[03:15] rcy (~rcy@DSL-209.90.181.46.Vancouver.primus.ca) left irc: Remote host closed the connection [03:54] rcy (~rcy@DSL-209.90.181.46.Vancouver.primus.ca) joined #fai. [06:33] MT (~MT@ppp-82-135-14-146.dynamic.mnet-online.de) joined #fai. [06:37] so now i understand what all this "red logo" stuff is about [06:37] i thought people were talking about the red box at the top of the screen during the install [06:38] i _definitely_ want an enterprise license ;) [06:40] sheldonh: or read the documentation how to get rid of it :-) [06:41] (hint: search for nocolorlogo) [06:41] have no fear, it won't survive :) [06:41] MT (~MT@ppp-82-135-14-146.dynamic.mnet-online.de) left irc: Ping timeout: 480 seconds [06:42] i must admit, it looks nice. but i hear it doesn't play nice with serial consoles [06:44] not very, no [06:59] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [07:51] i'm now writing our post-install hooks. they need to copy some files onto the install client (i know, we should really be using packages). is there a de facto standard location for resources that $FAI/scripts copy onto the client, or shall i just create $FAI/resources? [07:51] $FAI/files ? [07:56] morgen [08:02] MT: oh. are those copied in automatically? [08:02] no, but fcopy helps you doing that [08:02] well, actually you could do a recursive fcopy [08:02] hmm, maybe they even are [08:02] MT: that's good. i don't want automatic copy of my goodies [08:03] search for fcopy in the scripts/* [08:03] or use man fcopy [08:06] interesting [08:07] meandtheshell (~markus@85.127.114.102) joined #fai. [08:08] should fai-cd work on 3.2.1? I am getting kernel panic with the following error msg: /init: .: 150: Can't open /scripts/live [08:09] that is when i create my own installation cd using the fai-cd command and try to install a machine [08:10] theBro: just a wild guess: do you have live-initramfs installed (inside the chroot system which is used for building the initramfs)? [08:11] that was my guess but i just installed it on host :( (gotta try under chroot) [08:12] where is set_disk_info defined (as used by 20-hwdetect.source)? [08:12] it has to be installed on the system where 'update-initramfs ...' is invoked (which builds the initramfs used for booting the system) [08:12] sheldonh: /usr/lib/fai/subroutines-linux [08:13] shot, thanks [08:14] mikap: i see it sets a variable disklist, but doesn't export it. i'm guessing that, because my 60-menu.source is sourced, the variable will be available anyway? [08:17] cool! disk-info skips cdroms [08:21] mikap: where exactly do I need to configure the live-initramfs for it to be installed? [08:22] that's defined in /etc/fai/NFSROOT [08:22] (I think...) [08:22] theBro, it needs to be apt-getable in the nfsroot-chroot [08:24] it is mentionet in NFSROOT and koeln repo is in sources.list (under /etc/fai/apt/) [08:25] do you have to specify dependencies in package_config/*, or will dependencies get installed automatically becuase i'm using "PACKAGE aptitude"? [08:25] sheldonh: dependencies get installed automatically [08:25] wicked [08:25] Mrfai (~lange@kueppers.informatik.uni-koeln.de) joined #fai. [08:30] whereis ifclass [08:30] hmm, either subroutines or subroutines-linux [08:31] grrr, i've got to stop being an idiot, sorry [08:31] idiot - why? [08:31] well i should have thought to check there :) [08:31] just because you're asking questions? [08:31] thanks, you're very kind :) [08:32] hey, we're here because we like to help wherever we can... [08:33] still getting the same error :( [08:33] this is so going to rock. i can pretty much handle the entire configuration of our standard authentication and backup systems as part of the install :) [08:34] in fact, you should be able to handle really _the entire_ config [08:34] that's what we do here [08:34] no manual config at all [08:34] ever. [08:34] MT: well i'm a little nervous about giving our fai server root on our backup servers :) [08:35] sure, migration could be painful [08:35] MT: i suppose i'll have to go there eventually. but for now, i reckon an hour's work will shave 40 minutes off every install :) [08:35] i would need som kind of tool to add newly installed machines to supervising servics like munin [08:36] installing and configuring the munin client is obvious [08:36] but altering the munin server with each client install? [08:36] :) [08:36] MT: that may depend on your security policy... [08:36] is there a tool for that? [08:36] gpg signed mails send to the respective services at install time? [08:36] theBro, did you check whether live-initramfs is installed? [08:37] stockholm: the trick is (well, will be for us) , for the fai install to write the new install client's details to a provisioning database, and then the provisioning server takes care of updating nagios, munin, backup, etc. servers [08:37] torkel, "that" - ehm - what? [08:37] stockholm, debian-edu does that. iirc it's done by sitesummary (another package) [08:38] MT: if you can do everything automatically. [08:38] h01ger: debian-edu seems to be cool. i should check that out. [08:38] MT: it should get installed...but i'll make sure... [08:38] :-) [08:38] stockholm, yeah, i you havent done so, you should really look at debian-edu ;-) [08:39] torkel, if your security policy does not allow for fully automated (and thus reproducible) installs there might be some problem in your policy... :-) [08:39] of course, some manual checks may be necessary [08:39] but manual changes is probably not what you want... [08:42] MT: depending of the security level of the machine, we have to do some steps manually. Not on all the cluster nodes though, they are installed automatically [08:42] MT: MT live-initramfs gets mirrored but there is nothing with the name live under the initrd.img [08:43] would guess that there should be something within the initrd image [08:43] h01ger, any idea about theBro's problem? [08:44] I've not yet started using live-initramfs, so I'm afraid I can't help that much [08:47] Action: h01ger thinks live-initramfs is not installed in the nfsroot [08:47] theBro, check the output of "make-fai-nfsroot -v" [08:47] (even if it ends in "nfsroot created successfully" [08:47] ) [08:48] ack [08:50] during the retriving phase there is no live-initramfs mentioned (script is still running though) [08:50] using 3.2.1? [08:50] h01ger: yep [08:51] also inside the nfsroot? ;) [08:51] how do i make sure? [08:52] do I need to include koeln repository in make-fai-nfsroot.conf? [08:53] I currently have: FAI_DEBOOTSTRAP="etch http://localmirror/debian" [08:56] i think its enough if you have it in /etc/fai/apt/sources.list but i'm not 100% sure [08:56] Mrfai, ^ [08:57] i have it in that one and not working :( [09:01] someone has forgotten to run make-fai-nfsroot after updateing the sources.list [09:01] Action: Mrfai gruesst aus Zuerich [09:03] Mrfai, gruezi :) [09:03] theBro, and now it works? [09:05] h01ger: yep, seems to be starting the installation [09:05] thanks a bunch [09:18] can dialog(1) put multiple widgets (a checklist and a menu) in a single dialog? i know it can chain multiple dialogs, but i'd like everything in one if possible [09:22] sanso|work (~sanso@194.120.17.62) joined #fai. [09:22] sheldonh: no, if you need that take a look at STFL [09:23] ooooo [09:23] schweet [09:25] sanso|work_ (~sanso@194.120.17.62) joined #fai. [09:25] sanso|work (~sanso@194.120.17.62) left irc: Remote host closed the connection [09:34] siretart (siretart@213.239.237.3) got netsplit. [09:34] \sh_away (~shermann@server3.servereyes.de) got netsplit. [09:38] sanso|work_ (~sanso@194.120.17.62) left irc: Remote host closed the connection [09:38] sanso|work_ (~sanso@194.120.17.62) joined #fai. [09:40] siretart (siretart@213.239.237.3) returned to #fai. [09:44] mfl_ (~mfl@apoll.fnb.maschinenbau.tu-darmstadt.de) joined #fai. [09:44] mfl (~mfl@130.83.226.11) left irc: Read error: Connection reset by peer [09:45] \sh_away (~shermann@server3.servereyes.de) got lost in the net-split. [10:12] can i just exit with nonzero status in a class source ($FAI/class/60-menu.source) to abort the install, or must i call a magic helper [10:12] ? [10:13] define an EXIT class and hook all tasks away if that class is defined [10:16] ummm... [10:16] "hook all tasks away"? [10:16] hmm, I asked for such an enhancement (make it configurable if FAI should continue with an an install if a script exists with nonzero or not)... but I think I didn't file a wishlist bug yet [10:17] sheldon use the function skiptask TASKNAME [10:17] it's a workaround for the thing you want [10:17] currently, FAI just runs everything to the end as far as it can, whatever error occurs on the way [10:17] some people want that, and it's useful in some cases [10:19] sheldonh: but even with this skiptask, the scripts would all run through, I assume - this would only skip all the tasks _after_ the configure task [10:21] Action: h01ger wouldnt call it a workaround :) [10:23] what else [10:23] normal fai usage [10:24] he asked for a _solution_ of something he wants to... [10:24] and his proble is currently not solveable, but a worakround is possible [10:24] yeah. its a feature of fai that you can use hooks for everything. so its no workaround, but normal usage [10:24] from the point of view of his requirements, this is a workaround. [10:26] i guess i could print a message, read a line of input and then reboot -f [10:27] Mrfai: gruesse nach zuerich! [10:28] Mrfai: fai 3.2.1 ist in ubuntu nun. die letzte mail von mir haette eigentlich nicht an die listen gehen sollen, sorry fuer den spam [10:29] sheldonh: there's also the command faireboot, but I don't know excatly, in what it differs to reboot, and if it's good to call it in the middle of the configure task [10:29] only problem with my reboot approach is a log won't be saved back to the fai server [10:29] and yes, that's the way how you can skip all following scripts to run and any task [10:29] sheldonh: call savelog before reboot! [10:29] I'd like to include scripts/config from a class in another class; what's the proposed way to do that? is symlinking supported? so .../fai/config/scripts/GRML/32-xorg is a symlink to .../fai/config/scripts/GRML_X/32-xorg? [10:30] lazyb0y: aha! looks like faireboot calls fai-savelog if called with the -s option :) [10:30] mikap, better define that class if the other class is defined [10:31] mikap: hmm, not exactyl, theres no such thing as inheritance [10:31] mikap: what's the difference between the two classes? [10:31] mikap: if it's packages, then you can define additional class dependencies in a package config [10:33] and in scripts, you can also to "ifclass X", and do additionall stuff for the X class in your script [10:34] h01ger: lazyb0y: this sounds good, are there some recommended examples available to look at? [10:34] mikap, just use ifclass in a /class-script [10:34] mikap: in the simple examples [10:35] "ifclass FOO && echo BAR" [10:35] my problem is, that I've a base system, several different feature-sets and want to use several scripts depending of their "inheritance" [10:35] h01ger: thanks! [10:35] grep for ifclass, and look at package definitions that start with lines like "PACKAGES install CLASSNAME" instead of only "PACKAGES install" [10:35] lazyb0y: thanks, found it [10:36] excellent, I'll play with that [10:36] my example is to be used in fai-config-dir/class/50-host-classes or such... [10:36] great [10:49] grrr, stupid dialog's saving the selected class names with quotes around them [10:54] 6 minute installs are nice :) [10:56] skiold (~skiold@84.121.91.161.dyn.user.ono.com) joined #fai. [11:07] where can i find a list of hooks? i expected to find it in "7.11 Hooks". i want to scripts that are run after all packages are installed [11:09] ah i see... i could do 10-domagic.LAST, yes? [11:10] hmm, no [11:10] 10-domagic is not a task name [11:11] "after all packages" is before the default activity of configure [11:12] so you should instead do configure.LAST [11:12] okay. is there a list of these activities? [11:12] (well, I'd name it configure.DEFAULT) [11:12] I'm not sure whether there is one in the documentation [11:12] Mrfai? [11:12] ah, i see. i actually want configure.CLUE_STANDARD then [11:12] better look in lib/subroutines [11:13] search for task_install [11:13] the heart of it all :) [11:18] and if my configure.FOO hook wants $disklist, it should source subroutines-linux, yes? or is /usr/lib/fai in the path, in which case i can just call disk-info myself [11:23] all tasks are listed in the documentation [11:24] See chapter 7.2 in the FAI Guide [11:24] sanso|work_ (~sanso@194.120.17.62) left irc: Remote host closed the connection [11:24] sanso|work_ (~sanso@194.120.17.62) joined #fai. [11:29] torkel: thanks. and can my hooks trust that /usr/lib/fai is in the PATH? [11:35] I think so yes [11:36] sheldonh: If you haven't done it yet, I really suggest that you read the FAI Guide. It will give you a good understanding about FAI, and also some tips and tricks [11:45] sheldonh: variables are available in all hooks/scripts. HAve a look at /tmp/fai/variables.log [11:46] sanso|work__ (~sanso@194.120.17.62) joined #fai. [11:46] sheldonh: grep PATH /usr/sbin/fai [11:50] sanso|work_ (~sanso@194.120.17.62) left irc: Remote host closed the connection [11:54] asw (~asw@karuna.med.harvard.edu) left irc: Quit: Leaving [11:56] ErKa (keryell@m7.wifi.enstb.org) joined #fai. [11:58] siretart: is PPA only building i386 packages? I'm missing live-initramfs for dapper/amd64 [12:01] and the build records says nothing about amd64 only i386 :-( [12:04] skiold (~skiold@84.121.91.161.dyn.user.ono.com) left irc: Quit: skiold [12:04] ah now is it there [12:17] asw (~asw@karuna.med.harvard.edu) joined #fai. [12:20] torkel: live-initramfs is now arch all [12:32] torkel: right, live-initramfs is arch all. [12:33] torkel: PPA is btw building on both i386 and amd64. - that's because the builders are using xen, and xen doesn't work on anything but those 2 architectures [12:34] btw, did I mention that every member in the FAI team is able to upload packages to that ppa? [12:42] Mrfai, siretart: Yeah I noticed. My brain is a bit slow [12:44] siretart: You did. Though I'm not sure I trust my packaging skills enough to upload packages (yet) [12:46] sanso|work__ (~sanso@194.120.17.62) left irc: Remote host closed the connection [12:46] sanso|work__ (~sanso@194.120.17.62) joined #fai. [12:52] Mrfai: i keep reading, getting excited, having a look at something, and then getting lost because i should have read further :) [12:53] http://paste.grml.org/58 => huh? does ifclass work different than I expect it to? [12:54] gnaaaa, forget it :-/ fscking typo from myself ;( [12:56] if ifclass REMOVE_DOCS: ; then [12:56] :) [12:56] h01ger: yeah, what a stupid typo from the code-rework ;-( [12:56] skiold (~skiold@84.121.91.161.dyn.user.ono.com) joined #fai. [13:16] can i use ifclass in class/*.var files ? [13:18] i-R2D258 (~deed@tor-irc.dnsbl.oftc.net) left irc: Remote host closed the connection [13:19] i-R2D258 (~deed@tor-irc.dnsbl.oftc.net) joined #fai. [13:35] bleh. i've forgotten how setup_harddisks decides which class to use, and it's not documented in 7.7. pointer? [13:38] looks like it works through them in reverse order of definition [13:45] Robinayor_ (~robin@suw.rkara.de) left irc: Quit: going away [13:47] i-R2D258 (~deed@tor-irc.dnsbl.oftc.net) left irc: Remote host closed the connection [13:52] while running fai, the variables seem to be sourced with '-x', this means that I can see things like to rootpw hash [13:52] has anyone else noticed that? how can I prevent that? [13:59] sanso|work__ (~sanso@194.120.17.62) left irc: Remote host closed the connection [13:59] sanso|work__ (~sanso@194.120.17.62) joined #fai. [14:03] mfl (~mfl@apoll.fnb.maschinenbau.tu-darmstadt.de) joined #fai. [14:03] mfl_ (~mfl@apoll.fnb.maschinenbau.tu-darmstadt.de) left irc: Read error: Connection reset by peer [14:08] /srv/fai/config gets mounted on the install client as /fai, right? [14:09] siretart: is it worth trying to fix? surely anyone who can see it could have seen it anyway, given physical access? [14:12] sheldonh: yes, it is rather a cosmetical issue [14:21] phew, this is mangling my head. i can't find where fai chroots into $target. i want to see where it happens, because i'm trying to find out whether i have access to anything from the fai server or the nfsroot from inside the $target chroot [14:44] rgrep chroot * [14:45] looks ot me like /srv/fai/config is _not_ mounted as /fai by the install client. my prepareapt hook couldn't find files in it [15:32] Action: h01ger found a new bug in fai. bashism. [15:33] sanso|work__ (~sanso@194.120.17.62) left irc: Remote host closed the connection [15:34] sanso|work__ (~sanso@194.120.17.62) joined #fai. [15:34] sanso|work__ (~sanso@194.120.17.62) left irc: Remote host closed the connection [15:34] sanso|work__ (~sanso@194.120.17.62) joined #fai. [15:56] does fai have hooks for including my own package signing keys into the fai nfsroot and subsequently into the installed client? [16:01] sheldonh, yes, new in 3.2.1 [16:02] damn [16:05] h01ger: so for now i need an updatebase hook? i've been trying (with http://rafb.net/p/Vuh25n99.html ), but i keep getting gpg NO_PUBKEY errors for the supposedly installed key at apt-get update time [16:07] yes, i think so [16:09] skiold (~skiold@84.121.91.161.dyn.user.ono.com) left irc: Quit: skiold [16:17] meandtheshell (~markus@85.127.114.102) left irc: Quit: Leaving. [16:17] ha. updatebase was the right task to hook, prepareapt was the wrong one :) [16:23] wow! [16:23] i wish i'd heard about debian-edu-config earlier. it's got its own support for keeping etc under revision control with svk! [16:29] :) [16:36] fai is the shit [16:36] interestingly, a colleague was reading The Debian System the day before i started working on automated installs, and suggested i look at it. if he hadn't, i'd be busy with preseeding right now [16:37] madduck used to hang out here [16:39] is it possible to execute task_configure + later tasks only and not redo dirinstall completely btw? [16:39] fai -N softupdate [16:39] cool, thanks :) [16:39] np, yw ;) [16:40] cp: cannot create fifo `//var/log/fai/fuchsiatard/noaction-/logfifo': File exists - hm [16:40] ErKa (keryell@m7.wifi.enstb.org) left irc: Ping timeout: 480 seconds [16:41] when deleting the file (looks like on error the file isn't removed) I get: [16:41] Error: Provide the URL to obtain the fai config storage in $FAI_CONFIG_SRC - hm [16:41] what should that be? :) I'm using dirinstall only, no NFS stuff anywhere ;) [16:42] then you need to define $FAI_CONFIG_SRC in /e/f/fai.conf [16:42] where should it point to? never defined that before :) [16:42] nfs://server/path [16:43] or svn://svn.hostname.tld/path [16:43] or svn+ssh... [16:43] or git ;) [16:43] I use none of these ;) [16:43] what do you use? [16:43] I never used that so far, so I've no clue what this should be ;) [16:43] only dirinstall feature :) [16:43] ah [16:44] "normally" everybody starts with nfs... [16:44] use svn or git [16:44] for which task? [16:45] get_config(dir?) [16:46] i've never used nfs. :) [16:46] hm? I mean, the variable is described as "This variables described how to access the configuration space on the install clients." in the docs, but I don't have any "install clients" [16:46] mikap, svn co svn://svn.debian.org/svn/d-community/trunk/fai-config-dir ; cd fai-config-dir/doc ; less manual_adminwork.txt # read step 6 there [16:46] mikap, the machine you run fai on, is the install client [16:46] h01ger: ahhh, thx - I'll check out :) [16:48] instead of svn://svn.debian.org/svn/d-community/trunk/fai-config-dir it should point to my localhost now? [16:48] MT, re: 443198: maybe,probably,hopefully ;) [16:49] mikap, yes, to your svn repo [16:49] h01ger: hm ok, didn't want to use one but ok :) [16:49] h01ger, I think that trap code was developed during the first FAI workshop [16:49] but I've no idea what that logical OR is good for... [16:49] mikap, its trivial to write /usr/lib/fai/get-config-dir-file and file a wishlist bug, so that Mrfai includes it in the next release ;) [16:49] hmm, maybe eartoast knows about it [16:50] I'd guess it was him :-) [16:50] hehe [16:50] h01ger: hm ok, I think I've to take a closer look at that ;) [16:51] though I don't really understand yet why's that needed at all, hm (why not just skip the steps I don't need again?) [16:52] you need to access the config space somehow [17:18] i-R2D2 (~deed@tor-irc.dnsbl.oftc.net) joined #fai. [18:00] hm, I don't want to set up a full subversion server just to be able to skip the bootstrapping phase; isn't there a simple way how to skip the software installation? :-/ [18:06] commenting out the relevant functions inside task_dirinstall() seems to do the trick :) [18:07] mikap, use svn+ssh ;) [18:09] h01ger: that would be my fallback-solution ;) I don't want to have a dependency for a working subversion server, especially because I currently don't even know what would be the gain [18:09] rcy (~rcy@DSL-209.90.181.46.Vancouver.primus.ca) left irc: Remote host closed the connection [18:26] meandtheshell (~markus@85.127.114.102) joined #fai. [18:29] meandtheshell (markus@85.127.114.102) left #fai. [18:33] rcy (~rcy@DSL-209.90.181.46.Vancouver.primus.ca) joined #fai. [18:33] MT (~MT@dove.informatik.tu-muenchen.de) left irc: Read error: Connection reset by peer [18:33] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [18:41] oh, siretart away :( [19:02] - [ -f /etc/init.d/udev ] && umount $FAI_ROOT/dev [19:02] + [ -d $FAI_ROOT/dev/.udev ] && umount $FAI_ROOT/dev [19:03] => for /usr/sbin/fai, does this sound reasonable for you? [19:06] mikap, why do you propose this change? [19:06] sanso|work__ (~sanso@194.120.17.62) left irc: Remote host closed the connection [19:06] sanso|work__ (~sanso@194.120.17.62) joined #fai. [19:09] MT: using this plus http://grml.org/tmp/fai_patch_for_dirinstall2.diff (that's a hack for now, I'm aware of this) would allow to re-execute dirinstall without headaches [19:10] what do you mean be re-execute? [19:10] MT: I have a working chroot and just want to execute the class scripts without rebuilding the whole chroot [19:10] ah, ok [19:11] well, if this is the case -- file a bug report please :-) [19:11] I'll do some more testing before, but will do that for sure :) [19:12] I think at the time this code was introduced udev was still pretty new and not used by everyone [19:12] that is, it was before etch :-) [19:12] hehe, yes looks like this ;) [20:20] ErKa (keryell@keryell.pck.nerim.net) joined #fai. [20:40] sanso|work__ (~sanso@194.120.17.62) left irc: Remote host closed the connection [20:41] sanso|work__ (~sanso@194.120.17.62) joined #fai. [20:55] sanso|work__ (~sanso@194.120.17.62) left irc: Remote host closed the connection [20:55] sanso|work__ (~sanso@194.120.17.62) joined #fai. [21:10] Mhm on client: mount lists tmpfs on /cow type tmpfs (rw). But there's no /cow dir in / on the client [21:10] other way round: anyone tried instaling gutsy client from a gutsy server? ;) (fai 3.2.1-0ubuntu1) [21:29] MT (~MT@dove.informatik.tu-muenchen.de) left irc: Ping timeout: 480 seconds [21:34] sanso|work_ (~sanso@194.120.17.62) joined #fai. [21:34] sanso|work__ (~sanso@194.120.17.62) left irc: Remote host closed the connection [23:01] ErKa (keryell@keryell.pck.nerim.net) left irc: Ping timeout: 480 seconds [23:24] allee: not yet [23:27] allee: most probably because live-initramfs hasn't been synced yet [00:00] --- Thu Sep 20 2007