[ntp:questions] "pool" command operation in ntpd 4.2.6?
stenn at ntp.org
Tue May 26 06:20:32 UTC 2015
Nathan Stratton Treadway writes:
> I've been looking over the documentation for the "pool" command, and
> noticed that the descriptions for the "Server Pool Scheme" changed quite
> a bit between v4.2.6 (
> ) and the "current" docs (
> Looking through the ChangeLog file, I see the entry "New "pool"
> directive implementation modeled on manycastclient." mentioned for
> v4.2.7p22 (but other than that the only pool-related changes mentioned
> seem to be fairly minor).
> So, I just wanted to be sure I was correct in concluding that in back in
> NTP v4.2.6 the "pool" command will pull in multiple servers (assuming
> the DNS query on the host name returns than one address), but otherwise
> it behaves similar to a handful of "server" lines pointing to the same
> host name?
I don't remember what 4.2.6 did. Perhaps somebody else does.
> That is, in order to get repeated addition of new servers until the
> "maxclock" count is reached and automatic discarding/replacing of
> servers that have stopped responding (and the other helpful features
> previously discussed here on this list), I would need to be running the
> manycastclient-type pool implementation found in v4.2.7p22 or later?
The 'pool' directive in 4.2.7/4.2.8 will automatically spin up
associations as needed, including throwing away "bad" servers and
adding "new" servers as appropriate.
There is still work to do in this area, and there should be one or two
bug reports on this. I think we add one new server per poll interval,
and I think we should ask for a "whack" of servers at each poll
interval. I started to work on this issue and more important things
have pulled me away from it. It's not an easy problem to fix, but if
anybody else wants to take a crack at it please feel free.
Harlan Stenn <stenn at ntp.org>
http://networktimefoundation.org - be a member!
More information about the questions