Date: Fri, 3 Jul 2009 04:21:02 +1000 (EST) From: Bruce Evans <brde@optusnet.com.au> To: David Malone <dwmalone@maths.tcd.ie> Cc: Edwin Groothuis <edwin@FreeBSD.org>, freebsd-net@FreeBSD.org Subject: Re: NTP - default /etc/ntp.conf Message-ID: <20090703035833.E40558@delplex.bde.org> In-Reply-To: <200907020955.aa73037@walton.maths.tcd.ie> References: <200907020955.aa73037@walton.maths.tcd.ie>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 2 Jul 2009, David Malone wrote: >>> The local clock should only be configured on a single server in a >>> NTP domain that might be disconnected from the rest of the tree. >>> Since the default config is a client config, it doesn't make sense >>> to have the local clock configured. Even if it was a server config, >>> it still wouldn't make sense, because it is only useful if a single >>> server has it configured. > >> I do not see the point in removing it, it helps to keep the ntpd daemon >> running if for some reason it loses the "real" ntp servers. > > It's definitely a misconfiguration to ship it by default. If you > have many clients all with a local clock configured, then, when > disconnected, they all just follow themselves rather than following > the clock on a server. If you have it configured on several servers > you end up with some clients following each of the servers, but > they won't all stay together unless you're lucky. For this use of > the local clock, you only want one local clock per island that might > become disconnected. This is handled by having different local clocks in different strata. I only use 2 such strata -- 1 for server and another for all clients. But this is not so easy to set up as a default. ref8 FreeBSD cluster machines have the new ntp.conf, so they have the local clock with the same stratum and this misconfiguration is getting some official testing :-). It is probably actually unofficial, due mergemaster blowing away the old primitive ntp.conf which is still used on other FreeBSD cluster machines. This seems to break the server config generally -- ntptrace on ref8* hangs after localhost while ntptrace on ref7-amd64 hangs after hub. Bruce
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20090703035833.E40558>