[ntp:questions] Start of new GPS 1024 week epoch
magnus at rubidium.dyndns.org
Thu Aug 15 03:02:14 UTC 2013
On 08/14/2013 06:21 PM, David Taylor wrote:
> On 14/08/2013 17:14, David Malone wrote:
>> Possibly - this was a handy timestamp to use, but there may be a
>> better choice. It's also possible that this should be an optional
>> flag to the driver, so that people with perfectly good GPS units
>> won't ever trip over the code ;-)
> Yes, I would prefer to see a flag which enables this, with the default
> being "off". All being well, the number of NTP installations with
> "good" receivers should be increasing over the number with "problem"
This gives you the benefit of disrupted operations as they fail with the
flag typically being off, they will have to search the net to find this
thread to figure out what went wrong. Do consider that they will fail
typically as UTC Saturday turns to UTC Sunday, so it will typically
occur during the weekend and leave a nice little disaster for Monday
morning, but then again we do want to contribute to the Monday morning
feeling, right? The case when it is not typically occurring on UTC
Sunday, is when the receiver is woken up from it's beauty sleep on a
weekday, which most probably is manual labor involved and thus direct
attention to the problem.
An alternative approch is to have this default on, but allowing security
nerds to turn it off.
Regardless, a log event should occur to indicate that
1) 1024 GPS week wrapping was detected, and not adjusted for (to fix,
enable option foo-bar-gpsweekwrapping enable)
2) 1024 GPS week wrapping was detected and corrected for (to disable use
option foo-bar-gpsweekwrapping disable)
depending on wither the option was enabled or disabled.
I clearly would have default enabled, as that would cause least
operational concerns in the typical installations. There is so many
fiddly options to consider and this one can be easily missed if you are
not a ntp-pro.
More information about the questions