=== Amaranth [n=travis@ubuntu/member/amaranth] has joined #upstart === Amaranth [n=travis@ubuntu/member/amaranth] has joined #upstart === Amaranth [n=travis@ubuntu/member/amaranth] has joined #upstart === Keybuk [n=scott@quest.netsplit.com] has joined #upstart === treepio [n=pine@88-212-90-121.vl20-cph.dhcp.clearwire.dk] has joined #upstart === maro [n=mark@0x55511dab.adsl.cybercity.dk] has joined #upstart === Amaranth [n=travis@ubuntu/member/amaranth] has joined #upstart === Amaranth [n=travis@ubuntu/member/amaranth] has joined #upstart === Amaranth_ [n=travis@ubuntu/member/amaranth] has joined #upstart === wasabi__ [n=root@207.55.180.150] has joined #upstart [06:05] Keybuk: you around? [06:05] yup [06:05] what's up? [06:05] initramfs questions. Trying to boot a system, which seems to be fairly broken. At some point during the initram, the HD's get probed... repeatidly (as indicated by lights blinking on then off for 1 sec on the drives) [06:05] Trying to pinpoint it [06:06] break seems to get me there. [06:06] can I continue after breaking? [06:06] yeah, just exit the shell [06:06] hmm. gave me a panic last time I tried thta [06:06] which release? [06:10] feisty [06:10] looks like udev isn't kicking off. [06:10] it's um... the latest daily. I have this wonderful system here which has an intel d965. [06:10] So, nothing less than 2.6.18 works on it. ;0 [06:11] The problem i was having is the device nodes weren't being created before mdadm ran [06:11] I ran udevd --daemon;udevtrigger;pkill udevd manually during premount [06:11] and then exit 0, and it continued, and worked. [06:12] <_ion> There used to a problem with the run order of udev, lvm and md stuff, but i thought it was already fixed. === wasabi_ [n=administ@207.55.180.150] has joined #upstart [06:14] yay new working system [06:14] sorta. [06:18] oh [06:18] anything to do with lvm, md, evms, etc. is totally busted in feisty :p [06:19] nice. [06:19] well, got it working by hand... until i reboot again === wasabi_ sticks udevd --daemon; udevtrigger; pkill udevd into local-top/mdadm === wasabi_ snicker [06:19] doesn't it PREREQ="udev" ? [06:19] Yeah [06:20] Which doesn't exist in local-top [06:20] weird [06:20] Ahhhh. [06:20] udev is now in init-bottom and init-premount. [06:20] and not in local-top. [06:21] right [06:21] those happen first [06:21] well, premount does [06:21] how do the prereq's work though? [06:21] Aren't they only within the one script subdir [06:21] wonder if it's looping because udev doesn't exist :p [06:21] yeah [06:21] so, the prereq "udev" is unfulfillable. [06:22] ahh, yeah. [06:22] it's looping. [06:22] hence the HD noise. ;0 [06:22] right [06:22] that's the bug then === Amaranth [n=travis@ubuntu/member/amaranth] has joined #upstart === Amaranth [n=travis@ubuntu/member/amaranth] has joined #upstart [07:08] Trying to make upstart-0.3.0 but I get alot of errors with: ./configure --prefix=/tree/temp/upstart --sysconfdir=/tree/etc -->some of my errors: make[2] : ***[cfgfile.o] error 1, + alot of cfgfile errors [07:10] treepio: we'd need to see the actual errors [07:11] on another system, I hatve to type em all :) [07:11] unfortunately we can't see the other system's console, so don't know what they are unless you type them :-/ [07:12] ../nih/io.h:221: warning: 'warn_unused_result' attribute directory ignored [07:12] that's just a warning [07:13] cfgfile.c:880: error: 'RLIMIT_NICE' undeclared (first use in this function) [07:14] cfgfile.c:880: error: 'each undeclared identifier is reported only once), cfgfile.c:880: error: for each function it appears in), [07:15] cfgfile.c:880: error: 'RLIMIT_RTPRIO' undeclared (first use in this function) [07:15] missing include? [07:15] those are my 4 errors. Looks like I have made some basic mistake somewhere. alot of warnings and stuff [07:18] hmm [07:18] hmm [07:18] which version of kernel, glibc and gcc are you using? [07:19] does /usr/include/bits/resource.h exist for you? [07:20] those two limits were added in ~2.6.12 and I guess glibc 2.3.6 [07:21] resource.h exist [07:21] kernel 2.4.31, gcc 3.3.6 .. checking glibc. [07:22] ah [07:22] 2.4.31 probably won't work, if they were added in 2.6.12 [07:24] yeah, wayyyyy too old version of Linux :) [07:24] glibc: 2.3.5 [07:24] definitely too old :) [07:24] bah :).. I'll boot into another system and try :) [07:24] what distro are you using, out of curiousity? [07:24] this isslackware [07:24] which ver? [07:24] 10.0? [07:25] 10.2 [07:25] ah [07:25] upstart needs 2.6.12 or later, gcc 4, glibc 2.4 etc. [07:25] slackware still ships with 2.4 kernel?! [07:25] ok, let me see what fedora has [07:25] 11 ships with 2.6 afaik [07:25] but patrick really didn't want to have 2.6, don't ask me why :/ [07:26] upstart isn't much use without 2.6 anyway [07:26] probably because he had to do alot of work :) [07:26] the entire point is that you'd want to use it with udev [07:26] I just want to try and build a system, and decided upstart looked nice, ruby and all. [07:27] most recent version of Ubuntu, Fedora and SuSE are all suitable [07:27] Debian etch should be as well I believe [07:27] I have FC5 on the comp I'm testing with [07:28] btw Keybuk, when will development on all these fancy new upstart features start? ;) [07:28] AlexExtreme: I'm working on some of the lower level library changes at the moment [07:28] cool [07:28] I imagine I'll get to upstart next week, and probably do the bulk of them over december [07:29] sounds good [07:30] kernel: 2.6.18xxx, glibc: 2.4, gcc: 4.1.0 [07:30] also, how do you get udev to emit events upon device additions/changes? add calls to initctl in the rules files? [07:30] treepio: that should work [07:31] AlexExtreme: thanks, I will try again [07:31] AlexExtreme: yeah [07:31] I plan to patch udev to do it directly as well [07:31] ok [07:32] and one last question: will the development of the replacement initscripts take place in the respective packages themselves or in a bazaar/cvs/svn/whatever repo? [07:33] depends for which packages? [07:33] they'll certainly be shipped inside those [07:33] hmm [07:33] ok [07:33] but I suspect it'll be sensible to have a bazaar repository of them for people to use [07:33] yes, definitely [07:33] It's hard to learn the basics of linux.. [07:34] No errors this time though, thanks. [07:34] :) [07:37] can I jsut take the example files and throw in event.d and just change the disk layout? [07:39] no [07:39] the example files are for ubuntu [07:40] oo [07:40] :/ [07:41] replacing the init daemon of your system is quite a big undertaking for a novice [07:41] I have made my disk layout, fstab and inittab. now I decided to try upstart(looks nice). do I have to reqrite everything ? [07:42] Keybuk: you got it right. I am a nivice in this field.. so I decided to learn it.. Have to start somewhere [07:42] to do upstart properly, you'd need to write everything to boot your machine, yes [07:42] (in particular, upstart doesn't use inittab) [07:42] Problem is that upstart is so new, it's hard to find any info on it. [07:42] right [07:42] it's still being developed [07:42] hmm. [07:43] we're only just at the point where we're about to begin replacing init scripts with upstart jobs [07:43] Will I need to change much to have it work on my new system?. or should I just get the old unix or messy sysvinit stuff? (I will have to learn that too, so why not learn the good looking upstart) [07:44] Wouldn't it be nice to have 17 people in here instead of 16 ;) [07:45] Especially because I want to learn ruby too [07:45] if you're installing a Linux machine from scratch, without any distribution, I'd definitely start with upstart [07:45] it's just as useful as sysvinit today [07:45] Keybuk: the other inits got alot of help and howtos. Can you help me with the files I will need to change ? [07:46] For a start maby help me to find out how many and what files need changing [07:46] I'm confused a little [07:46] About what [07:46] are you taking an existing installation of Linux and attempting to modify it to use upstart [07:46] no [07:46] or are you trying to build a Linux install from scratch? [07:46] I'm using an existing distro to make a new [07:47] right, so you're modifying an existing installation? [07:47] no, I have made a seperate partition with empty dirs added fstab inittab and so. [07:47] ah I see [07:48] the first thing is you need to throw away that inittab file [07:48] instead you'd need to write /etc/event.d files to match what you originally put in there [07:48] I'm a pretty good sysadmin if I should say it myself.. but this, the basics are really hard. [07:48] Probably easy once I have made it work just once.. === AlexExtreme sits and waits while KDE SVN builds... [07:48] you'd need to write shell scripts to check and mount the disks, etc. [07:49] all the usual boot-time tasks [07:50] Keybuk: I can just change a few parameters in the example jobs right. I don't need something fancy. I just want to have it working and end up with a prompt someday.. then I can check and learn the scripts [07:50] right [07:50] the example jobs are based on the Debian / Ubuntu inittab [07:50] you'd need to tweak them to match a different distribution (init is gloriously inconsistent) [07:50] that'd make sure that the existing SysV scripts (/etc/rc*.d and /etc/init.d) are still run [07:51] I will try to look up an inittab from debian/ubuntu [07:51] Do I need those?, just removed the rc.d folder I had set up .. :) [07:51] you don't need them [07:51] but you do need what they did [07:52] if you don't want the rc.d scripts, you'd need to rewrite them as /etc/event.d jobs [07:52] Thanks, I want it real basic. don't care about compatibility . [07:52] tbh, if you wanted a REALLY quick start ... [07:52] you only need one file in /etc/event.d [07:52] :) [07:52] ----8<--------8<--------8<--------8<---- [07:52] on startup [07:52] respawn /bin/bash [07:52] ---->8-------->8-------->8-------->8---- [07:52] :p [07:52] pff :) [07:53] huh? :) [07:53] that'd give you a root shell, and keep it running [07:53] hehe [07:53] I want that. [07:53] you can just put init=/bin/bash on the kernel command line for that! [07:53] AlexExtreme: true, but then it wouldn't be upstart supervised [07:53] although, that wouldn't respawn it and it would kernel panic when the shell exits :p [07:54] oh, and you'd need console owner ;) [07:54] <_ion> ;-) [07:54] No more panics, got enough problems with my cpu fan I died up with speaker cables.. the retention bracket is broken in all 4 corners :) so I tied it up with wires.. complains about heat problems.. hehe [07:55] <_ion> treepio: Upstart isn't made in Ruby, it's C code. [07:55] _ion: cute [07:56] <_ion> Wouldn't that example also need 'console owner', btw? :-) [07:56] yes [07:56] heh [07:57] i said that ;) === _ion is blind. [07:58] rebuilding upstart to the right location dunno why I made it in /tree/temp. :) [08:00] use --prefix=usr/ --exec-prefix=/ --sysconfdir=/etc [08:00] then install with make DESTDIR=/tree install [08:00] that'll do the right thing [08:00] ok built and ready. [08:00] <_ion> /usr [08:00] likewise for things like libc, etc. [08:00] (admittedly, if I were building a Linux from scratch, I would do away with /usr ) [08:01] I did: ./configure prefix=/tree/usr sysconfdir=/tree/etc (where /tree is a mounted partition) [08:01] treepio: the problem with that is that upstart will be looking for /tree when it boots [08:01] which won't exist if you mount that partition as the root [08:01] (/tree/usr will become /usr) [08:01] auch. [08:02] ./configure --prefix=/usr --sysconfdir=/etc --exec-prefix=/ && make clean && make && make DESTDIR=/tree install [08:02] I will just rm -rf stuff again [08:02] AlexExtreme, Thanks === Amaranth [n=travis@ubuntu/member/amaranth] has joined #upstart [08:37] I will get on to installing bash, etc etc. Thanks for all your help so far. === mbiebl [n=michael@e180085091.adsl.alicedsl.de] has joined #upstart === wasabi_ [n=jhaltom@207.55.180.150] has joined #upstart === wasabi_ [n=jhaltom@207.55.180.150] has joined #upstart === theCore [n=alex@modemcable229.181-131-66.mc.videotron.ca] has joined #upstart