Gentoo Archives: gentoo-dev

From: 320095285153-0001@t-online.de (Achim Gottinger)
To: gentoo-dev@g.o
Subject: Re: [gentoo-dev] ebuild, ip6tables headaches, kernel confusion, xinetd vs. tcpwrappers
Date: Wed, 10 Jan 2001 03:41:18
Message-Id: 3A5C343E.AA26F8E2@gottinger.de
In Reply to: [gentoo-dev] ebuild, ip6tables headaches, kernel confusion, xinetd vs. tcpwrappers by Justin Vander Ziel
Justin Vander Ziel wrote:

> Hello folks. > > Some more questions for the eventual Gentoo/linux dev-newbies if nothing > else (and for myself, of course B) > > 1. I managed to revise the iptables-1.1.2.ebuild script to handle > iptables-1.2 -- and it worked! That is, except for one small problem: I > forgot to build a package list for ebuild and BTW, how do I do this? > > 2. the ebuild script chokes on versions such as in autofs-4.0.0-pre9.tar.bz2 > (with accompanying ebuild file) -- when I do > > # ebuild autofs-4.0.0-pre9.ebuild merge > > I get the message > > 4.0.0pre9 doesn't appear to be a version or rev string > !!! packagename is not a valid packagename > Exiting
autofs-4.0.0_pre9.ebuild should work. Maybe Daniel can document the naming pattern somewhere on the net.
> > > 3. iptables > when I do > > # modprobe ip6table_filter > # ip6tables -t nat -A POSTROUTING -o ppp0 -j MASQUERADE > > (this is from from Rusty Russel's Linux 2.4 NAT HOWTO from > netfilter.kernelnotes.org) I get the error message > > iptables: v1.2: can't initialize iptables table 'nat': iptables who? (do > you need to insmod?) > Perhaps iptables or your kernel needs to be upgraded
I think that is just a kernel-configuration issue.
> > > (Yes, I upgraded to v1.2 (see number 1 above) thinking perhaps something was > missed and then I did a portage-unmerge of the original sys-apps/iptables... > but I get the exact same error message as I did with 1.1.2 ). > > 4. kernel confusion > On Sun, 7 Jan 2001 08:55:23 -0700, drobbins@g.o wrote: > > To: achim@g.o > > > Another question, what kernel do you use at the moment test12, > prerelease or > > > final ac1/2/3.? > > > > I'm using test12. > OK, could someone please clarify the *NEW* and bungled kernel naming system? > - Is there a match between 2.3.X and the 2.4.0-testX kernels? > - What does final ac1/2/3 refer to? > When all is said and done why not just go with 2.4.0 stable?
2.4.0 does not work correct with reiserfs. Thats why I made packages for ac2/3/4. (2.4.0.2, 2.4.0.3 ...). I use 2.4.0.3-r1 at the moment here. 2.4.0.3 includes a buggy nfs-patch, this patch gets not applied in 2.4.0.3-r1. In 2.4.0.4 there where a few changes to LFS in the kernel and there is only a quick fix for that to work with reiserfs available at the moment, so 2.4.0.3-r1 is the best choice at the moment but you can't use nfs with reiserfs.
> > Ok, so I'm sure you get/will get suggestions until the cows come home and > then some. However, I did notice that you have a bootscripts alternative in > the package list that I did not find particularly impressive (Though I seem > to recall that it was mostly because nothing was written for it, yet or else > I can't remember why). Here is my suggested alternative - Linux Boot > Scripts. In fact, it seems such a good idea to me that I think it should be > THE init script. A full description is available here > (http://www.atnf.csiro.au/~rgooch/linux/boot-scripts/) but here are the > highlights IMHO: > * There is no master script which orchestrates everything. The mini scripts > are kept in /sbin/init.d and init(8) runs all of them, in random order > except where there is a script dependent on another. > * Each script runs any other scripts it depends on, using the need(8) > programme which ensures that a script is only run once. > * runlevels can be supported in one of two ways. The traditional method > would require that the script specified be a directory in which case all the > scripts in the directory would be run. This uses rc1.d etc. with symlinks > back to init.d > R Gooch says that a "more elegant solution is to have a script for each > runlevel, which would look something like this: > #! /bin/sh > # /sbin/init.d/runlevel.3 > case "$1" in > start) > need runlevel.2 > need portmap > mount -vat nfs > ;; > stop) > umount -vat nfs > ;; > esac > # End > which seems to be pretty elegant to me as well. Expert comments?
Sounds interesting, more comments later. :-)
> > > 8. cfengine (http://www.gnu.org/software/cfengine/cfengine.html) > This also seems to be the way to go from what I have heard (mostly from > Christopher Brown at http://vip.hpnc.com/~cbbrowne/linuxsysconfig.html. > However, the e-mail responses from him are all at work). Any thoughts on > this? > > 9. Nifty cd package label in ps if your interested. > I spent a little time in Photoshop LE (I don't have $600 smackers to pay for > the real thing!) and whipped this up. I think it's attractive but it's not > exactly 'production quality'. Anyways if you want a copy I'll send you a > postscript file with the cd label and jacket for gentoo 1.0 rc3 > > Justin (Just enough to be dangerous) Vander Ziel > mailto:zielot@××××××××××.com > > _______________________________________________ > gentoo-dev mailing list > gentoo-dev@g.o > http://www.gentoo.org/mailman/listinfo/gentoo-dev

Replies