[pkg-ntp-maintainers] Bug#498992: Bug#498992: kernel time sync status change and log levels
tim at buttersideup.com
Mon Sep 21 09:30:40 UTC 2009
Kurt Roeckx wrote:
> On none of the server I run I see the "status change" messages,
> except for one about 5 minutes after booting. At any other time
> a change of the status for me is actually an indication that
> something is wrong. I expect it to always be in PLL mode (0001).
> I guess it comes down to how much you care about your clock,
> and expect that people using the default pool servers see
> those errors more. Looking at some of those pools servers,
> their clocks are regarly off in the range of 100 ms, and half
> an hour later -100 ms. And I expect ntp to have a problem
> with such servers.
. I get these messages in the default Debian configuration (i.e. , even
on a good quality low jitter/low-latency network connection), this may
be due to pool servers, as you say.
. Upstream consider these messages to be "normal, but significant" -
they do not indicate a fault.
. Currently the logcheck rules are set to ignore "kernel time sync
disabled messages", but not these mode-change messages!
. I would imagine that even with perfect ntp servers, users with jittery
or congested network connections will frequently see these messages, and
they should be considered normal behaviour.
. If the majority of users should care about these messages, then they
should instead be changed to something a lot more clear (i.e. the user
doesn't need to have knowledge of PLLs and refer to the ntp source to
understand what they might mean - let alone whether the condition may
indicate a problem).
. I am not asking that these messages should not be logged, only that
they should not be emailed to me by default....
More information about the pkg-ntp-maintainers