MT ( left irc: Ping timeout: 480 seconds
Lin_ (~igor@ joined #fai.
Lin_ (~igor@ left irc: Remote host closed the connection
Mrfai ( joined #fai.
itais ( joined #fai.
i-R2D2101 ( left irc: Remote host closed the connection
i-R2D2 ( joined #fai.
MT ( joined #fai.
09:20 MTMrfai, I think fai-nfsroot should depend on live-initramfs
09:21 MTstill, of course, #452363 is a no-bug
09:29 Mrfaicurrently I will just ognore this bug
09:31 Mrfaia new depends for fai-nfsroot? Mmm. I have to think about it and it make a short test later today.
09:41 MTobviously there is a hard requirement on live-initramfs (independent of any local configuration)
09:42 MTso I guess it would be valid to depend on it
09:42 MTthis would also keep fai from entering testing if live-initramfs ever were unavailable in testing
09:49 lazyb0yAs it doesn't make sense, and doesn't work without, there is no resons to make fai-nfsroot not depending on live-initramfs
09:51 MTI just check /etc/fai/NFSROOT
09:51 lazyb0yAnoither question for the 3.2.X changes howto: is it correct, that make-fai-bootfloppy is not (yet) made to work with fai 3.2? at least it seems it doesn't use the standard debian kernel:
09:51 lazyb0y...
09:51 lazyb0yWriting boot data to floppy.
09:51 lazyb0yThe kernel configuration is /srv/fai/nfsroot/boot/config-2.6.18-fai-kernels.
09:51 lazyb0yAdditional kernel parameters:
09:51 lazyb0y...
09:51 MTI think live-initramfs is really the only package that fai-nfsroot should depend on instead of having it in the NFSROOT file
09:52 MTlazyb0y, yes, I think current SVN trunk even has a die() included
09:52 lazyb0yga
09:52 MTsuch that make-fai-bootfloppy immediately bails out
09:52 lazyb0ygreat...
09:53 MThmm, indeed the bug may be more valid than we thought ...
09:53 lazyb0ymy old new fileserver doesn't boot from pxe as soon as the disk controller is enabled (and wit disabled disk controller, it's not really useful to try an install)
09:53 lazyb0ygogin to try again etherboot, but it never worked for me when i tried it...
09:53 lazyb0ywhat's needed to get the bootfloppy going again?
09:54 lazyb0yMT: the live-initramfs bug? yes, but it's probably only the fai sources.listthat is still, set to etch :)
09:56 MTI guess the debootstrap settings are set as such too
allee ( left irc: Ping timeout: 480 seconds
Bokeh ( left irc: Quit: Leaving
Bokeh ( joined #fai.
10:29 Mrfailazyb0y: booting from floppy will not work, since the kernel and initrd will not fit onto a floppy
Action: h01ger already suggested to remove make-fai-bootfloppy
Bokeh ( left irc: Quit: Leaving
Bokeh ( joined #fai.
10:51 MrfaiI have a nice little problem with make-fai-nfsroot
10:51 MrfaiOn one machine it stalled asking me which live.conf version I like to use. On another machine it just works without asking.
10:52 MrfaiThis is because make-f-n creates /etc/live.conf before installing line-initramfs.
kasimon (~user@ joined #fai.
10:52 kasimonhello :)
10:52 kasimonhow can I specify an alternate nfsroot when doing a dirinstall?
10:53 MrfaiAny solutions? There's an option for dpkg force-confdef and/or force-confold or force-confnew.
10:53 MrfaiSince I do not know them well, I do not like to change things without asking.
10:53 MrfaiCan someone reproduce this problem?
10:56 lazyb0yMrfai: hmm, yes, then h01ger's suggestion might be right. but then, as it is for me now, some people cannot use FAI anymore for some machines...
10:56 lazyb0yand yes, as always, etherboot doesn't work here, again
10:57 lazyb0yaehm, but wait: the argument with space on the floppy doesn't count for iso's!
10:57 Mrfaisorry, but bootfloppy is low priority for me today. FAI 3.2.2 should be released today.
10:58 h01gerlazyb0y, right. (cds)
10:59 Mrfaiguys, I have plans to enhance fai-cd to it can create boot CD's that only contain the kernel and initrd. But now back to the live.conf problem
11:01 kasimonor alternate question: how can I create a i386 dirinstall on a AMD64 system?
allee ( joined #fai.
11:10 lazyb0yMrfai: hmm, no idea about live.conf
11:11 lazyb0yMrfai: but, make-fai-nfsroot should really be the same on all machines, the nfsroot is deleted at beginning
11:11 lazyb0yso it sound a bit strange that it behaves different on different machines...
11:11 lazyb0yare you sure, /etc/fai is exactly the same on both machines where you test it?
11:12 Mrfaino /etc/fai may be different. I did not check it
11:12 lazyb0ykasimon: hmm differnet nfsroots for dirinstall is an interesting questions
11:12 lazyb0yMrfai: you should, then :)
11:13 lazyb0ykasimon: one thing I noticed just recently: dirinstall calls debootstrap on it's own when it doesn't receognize an nfsroot
11:13 MrfaiJust did a complete purge of all fai package, then installed beta16 pacvkages. m-f-n does not ask. So it might be a problem of the other maching using some old configs.
11:13 lazyb0ykasimon: so, if you can make the debootstrap run with arch=i386, ithout using the nfsroot, that might help
11:14 lazyb0ykasimon: but I am not sure, how debootstrap is run from dirinstall, especially interesting would be, if the deboostrap config stuff in make-fai-nfsroot.conf ist used...
11:15 lazyb0yMrfai might know, but he doesn't want to talk about it now, porobably :)
11:18 lazyb0ygah, my bootfloppy problem might apply to 3.2.x, but I just realized, I am still on 3.1.8 where I need it now!
11:19 lazyb0yand, in addition tro that, even etherboot seems to work the very first time for me! (it seemded like it did not, because the first boot device was still the net when I tried, and that just hangs with a blinking cursor on the upper left)
Action: Mrfai SIGLUNCH
allee ( left irc: Remote host closed the connection
allee ( joined #fai.
meandtheshell (~sa@ joined #fai.
sheldonh (~sheldonh@ joined #fai.
12:08 sheldonhdoes make-fai-nfsroot have hooks for installing my own discover-modprobe.conf?
12:08 sheldonhi need to exclude some modules
12:09 h01germakef-fai-nfsroot in >=3.2 has hooks
12:09 h01germaybe earlier too
12:10 sheldonh*sigh*
12:10 sheldonhi wish i'd held out a little longer :)
12:14 h01gershould be easy to backport if you need it
12:15 sheldonhbackporting 3.2 should be easy.  my worry is forward-porting my FAI config and scripts to fai-3.2 :)
12:17 sheldonhit's unfortunate that discover-modprobe.conf doesn't allow types="all -display"
lola22 ( joined #fai.
lola22 ( left irc:
12:27 sheldonhcan i write to files in /etc from within 20-hwdetect.source, or will that mess up the nfsroot for the next guy?
12:27 kasimonlazyb0y: thx for the dirinstall/nfsroot pointers. I'll have a second look
12:29 lazyb0ysheldonh: the nfsroot is read-only normally. but it might be different with unionfs and fai 3.2.x - with unionfs writing might be supported, but it will be only in ramdisk, then - so nothing is messed upo fpr the next install
12:31 sheldonhlet's give it a whirl, then... :)
MT ( left irc: Remote host closed the connection
MT ( joined #fai.
12:53 Mrfailive.conf problem. It does not happen with fai-nfsroot beta13 but with beta16. :-(
12:53 MrfaiOn both machines.
13:03 lazyb0yMrfai: then release bet13 as 3.2.2 :)
13:04 MrfaiI think I found the problem. Currently working on a fix
13:22 Mrfaiok, this is fixed.
Mrfai ( left irc: Quit: leaving
13:25 stockholmwhere should i put "configuration" to fai scripts?
Mrfai ( joined #fai.
13:25 stockholmlike the names of the admins of the servers
13:26 stockholmi would like to have a script that adds users with different levels of authority and a config file to this script
13:27 stockholmwhere would that config file go?
13:27 lazyb0ystockholm: isn'tr something like usually solved with ldap and pam?
13:27 stockholmlazyb0y: pam?
13:27 lazyb0ypluggable authentication modules
13:27 stockholmi need a list of logins at least even in that case
13:27 stockholmthanks, i have heard of pam
13:28 lazyb0yapart from that, I create a default user with some script
13:28 lazyb0yjust a script for my BASE class or something like that
13:28 stockholmand just one user?
13:29 lazyb0yyes, for my personal systems
13:29 lazyb0yif I had more, I'd use pam/ldap
13:29 stockholmi dont how you would want to do that
13:30 lazyb0ybut it shouldn't be a problem to write more useradd statements, or read the users from a file
13:30 stockholmcan you be more specific?
13:30 lazyb0yabout what? pam/ldap or my useradd script?
13:31 stockholm"read the users from a file", right. that is the configuration file i talk about
13:31 lazyb0yok, but that doesn't exist - should be a 5-liner in bash, or something like that
13:31 lazyb0ya little more complex if you want it to have passwords, then perl might be better
13:32 lazyb0ybut we don#t nwant to have passwords in the nfsroot, won't we?
13:32 stockholmlazyb0y: i know how ldap works, but for pam you need to create users, dont you? and if you create users you need a config file
13:32 stockholmlazyb0y: password hashes would do fine
13:33 stockholmlazyb0y: the question is WHERE would the config file for such a script go?
13:33 stockholmlazyb0y: something like /etc within fai
13:33 lazyb0yhmm
13:34 lazyb0yah, you mean, you will then have a script for the BASE class, but depending on other classes, you want it to look different?
13:35 lazyb0yand therefore, the usernames cannot be hardcoded in the scripot itself - something like that?
13:35 stockholmlazyb0y: yes
13:35 stockholmand i might want to use the user names in other scripts
13:36 lazyb0yabout pam/ldap: not sure, I didn't use it yet - I'd expect it to work without having the users created on the machine, and "something" should take care of creating a new home, if a new, allowed user logs in
13:37 lazyb0yotherwise, in a bigger organization, you'd have to create 1000 home dirs, even if only 5 people will ever login on the system...
13:37 lazyb0ybut maybe that can only be used with a nfs-home...
13:38 lazyb0yback to the configs for users, how about a variable, in class/BASE.var ?
13:38 stockholmright
13:38 stockholmthat file can define several variables, right?
MT ( left irc: Remote host closed the connection
MT ( joined #fai.
13:43 MTMrfai, is there any reason for using aptitude instead of apt-get (r4788)
13:44 stockholmMT: ja, aptitude keeps track of what was a required package (leaf package) and what were dependencies
itais ( left irc: Quit: See you
13:44 MT... so ... what?
13:45 MTwhat impact does that have?
13:45 stockholmMT: when removing leaf packages their unused dependencies get removed too
13:45 stockholmyou dont bloat your system
13:45 stockholmor can un-bloat it
13:45 MTah, cool
13:45 MTdidn't know that
13:46 MTjust new deb-orphan
13:46 MrfaiMT: I thought aptitude is better because I can say isntall without-recommends. But it didn't change anything. But I will let it unchanged now, since also /etc/fai/NFSROOT uses aptitude.
13:46 Mrfaibetter that deb-orhpan
13:46 Mrfais/that/than/
13:47 MrfaiJust testing an isntallation from usb stick.
13:47 lazyb0ystockholm: yes, AFAI - not user .var files a lot...
13:47 MToh, I think aptitude is fine, I was just a bit surprised
13:47 h01geraptitude is better becomes it handles conflicts much better. on stable, the diff is not that big, on testing+sid it is
13:48 h01geralso, /me thinks we should use aptitude, as aptitude is the recommended tool, apt-get is deprecated
13:48 stockholmi use apt-get only for apt-get update
13:49 oz_anyone got a good documentation howto build a pxemenu?
13:50 sepoz_, i used /usr/share/doc/syslinux,
13:52 Mrfaioz_: you mean using menu.c32?
13:53 sepor vesamenu.c32
13:57 oz_Mrfai, others: yes, exactly that
13:58 oz_sep: thx...hope that helps
13:58 MrfaiI have an example here, wait
13:58 oz_i don't find menu.32 as a file...
13:58 oz_and apt-file does not find it, too
13:59 oz_ menu.c32, even
13:59 Mrfaikueppers[~]> zgrep menu.c32 $ECONTENTS
13:59 Mrfaiusr/lib/syslinux/menu.c32                                   utils/syslinux
13:59 Mrfaiusr/lib/syslinux/vesamenu.c32                               utils/syslinux
14:00 Mrfai$ECONTENS = .../debian/dists/etch/Contents-i386.gz
14:00 MrfaiIT's nice to have a local mirror
14:01 Mrfaioz_:, but I did not test it. Got it from someone else
14:01 MrfaiI just enhanced my menu.lst on my usb stick to just boot the kernel and initrd, then doing ip=dhcp and the rest via NFS.
14:02 MrfaiNow I got: short read 0<28.....
14:02 MrfaiFine, now I can reproduce the problems of others :-(
14:05 MrfaiANY wishes for FAI 3.2.2? Speak now or remain silent forever ;-)
14:06 oz_:-)
Action: oz_ wants FAI 3.2.2 to cook coffee and install all my App Servers without a flaw :)
14:07 oz_hm. do I have to replace pxelinux.0 with menu.c32 in dhcpd.conf?
14:08 lazyb0yFai 3.2.2 should make me rich and famous!
14:09 lazyb0yand, in general, it should male the world a better place! no wars anymore, enough to live for everyone!
14:09 lazyb0ys/male/make
14:10 sheldonhdon't suppose there's a 3.1 -> 3.2 migrating guide?
14:10 sheldonhcome on, say the changelog, i dare you :)
14:11 MTthere is the NEWS file
14:11 lazyb0ysheldonh: I wrote some things into the wiki
14:11 sheldonhcool
14:11 lazyb0yMT: but AFAIK it only says what has changed, not exactly how a user has to deal with it
14:11 MThmm, yes
14:11 sheldonhthat's fine, i can use the stuff between my ears for that :)
14:12 oz_Ha, /me got pxemenu running. that was easy. :-)
14:12 lazyb0ysheldonh: look at
14:13 lazyb0ysheldonh: I don't think you come up with the idea for the fai sources.list too easy - reading saves you quite some thinking time :)
14:16 sheldonhah, yes i see :)
14:17 sheldonhlooks like the biggest change to worry about is /tmp/target -> /target
14:19 lazyb0ysheldonh: ? do you /tmp/target hardcoded in your scripts?
14:22 lazyb0yMrfai: will you fix the "etch" string in the FAI 3.2.2 packages?
14:22 sheldonhlazyb0y: hmmm, looks like only one place: sed -ie '/^\/dev\/ /d' /tmp/target/etc/fstab
14:23 lazyb0ysheldonh: there is a variable for it - so you are safe for future changes!
14:23 sheldonhhoorah!
14:23 lazyb0yand, this won't work with softupdates, too...
14:23 Mrfailazyb0y: no, the etch string will remain
14:23 lazyb0yMrfai: then it's broken in debian, by default!
14:25 Mrfailazyb0y: it's not broken, but it will only create an etch nfsroot
14:25 sheldonhdoes MD+LVM still require hackery in 3.2?  i'm using peter garvai's mountdisks.LVM hackery
14:25 lazyb0yI know why you do it like that, but we should think about this point again
14:25 MTthis is broken.
14:25 MTMrfai, one can't currently create an etch nfsroot using Debian etch alone
14:25 lazyb0yMrfai: it's broken with the defaulkt config, as the etch repo doesn#t have live-initramfs - as the bug from today said
14:26 MTsheldonh, FAI 3.3 will hopefully contain our new solution
14:26 sheldonhah
14:27 MTfor the moment, you might want to have a look at
14:27 sheldonhi should probably wait for that, and just keeping hacking /srv/fai/nfsroot/etc/discover-modprobe.conf
14:27 Mrfaimm, so FAI 3.2 alread was broken and noone noticed it.
14:27 MTtrue
14:27 sheldonhmind you, i have a feeling _no_ fai users want discover-modprobe to load display drivers
14:27 Mrfaiwhat about adding this line to /etc/fai/apt/sources.list: deb etch koeln
14:27 MTI think a possible fix would also be a use of your repository in /etc/fai/apt/sources.list
14:27 MT:-)
14:28 MTI guess that should work
14:28 lazyb0yhmm, sound quite ugly to have this repo in a debian package...
14:28 MThmm, why not?
14:28 MTof course, at some point we will drop etch compatibility anyway
14:29 lazyb0yI cannot explain it, it just "sounds/feels" ugly
14:29 MTthen this will be removed again
14:29 MTsomebody also pointed at backports I think
14:29 lazyb0yhow about mrfai's debian upload script replacing etch with lenny?
14:32 Mrfaiwhich upload script?
14:33 lazyb0yMrfai: whatever you use when uploading to the debian repo (I don't know how this exactly works)
14:34 MrfaiI just upload to unstable, then packages will migrate to testing after a while
14:35 lazyb0yI know that, but at some point, you upload sour sources - that's the place where etch could be replaced with lenny
14:37 lazyb0ybut, whatever, we probably waste too much time discussing this - it gets more interesting, and a real solution will be needed when etch and lenny differ in more than this single string at two places...
14:37 MrfaiI will add my repository to the sources.list line
Action: h01ger is with lazyb0y. adding an external, unsigned repo is ugly, IMNSHO. its fine in a branched version from an unsigned repo, but in main...
kugg ( left irc: Ping timeout: 480 seconds
15:04 Mrfaican someone check if the koeln etch repository is correct?
15:05 MrfaiOK FAI 3.2.2 was uploaded. Announcement will be sent later.
15:06 it boots a not so nice ubuntu kernel.
jeedi ( joined #fai.
15:23 jeedihi
15:23 h01gerhi jeedi
15:24 jeedinext problem with my fai-stuff...
15:24 jeedii was trying to set up two machines with different classes today
15:26 jeediadded basetest*) and webtest*) to config/class/50-host-classes
15:27 jeediand added BASETEST and WEBTEST package lists to config/package_config/
15:28 jeedistrange thing is: the machines don't seem to get a hostname when booting..
15:28 jeedithey simply assume their ip-address is their hostname.
15:37 jeediany ideas how to get the machines to set their hostname properly?
kasimon (user@ left #fai (ERC Version 5.1.4 (IRC client for Emacs)).
sheldonh (sheldonh@ left #fai (
MT ( left irc: Remote host closed the connection
MT ( joined #fai.
jeedi ( left irc: Quit: Leaving
sepski (~sep@ joined #fai.
17:01 Mrfaiuse-host-decl-names on; in dhcpd.conf
Action: h01ger forwarded that to jeedi
MT ( left irc: Remote host closed the connection
MT ( joined #fai.
sepski (~sep@ left irc: Remote host closed the connection
itais (~itais@ joined #fai.
Lin (~igor@ left irc: Remote host closed the connection
Mrfai ( left irc: Quit: leaving
itais (~itais@ left irc: Quit: See you
20:14 MTh01ger, could you take a look at #419788 (and #360183)?
20:15 MTI just checked once again that this problem is indeed fixed
Lin (~igor@ joined #fai.
20:28 h01gerMT, its on my list, but currently fai is not on the top there.. :/
20:28 h01gerbut all my scripts use -i so i'm interesting it checkin this
MT ( left irc: Ping timeout: 480 seconds
meandtheshell (~sa@ left irc: Quit: Leaving.
MT ( joined #fai.
allee ( left irc: Remote host closed the connection
--- Fri Nov 23 2007

Generated by 2.1_h01ger1 by Jeff Waugh - find it at! Customized by Holger 'h01ger' Levsen