Date: Wed, 7 Oct 1998 13:38:04 +0930 From: Greg Lehey <grog@lemis.com> To: Eivind Eklund <eivind@yes.no>, doc@FreeBSD.ORG Subject: Re: Dos and Don'ts Message-ID: <19981007133804.U27781@freebie.lemis.com> In-Reply-To: <19981007053916.36507@follo.net>; from Eivind Eklund on Wed, Oct 07, 1998 at 05:39:16AM %2B0200 References: <19981006071237.02443@follo.net> <19981006155341.C27781@freebie.lemis.com> <19981006083809.00946@follo.net> <19981007123122.O27781@freebie.lemis.com> <19981007053916.36507@follo.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wednesday, 7 October 1998 at 5:39:16 +0200, Eivind Eklund wrote: > On Wed, Oct 07, 1998 at 12:31:22PM +0930, Greg Lehey wrote: >> On Tuesday, 6 October 1998 at 8:38:09 +0200, Eivind Eklund wrote: >>> On Tue, Oct 06, 1998 at 03:53:41PM +0930, Greg Lehey wrote: >>>> On Tuesday, 6 October 1998 at 7:12:38 +0200, Eivind Eklund wrote: >>>>> Dos and Don'ts of FreeBSD >>>>> ------------------------- >>>>> >>>>> DON'T run pppd unless you either >>>>> (a) already have a working setup, or >>>>> (b) absolutely need the 2% reduction of CPU usage it will give >>>>> you. >>>> >>>> Why this? With all respect for Brian, I've found pppd to be more >>>> reliable. >>> >>> Because it often lead to a large amount of pain (in setup), especially >>> when somebody get the idea that they want to use NAT. In the cases >>> where there are problems with iij-ppp, my impression is it usually get >>> fixed pretty quickly (personally, I've never had a problem except when >>> I've been hacking the code myself, so I can't give more than a >>> second-hand impression). >> >> OK. I had to go through installing both in painful detail for my >> book. Everybody has always said "don't use Kernel PPP, it's painful", >> but I didn't find it so. In fact, the difficulty of installation is >> about equal. What *is* deficient is the documentation. > > ... and the lack of a term mode. When I set up ppp, I found this very > convenient, as I could trivially test that PPP worked in itself before > I had to make sure chat scripts etc worked. Well, yes, that's nice, though I found it rather complicated to use. > [... sad story involving iij-ppp removed ...] >> I moved (quickly and painlessly) to kernel PPP, and since it works, >> I've been using it ever since. Note also that most PPP problems >> reported to -questions are for user PPP, not kernel PPP. This is >> almost certainly because most people try user PPP, but it does >> suggest that this rule is unnecessary. > > The most severe problem is the introduction of NAT, which it seems a > large amount of people are doing. I regularly see people struggle > with setting up natd (due to lacking network understanding, mostly), > and people that have pppd running try to set up natd instead of > switching to iij-ppp. This is a shame, both because setting up natd > is a pain, and because natd will NAT wrongly for any packets coming > before with dynamic IPs are assigned (and this is non-fixable). Hmm. I set up natd for the book as well, and I didn't have much trouble. At the time, I think it was the only game in town. I haven't been watching what Brian's put into userland PPP, but I'm sure that we could find some reason to still want to use natd (for example, its symbiosis with ipfw). I still think that there's not enough pain in kernel PPP to warn against using it. > However, I'll remove the statement if you accept that I give your > phone-number to anybody that come asking me how to fix their problems > with setting up natd to run with their already setup pppd (this is not > more than two or three people each day so far) ;-) How come so many ask you? I haven't seen any worth talking about. Anyway, sure. Also point them at http://www.cdrom.com/titles/os/bsdbook2.htm :-) > Seriously: I see a _lot_ of users drop by #freebsd and asking for help > with this, which is why I included it. It was the one thing "This > _really_ don't work" which led to me starting the Do/Don't list. Bounce me some of these messages, if that work in IRC. >>> DON'T send questions about anything (beyond the exceptions noted >>> above) to any mailing list you have not read at least two days >>> of traffic from. Yes, this implies you should be or have been >>> a member. >> >> I suppose so. We have a policy that non-members can post to >> -questions; how would you factor that in ("you're allowed to, but it's >> better not to"). > > The present text is > > DON'T send mail to any of the FreeBSD mailing lists not listed above > (plus freebsd-newbies@freebsd.org) before you have read the s/plus/including/ > mailing list you're thinking of mailing for at least a couple > of days. Which types of mail that is OK for which list varies > a lot, and it take a little while to pick up the 'culture' > (even when you've read the charters). Yes, I suppose so. How about adding "the charters of some lists allow you to send messages without being subscribed, but this doesn't make it a good idea". Greg -- See complete headers for address, home page and phone numbers finger grog@lemis.com for PGP public key To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-doc" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19981007133804.U27781>