[ntp:questions] NTP 4.2.6 Released
martin.burnicki at meinberg.de
Thu Dec 17 13:53:36 UTC 2009
Dave Hart wrote:
> On Dec 15, 10:25 UTC, Martin Burnicki wrote:
>> 4.2.6 contains a real bunch of changes against 4.2.4p8 which have not
>> been mentioned here, but significantly change the way ntpd works, so
>> users should carefully check if they want to upgrade to 4.2.6 right now.
> That sizable message spells out the three years of -dev changes
> between 4.2.4 and 4.2.6 in terms of file and changeset comments.
> There was no corresponding bk-ntp-stable-send at lists.ntp.org message,
> which would have been truly gargantuan as it would have included all
> the diffs. Harlan experienced a bk hiccup during the process that I
> suspect was the failure of the attempted generation of that monster-
> diff message.
> A more succinct copy of the most of same information can be seen in
> the ChangeLog file in the distribution, though it's maintained by hand
> and incomplete by comparison. Since this is a unified file for both -
> dev and -stable branches, new entries appear in two places, at the top
> for -dev and somewhere in the middle (ahead of the most recent prior -
> stable entries).
What I've disliked is that the announcement of 4.2.6 emphasizes the latest
security fix only, which is also in 4.2.4p8.
There should be a simple summary of basic changes in 4.2.6 compared to
4.2.4, which are important for the default user, e.g. (from memory):
- the default logging mode does not include messages when the system time
has been stepped, or which source is selected as sys peer
- even if logging is explicitely configured to include this information, the
resulting messages are not really human-readable
- leap second handling has been changed such that a leap second announcement
is valid is accepted only if a majority of the upstream sources provide the
- evaluation of a leap second file does not depend anymore on having autokey
enabled, and IIRC the default location of the file has changed
- IIRC there have been changes to autokey which may prevent autokey
interactions of 4.2.6 nodes with 4.2.4 nodes
More information about the questions