Date: Tue, 11 Oct 2005 13:56:26 -0700 From: Brooks Davis <brooks@one-eyed-alien.net> To: Joshua Coombs <jcoombs@gwi.net> Cc: freebsd-current@freebsd.org Subject: Re: FreeBSD 6.0-RC1 available Message-ID: <20051011205626.GA13461@odin.ac.hmc.edu> In-Reply-To: <digtbv$9jk$1@sea.gmane.org> References: <434BCDF6.3090303@samsco.org> <digpfu$r1b$1@sea.gmane.org> <434BEC96.4050801@samsco.org> <digtbv$9jk$1@sea.gmane.org>
index | next in thread | previous in thread | raw e-mail
[-- Attachment #1 --] On Tue, Oct 11, 2005 at 01:39:10PM -0400, Joshua Coombs wrote: > > "Scott Long" <scottl@samsco.org> wrote in message > news:434BEC96.4050801@samsco.org... > >Joshua Coombs wrote: > > > >>Given that we're in the RC stage, is it too late to get a PR > >>stuffed in to correct ntpd/ntpdate behavior in /etc/rc.d? > >> > >>Joshua Coombs > >> > > > >Depends on the kind of problem that you are seeing. Any details? > > > >Scott > > Two issues I spotted. > > First, ntpdate is still being used, despite it being marked as > depreciated for quite some time. The preferred replacement is to call > ntpd with the -q switch. This causes ntpd to match ntpdate's > behavior, it sets the time, and then exits with a status code > indicating success or failure. > > Second, when ntpdate, or what ever is set in ntpdate_program is > called, there is an IP appended to the end of the arguments, which is > not controlled by ntpdate_flags. This means you cannot setup ntpd -q > to be used in place of ntpdate without editing the ntpdate rc.d script > or creating a new one, a regression from 4.x's setup. > > I see some effort was put into the new ntpdate rc.d script, having it > pull potential servers from /etc/ntp.conf rather than require the user > specify one in rc.conf using ntpdate_flags. ntpd called with -q uses > the ntp.conf server entries automatically, so the extra work by the > rc.d script isn't required if we switch to ntpd -q in place of > ntpdate. > > I was going to work up a tweaked ntpdate rc.d script that included a > new option, ntpdate_use_ntpd, that when set, would use the preferred > practice of calling ntpd -q after verifying a valid ntp.conf exists. > If one isn't present, I was going to have it throw a warning, and > reference an example conf using pool.ntp.org servers to get baseline > time established. The ntpd rc.d script would receive the same check, > the example conf would lock ntpd down such that it would only operate > as a client for the local machine, and not act as a server for > external hosts, or respond to external ntp query/command/conf > requests. > > Unfortunately, I'm getting into this rather late, I just moved my 386 > to 6.0b5 this weekend, hence my tardiness discovering ntpdate still in > use in later releases of FreeBSD. I would like to see the correct > behavior implemented for release, but if I'm beyond the deadline for > this level of change, I'll accept that and work on making it the norm > for 6.1. Since this is a matter of mostly pedantic correctness, I can't see risking the release on these changes. -- Brooks -- Any statement of the form "X is the one, true Y" is FALSE. PGP fingerprint 655D 519C 26A7 82E7 2529 9BF0 5D8E 8BE9 F238 1AD4 [-- Attachment #2 --] -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.1 (GNU/Linux) iD8DBQFDTCb6XY6L6fI4GtQRApNQAJ9fNYj27SXEy6MH/RrkFPaNCSrVrwCg1GbQ MG7tlmaLhIpRR/DLnspJlxg= =p6GL -----END PGP SIGNATURE-----help
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20051011205626.GA13461>