Date: Wed, 27 Oct 2004 15:17:00 +0200 From: Marton Kenyeres <mkenyeres@konvergencia.hu> To: freebsd-security@freebsd.org Subject: Re: please test: Secure ports tree updating Message-ID: <200410271517.00682.mkenyeres@konvergencia.hu> In-Reply-To: <xzp654wiffv.fsf@dwp.des.no> References: <417EAC7E.2040103@wadham.ox.ac.uk> <xzp654wiffv.fsf@dwp.des.no>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wednesday 27 October 2004 13:11, Dag-Erling Sm=F8rgrav wrote: > Colin Percival <colin.percival@wadham.ox.ac.uk> writes: > > CVSup is slow, insecure, and a memory hog. > > if cvsup is slow, you're not using it right. Does using CVSup over an asymmetric link qualify as `not using it=20 right`? [From http://www.cvsup.org/howsofast.html ] "The volume of data sent by the client is comparable to that sent by the=20 server. On a typical full-duplex link, this effectively doubles the=20 usable bandwidth." It still can be quite fast due to it's diff based nature. Also it is=20 more widespread than portsnap, which is not really surprising, but=20 makes the probability of finding a fast mirror higher. (For example,=20 from my office the avg roundtrip to the portsnap site is 7 times the=20 roundtrip to the local CVSup mirror.) I'm thinking about making some mesurements with different updating=20 methods (AnonCVS, CVSup, CVSync, rsync, portsnap come to mind) over=20 symmetric and asymmetric lines. Any suggestions on what typical usage scenarios and updating practices=20 might be are welcome. (e.g. once a day / once a week / when freshports=20 notifies me that something on my watchlist has changed). > > I'm sure portsnap is a wonderful piece of software, but there's no > need to spread FUD about cvsup to promote it. I agree with that. > > DES m.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200410271517.00682.mkenyeres>