[00:54] thpo_ (~thpo@dotrc.de) left irc: Ping timeout: 480 seconds [00:54] thpo (~thpo@dotrc.de) joined #fai. [01:35] MT (~MT@dove.informatik.tu-muenchen.de) left irc: Ping timeout: 480 seconds [07:33] fai-guy (~fai-guy@p57B9C1D8.dip0.t-ipconnect.de) joined #fai. [07:55] meandtheshel1 (~sa@85-127-137-224.dynamic.adsl-line.inode.at) joined #fai. [08:03] allee (~ach@allee.mpe.mpg.de) joined #fai. [08:14] Nick change: tokkee_ -> tokkee [08:20] Nick change: \sh_away -> \sh [08:31] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [08:43] MT (~MT@dove.informatik.tu-muenchen.de) left irc: Ping timeout: 480 seconds [09:07] Mrfai (~lange@kueppers.informatik.uni-koeln.de) joined #fai. [09:11] nocturn (~nocturn@d51A4378A.access.telenet.be) joined #fai. [09:11] Morning guys [09:16] ErKa (keryell@193.50.97.163) joined #fai. [09:39] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [09:49] I was in here last week because I'm stuck setting up FAI on Debian etc [09:49] s/etc/etch [09:49] My client does a PXE boot but stops on Trying to load pxelinux.cfg/C0A80145 [09:50] I can download the file fine with a tftp client [09:55] I only have one suspicious log entry, tftp says [09:55] tftp: client does not accept options [09:57] but you had tftpd-hpa installed, right? [09:59] jcastang (~jcastang@laptop-147-210-128-147.labri.fr) joined #fai. [09:59] hi all [10:00] oz_: yes [10:00] And it works well with the tftp client [10:01] you can ignore the message: .. does not accept optionss [10:01] strange thing is that when I run in.tftpd with -vvv, I get an entry when the client requests pxelinux.0 [10:01] but no entries for pxelinux.cfg/C0A80145 [10:02] I have a problem with java using FAI ;) [10:02] It can't be automaticly installed because of the license. [10:02] Mrfai: Ok, I thought it indicated a problem with the PXE loader code in the node's firmware... [10:03] as anyone fixed this problem ? [10:04] jcastang: IIRC someone had solved this. But I have no pointer currently. [10:04] Action: oz_ had a solution for it [10:04] nocturn: check if you install client really tries downloading from you tftp server. [10:05] but I wonder that it's not in the wiki jet [10:05] momplz [10:05] don't know, juste take a look [10:05] oz_: do not wonder about it, but add it :-) [10:05] https://lists.uni-koeln.de/pipermail/linux-fai/2007-May/004954.html [10:06] Mrfai: ;) [10:06] thanks [10:06] I thought got the info from there [10:06] Mrfai: I see the entries for pxelinux.0, which downloads succesfully: [10:06] RRQ from 192.168.1.69 filename pxelinux.0 [10:06] nocturn: so you should also see request for pxelinux.cfg/C0A80145 [10:06] But no entry for the configuration file, so presumably, the client didn't send it correctly [10:07] nocturn: maybe this pxelinux.0 comes from somewhere else than /srv/tftpd/fai [10:07] Mrfai: next entry is the one about the options [10:07] yep. Or it send it to another server [10:07] no, it's the only place there... [10:07] nocturn: did you double check it? [10:07] There's no other server, they are on an isolated network [10:07] use tcpdump to see if those tftp requests are really send to your server [10:08] Action: Mrfai has to drive by bike through the ugly rain now. brb [10:08] I mean, on the filesystem on your host [10:08] I'll try that, I'm first going to try a second node [10:08] Mrfai: though guy [10:09] oz_: There are 2 pxelinux.0 files on it [10:09] /srv/tftp/fai/pxelinux.0 [10:09] /usr/lib/syslinux/pxelinux.0 [10:09] tftp has /srv/tftp/fai/ as root [10:09] /usr is not shared by any means, ftp, nfs, ... [10:10] nocturn: can you prove that your tftpd does agree to this? :) [10:10] like Mrfai said, did you look with tcpdump at it? [10:10] did you strace the daemon? [10:11] I'm going to try a second node first (to see if the results are identical), then I'll run tcpdump [10:20] Same result on another node [10:20] And I got a dump file [10:21] which says what? [10:21] loading it in wireshark [10:24] I see the read request for pxelinux.0 [10:25] and the subsequent answer and data stream [10:25] then a read request for pxelinux.cfg/xx-xx-xx (MAC address) [10:26] OK! [10:26] the request for pxelinux.0 is to 192.168.1.1 [10:26] which is the master (tftp server) [10:26] the other requests are to 0.0.0.0!!! [10:28] How is this possible? [10:30] interesting... [10:32] Nick change: MT -> Guest795 [10:32] Guest795 (~MT@dove.informatik.tu-muenchen.de) left irc: Read error: Connection reset by peer [10:32] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [10:33] some msgs on the web say it could be a problem with your standard-route [10:35] nocturn: seems to be a dhcp prob [10:35] Strange that it would take pxelinux.0 [10:35] but I can't dig into it any further [10:35] cos I am @ work [10:35] dhcpd config says: [10:36] option routers 192.168.1.1; [10:36] the FAI server is dhcp and DNS server and firewall [10:38] oz_: what search did you run to find the standard-route answer? [10:39] http://www.google.de/search?hl=de&q=pxelinux+0.0.0.0&btnG=Google-Suche&meta= [10:40] I have another problem, more complex [10:41] I want to setup a pxe menu (that's ok) to select wich type of installation I want to make [10:41] here I want to chose between destroy /home and preserve /home [10:41] jcastang: that's easy! nocturne's prob is quite a thing [10:42] don't understand [10:48] I'm getting somewhere... It seems related to the next-server option in dhcpd in newer versions [10:51] jcastang (~jcastang@laptop-147-210-128-147.labri.fr) left irc: Remote host closed the connection [10:52] Found it! now got the next error [10:53] jcastang (~jcastang@laptop-147-210-128-147.labri.fr) joined #fai. [10:54] Now it says: could not find ramdisk image... [10:54] which indeed is not there... [10:54] Do I need to generate this manually? [10:57] ErKa (keryell@193.50.97.163) left irc: Ping timeout: 480 seconds [11:05] Mrfai (~lange@kueppers.informatik.uni-koeln.de) left irc: Quit: leaving [11:05] Mrfai (~lange@suenner.informatik.uni-koeln.de) joined #fai. [11:14] can I use some boot parameters in pxe config to define classes for hosts ? [11:14] in order to select wich type of installation I want ? [11:22] yes. All kernel parameters are also defined as variables. So you could add an option like MYC=A,B,C [11:22] Then you have to write a script, that defines classes depending on this variable definition. [11:23] so MYC can be reached in booted OS as $MYC variable ? [11:24] yep [12:36] Mrfai, you still use "Homepage:" in the description of the fai binary packages. as its in the source packages control header this is redudant information [12:36] I know [12:36] i thought so, but wasnt sure :) [12:51] ctp (~ctp@brsg-4d079517.pool.mediaWays.net) joined #fai. [12:53] moin allerseits. bin langsam am verzweifeln, was unser fai setup angeht (ubuntu). bekomme beim booten des clients folgende meldung: [12:53] registering unionfs 1.3 [12:53] uups. 1.4 [12:54] unionfs: debugging is not enabled [12:54] unable to handle kernel NULL pointer dereference at xxxxxxxxx [12:54] und einen dicken call trace dahinter [12:55] muss man bei fai unionfs nutzen? [12:55] ubuntu gutsy? [12:56] ja [12:57] ctp: Da ist unionfs via NFS defekt. [12:57] es gibt aber einen workaround [12:57] wollte urpsrünglich debian etch, aber der standardkernel hat enorme probleme mit intels ich9 chipsatz. noch nichtmal sata erkannt. [12:58] ah, moment. gerade was neues erreicht [12:58] ja,ja. Eine Installation faengt beim HArdwarekauf an. [12:59] meandtheshel1 (~sa@85-127-137-224.dynamic.adsl-line.inode.at) left irc: Quit: Leaving. [12:59] und 2.6.22 auf backports.org [12:59] Mrfai, es geht um ein cluster mit 8 quads. haben uns für asus pundits entschieden. wie gesagt, intel ich9 [13:00] h01ger, das problem ist, wie bekomme ich etch installiert, wenn der installer noch nichtmal sie sata platten erkennt? vermutlich irgendein hickhack über debootstrap? [13:00] I don't understand german, but are you saiyng that FAI kernel with ubuntu gutsy as been fixed ?? [13:00] ctp, fai?! [13:00] h01ger, der fai server ist einer der knoten [13:01] http://kmuto.jp/debian/d-i/ hat d-i etch images mit 2.6.23 [13:01] jcastang, i had some hw detection trouble with etch so switched to gutsy based on 2.6.22 [13:01] h01ger, ah, ok [13:02] jcastang, no. so far they've been saying gutsy is buggy, bu there is a workaround. (but not where the workaround is :) [13:02] h01ger: yes I used some tricks to use FAI with gutsy [13:02] h01ger, ich versuche es noch 1-2 std mit gutsy und dann werde ich definitiv wieder zurück zu etch. bin auch kein fan von *buntu [13:02] and it's working quite fine [13:03] jcastang, /me doesnt want to hear about it. (i hear more than enough about ubuntu problems here.) i'm happy with debian :) [13:03] ;) [13:03] jcastang, i began to write this howto: http://wiki.slashconcept.com/index.php/Fully_Automatic_Installation_FAI_with_Ubuntu_7.10_Gutsy_Gibbon_Howto [13:04] fine, I may have a look (and complete if so) [13:04] jcastang, i always document the non trivial stuff to help others to get it working ;-) [13:04] gutsy workaround: create a feisty nfsroot and replace the base.tgz inside this nfsroot with a gutsy base.tgz. Then adjust the sources.list to use gutsy. [13:04] jcastang, what did you make to get fai working? [13:05] ah, ok ;-) [13:05] so you are running feisty _during_ isntallation (also the kernel), but you will install a gutsy ssytem [13:05] yurk ! I don't understand german so I can't read this wiki [13:05] To use gutsy with fai, I set up a working FAI installation with feisty [13:05] and so update to gutsy [13:05] jcastang, i translate most of the docs beeing finished with the german version ;-) [13:06] So I use feisty kernel with FAI and configure Fai to make gutsy installation (in source.list) [13:07] ok, so i'll switch back to etch [13:07] as my favorite ditro [13:07] lol :) [13:07] but thx for info ;-) [13:07] no pb [13:08] by the way: anyone here knows about some trouble running fai on amd64 (under etch)? [13:09] ctp: no, amd64 should work. [13:11] great. [13:11] mni thx guys ;-) [13:21] ErKa (keryell@m2.wifi.enstb.org) joined #fai. [13:28] I got to booting everything, but now it complains about: [13:29] /lib/modules/2.6.18-fai-kernels [13:29] Something wrong with my root path? [13:30] nocturn: which FAI version are you using? fai-kernels is not used any more in FAI >=3.2 [13:31] fai-server 3.2.4 [13:35] I can set root-path in dhcpd.conf, but that does not clear the error... [13:36] nocturn, apt-get remove fai-kernels [13:36] read NEWS.Debian [13:36] my nfsroot has that directory: /export/fai/nfsroot-i386/lib/modules/2.6.18-fai-kernels [13:36] nocturn, recreate the nfsroot too [13:44] h01ger: hast du mir den request geschickt bzgl. alioth? [13:44] hab so ne komische mail bekommen, wo kein absender drin steht [13:45] ja [13:45] fuer holger, ich bin nur als h01ger-guest member... [13:45] komisch, das der username leer war. Ganz leer. [13:45] #alioth :) [13:46] ok, holger geherot nun dazu, holger-guest nicht mehr [13:46] danke [13:47] MT: ich werd dich auch mal mit DD namen hinzufuegen und michael-guest loeschen [13:47] wollt ich gerade sagen :-) [13:47] mt@d.o ... [13:47] thx! [13:47] Mrfai, username feld ist auch bei ner mail zu nem anderem projekt leer, ich melde es mal in #alioth,bin eh noch da :) [13:47] meandtheshel1 (~sa@85-127-137-224.dynamic.adsl-line.inode.at) joined #fai. [13:47] What should I set TFTPROOT to? [13:47] d.h. ich sollte das svn neu auschecken, nehme ich an!? [13:48] MT: keine Ahnung [13:48] hast Du mich schon hinzugefügt? [13:48] dann versuche ich es mal ... [13:48] TFTPROOT=/srv/tftp/fai [13:48] MT: yep [13:48] MT: weiss aber nicht ob alioth ne Zeit lang braucht um die Daten vielleicht zu synchronisieren [13:49] I'll try ... [13:50] MT, svn rebase oder sowas gibt es [13:50] sach mal bescheid, wenn du es hast ;) [13:51] svn switch vermutlich [13:51] alioth syncronisiert jetzt "schnell" [13:51] jup, das wars [13:51] und "schnell" ist durchaus schnell.. [13:53] hmm, scheint aber noch nicht zu funktionieren [13:55] h01ger, muss man da auf alioth erst einen neuen Account anlegen? [13:56] ich _glaub_ nicht, aber da gibs ne faq, momang [13:56] http://wiki.debian.org/Alioth/FAQ [13:56] #5 [13:58] muss ich mir dann für mt ein neues Passwort schicken lassen? [13:59] ... ja ... :-) [14:00] wo kann man denn den Account löschen lassen? [14:01] #4 iirc :) [14:02] I removed fai-kernels, rebuild nfs-root [14:03] but still the same, at boot time, the client looks for those files in /lib/modules/2.6.18-fai-kernels [14:04] maybe you have fai-kernels still mentioned somewhere in /etc/fai/ [14:04] ?` [14:04] you have to copy the new kernel to your tftproot directory. [14:05] Mrfai: from where? [14:06] from the nfsroot. Normally make-fai-nfsroot is doing that [14:06] call make-fai-nfsroot -v again and paste the log to paste.debian.net [14:08] Action: nocturn is making nfsroot [14:10] svn switch --relocate svn+ssh://michael-guest@svn.debian.org/svn/fai/trunk svn+ssh://mt@svn.debian.org/svn/fai/trunk [14:11] h01ger ... [14:12] MT, thanks! [14:16] There's no vmlinuz* on the nfsroot now... [14:18] result of make-nfsroot [14:18] http://paste.debian.net/44585 [14:20] WARNING: These unknown packages are removed from the installation list: linux-image [14:20] this is wrong. [14:20] that's why you do not have a kernel inside the nfsroot [14:21] Oops... [14:21] if you have upgraded from an older FAI version, check all files in /etc/fai. Especially /etc/fai/NFSROOT [14:23] Mrfai: I switched from the Debian etch version to that on the official fai repos [14:23] Corrected and building again... [14:23] so you have updated from 3.1.8 to 3.2.x and maybe you are still using the old version of files in /etc/fai. [14:33] Mrfai: you've received my anwsers to questionaire? :) [14:33] +the [14:34] MT (~MT@dove.informatik.tu-muenchen.de) left irc: Remote host closed the connection [14:34] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [14:43] Mrfai: Now it stops at: can't open /scripts/live.... [14:49] Action: h01ger thinks we need better upgrade documentation [14:49] nocturn, you do have the koeln repo in your nfsroots sources.list? [14:50] checking... [14:51] h01ger: no, it wasn't there... [14:51] Do I also need to deal with the koeln PGP key? [14:51] i dont think so, fai uses insecure apt on default [14:52] Ok, rebuilding nfsroot [15:09] Better... but now it dumps me to a busybox shell [15:09] What could be wrong? [15:16] ctp (~ctp@brsg-4d079517.pool.mediaWays.net) left irc: Ping timeout: 480 seconds [15:25] ctp (~ctp@brsg-4d0790e7.pool.mediaWays.net) joined #fai. [15:33] etch installs 3.1.8 as default (fai-kernels suck). which source do you prefer to get 3.2 running on an etch box? backports? [15:34] backports [15:49] And now it stops at Calling 10-base-classes OK [15:51] nocturn: what does fai.log tell you? [15:57] oz_: I can't get a shell [15:59] I'm giving up for today... [15:59] Time to go home anyway [15:59] Thanks for all your help guys [16:00] nocturn (~nocturn@d51A4378A.access.telenet.be) left irc: Quit: leaving [16:22] ErKa (keryell@m2.wifi.enstb.org) left irc: Remote host closed the connection [16:28] ed: yes. Email received. [16:29] ctp: use the koeln repository as described on the main fai web page [16:34] Mrfai, thx [16:38] I cannot set keyboard to fr-latin9, it always set us keyboard instead on console (tty) [16:38] as anyone got this error (with other keyboard too) [16:48] jcastang (~jcastang@laptop-147-210-128-147.labri.fr) left irc: Quit: Ex-Chat [17:01] fai-guy (~fai-guy@p57B9C1D8.dip0.t-ipconnect.de) left irc: Quit: leaving [17:12] Nick change: \sh -> \sh_away [17:24] meandtheshel1 (~sa@85-127-137-224.dynamic.adsl-line.inode.at) left irc: Quit: Leaving. [17:50] MT (~MT@dove.informatik.tu-muenchen.de) left irc: Ping timeout: 480 seconds [18:01] Mrfai (~lange@suenner.informatik.uni-koeln.de) left irc: Quit: leaving [18:44] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [19:05] hi folks. what must i pay attention to building a fai kernel? i'm trying to get 2.6.23.1 running as fai kernel but booting a new client results in a BusyBox shell. [19:05] Begin: Mounting root file system ...... [19:06] eth0: e1000_watchdog: NIC Link is up 1000 Mbps Full Duplex [19:06] ---> BusyBox shell opens [19:06] cat live.log [19:07] to get a kernel running fai you must patch it for unionFS [19:07] you must have nfs root [19:07] tcp nfs [19:07] network built in [19:07] nfs works fine [19:07] and the ramfs enabled [19:08] ok, the last lines from live.log: [19:08] Begin: Trying netboot from 192.168.4.1:/srv/fai/nfsroot ... [19:09] Begin: Trying nfsmount -o nolock -o ro 192.168.4.1:/srv/fai/nfsroot /live/image ... [19:09] Done. [19:09] mount: Mounting unionfs on /root failed: No such device [19:10] hm, unionfs on /root? [19:12] by the way: there's no way to install fai clients without unionfs? [19:25] no./ [19:25] unionfs apply nicely on 2.6.23.9, I've compiled it 2 days ago [19:27] ok. that's the one i prefer. 2.6.23.9 or 2.6.24-rc4. so you patched 2.6.23.9 against unionfs patches? [19:30] yep [19:30] works nicely [19:31] just installed 5 box with that kernel today [19:31] oh, andrew morton's mm branch is already patched against unionfs 2.x. i am using hos patches on my desktops ;-) [19:32] you love instability! :) [19:32] ed: ok, mni thx for your infos [19:32] np [19:32] if you want I've a dot deb kernel package [19:32] patched against new e1000 driver [19:33] ed: oh, ok. running intel board here with intel gigabit, so your kernel seems to be the one i wanna build ;-) do you have webspace to upload the deb? [19:33] yeah [19:34] the kernel is build mainly for vostro 200 [19:34] (dell system) [19:34] give me a minute to upload [19:34] ed: great ! [19:34] do you wanna initrd & vmlinuz for pxe ? [19:34] we're running 8 quads (asus pundit series) [19:35] ed: all you have ;-) [19:35] i've not included all sata & eth drivers, so perhaps it'll not work on your system.. :) [19:35] give it a try [19:35] 'k [19:35] well, fine ;-) [19:39] 1 min left [19:41] http://daboog.zehome.com/~ed/fai_kernel_2.6.23.9-2.tar.gz [19:42] hm, the unionfs diff applies on the vanilla 2.6.23.9 from kernel.org? there's lot of git info (git-clone etc.) on the unionfs site. [19:42] ed: thx again ;-) [19:42] fast download. ready ;-) [19:43] :p [19:43] the dot deb can be used as host & boot kernel [19:43] the initrd and vmlinuz should only be used for boot [19:43] ('cause live) [19:43] really fine [19:53] meandtheshel1 (~sa@85-127-137-224.dynamic.adsl-line.inode.at) joined #fai. [20:02] does it works? [20:05] ed: hm, the client hangs during a "request_module: runaway loop modprobe binfmt-464c" [20:06] don't know what's that binfmt-464c ! [20:06] strange [20:06] maybe i2c stuff [20:06] i try to patch 2.6.23.9 [20:06] you used a simple patch -p1 foo.diff ? [20:08] yeah [20:09] (prefer using stdin ;)) [20:09] patch -p1 < foo.diff :) [20:13] ok, compiles now. time for cup of coffee ;-) [20:13] within 8 quad? \o/ [20:13] I can compile it in less than 1 minute with my quad xeon :p [20:15] we're running 8 barebones with (8xq6600). gcc is running on one of them (you know, security aspects ;-) so no distributed cc here ;-) [20:15] :D [20:16] a Q6600 in a barebone, wow, instant fire built in ! :D [20:16] did you compile unionfs extended attrs within the kernel? [20:17] yep [20:17] don't know if thoses features were used or not, so, as I've done it on my spare time, a sunday, I've all enable lol :) [20:17] +d [20:17] ed: the asus boxes have really good air flow. no trouble with the quads. made a performance test running them one day @ 100% [20:17] :) [20:18] ctp: I said that, as we sold thoses kind of barebones (pundit also) [20:18] they all went on fire [20:18] no exception :) [20:19] ed: ok. we're using p5g33 [20:19] was 2 years ago :p [20:19] ed: any experiences with them? [20:19] ah [20:19] nope [20:19] We now are running with nec & dell [20:19] Really, really cheap, and very good hardware with nec [20:19] much cheap on dell systems [20:20] dell is our desktop choice [20:20] now, for tiny systems, we use xdmcp on pcengines.ch alix systems :p [20:22] ah, ok. pcengines rocks ;-) [20:22] yeah :D [20:23] i've played around with wrap's ;-) [20:24] alix are better now :) 10 times more powerfull than soekris, 2 times cheaper, 2 times smaller, really good products :) [20:24] PXE is working for now on alix :) [20:25] I was frustrated with wrap that I can't PXE boot :p [20:25] ed: hm, cool. i need something new for playing around. xmas is my time ;-) [20:25] perhaps a day, I'll setup FAI for my alix systems :D [20:25] haha :) [20:26] http://rapaces.zehome.com/autophoto/view/Clariboite/dsc00024.jpg << transition between wrap / alix :) [20:26] now there is the same wall with alix on my desk at work :-) [20:28] our next project is fai for xen setups. with loadbalancing/failover on different xen clusters. really good man's toy ;-) [20:28] nice pic [20:28] ^^ [20:29] I don't work with xen for now, as I hate the performance hit it :( [20:29] but that's always interesting such technologies :p [20:29] definitely [20:30] ok, i'm out for half an hour. have to eat something ;-) [20:31] have a nice lunch [20:31] or diner, don't know how to say that in english, bref bon apetit :) [20:49] ctp (~ctp@brsg-4d0790e7.pool.mediaWays.net) left irc: Ping timeout: 480 seconds [20:58] ctp (~ctp@brsg-4d079e85.pool.mediaWays.net) joined #fai. [21:07] meandtheshel1 (~sa@85-127-137-224.dynamic.adsl-line.inode.at) left irc: Quit: Leaving. [21:12] ErKa (keryell@keryell.pck.nerim.net) joined #fai. [21:13] ctp (~ctp@brsg-4d079e85.pool.mediaWays.net) left irc: Ping timeout: 480 seconds [21:21] stockholm (~andreas@petrus.schuldei.org) left irc: Quit: leaving [21:22] ctp (~ctp@brsg-4d079ee0.pool.mediaWays.net) joined #fai. [22:04] ctp (~ctp@brsg-4d079ee0.pool.mediaWays.net) left irc: Ping timeout: 480 seconds [22:13] ctp (~ctp@brsg-4d079f93.pool.mediaWays.net) joined #fai. [22:14] ErKa (keryell@keryell.pck.nerim.net) left irc: Remote host closed the connection [22:52] kugg (~kugg@user184.77-105-210.netatonce.net) left irc: Remote host closed the connection [22:55] kugg (~kugg@user184.77-105-210.netatonce.net) joined #fai. [23:49] ctp (~ctp@brsg-4d079f93.pool.mediaWays.net) left irc: Ping timeout: 480 seconds [23:57] ctp (~ctp@brsg-4d0795ba.pool.mediaWays.net) joined #fai. [00:00] --- Tue Dec 11 2007