoo-user-accessible code paths
> > compiled. 
> 
> There is a script that tries most of the configure time options.
>   tests/option-tester.sh
> There is another one that tries building with python3
>   tests/python3-tester.sh
> Both include minimal testing - run enough to print out the version string.  
> That mostly checks that it can find the libraries (pre install)
This is mostly in terms of mapping packages to Gentoo's ebuilds and
ensuring dependencies are correct:
- spin up a test environment with nothing installed
- add the declared dependencies
- test building & possibly running with the kitchen sink turned on

> > - BUG: figure out why refclock_local being simply compiled-in and not in
> > the config at all prevented sync.
> I'm sure we can do something in that area.
> 
> > - MAYBE: log if the clock would have been stepped but isn't because $reasons
> I don't think we can (easily) do that, but it's worth investigating.  We 
> should also log long slews.  That may be tricky to do without adding too much 
> clutter.
> 
> > - DEBUGGING: during startup AND version output, log what refclocks are comiled-in
> We'll have to think about that.  It would break some stuff and/or add a lot of 
> clutter.
> How about another command line option?
Was thinking of log entries like: 
'INIT: compiled refclocks: ...'
'INIT: configured refclocks: ...'

I did find it in the 'ntpd -v' output as well:
| This version was compiled with the following clock drivers:
|      GENERIC        NMEA         PPS         SHM        GPSD
(very generous with whitespace I see)

I'd add it to ntpd -V output.

> > - DEBUGGING: log slewing changes
> That's in loopstats.
I mean when timing changes are applied to the kernel.

step_systime emits a WARN on changing time, but adj_systime emits
NOTHING unless there is an error. I'd add a DEBUG-level output after the
change, identical to the step_systime output.

-- 
Robin Hugh Johnson
Gentoo Linux: Dev, Infra Lead, Foundation Treasurer
E-Mail   : robbat2 at gentoo.org
GnuPG FP : 11ACBA4F 4778E3F6 E4EDF38E B27B944E 34884E85
GnuPG FP : 7D0B3CEB E9B85B1F 825BCECF EE05E6F6 A48F6136
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 1113 bytes
Desc: Digital signature
URL: <https://lists.ntpsec.org/pipermail/devel/attachments/20181218/dcb9e9e9/attachment.bin>