[00:05] simNIX (~simNIX@156-60.bbned.dsl.internl.net) left irc: Quit: Ik ga weg [05:56] Mrfai (~lange@kueppers.informatik.uni-koeln.de) joined #fai. [06:21] torge (~torge@blfd-4db0e0e5.pool.mediaWays.net) joined #fai. [06:31] fai: 3 mt committed revision 5795 to 7 fai: added fai-chboot patch (mac address support) [07:07] verwilst (~verwilst@router.begen1.office.netnoc.eu) joined #fai. [07:13] siggg (~jean@tra78-3-82-244-85-196.fbx.proxad.net) joined #fai. [07:16] Sp4rKy_ (~Sp4rKy@rennes1.dunnewind.net) left irc: Ping timeout: 480 seconds [07:41] Sp4rKy (~Sp4rKy@rennes1.dunnewind.net) joined #fai. [08:24] <\sh> mikap, problems with lenny and klibc ipconfig + initramfs dhcping on all interfaces, and while that happens -> kernel panic [08:26] \sh: ouch, do you have dmesg output available? i received similiar bugreports but without kernel panic though i need to find a system where i can reproduce this issue [08:26] <\sh> mikap, it's inside initramfs not being able to mount root via nfs...I know where the bugger comes from, and I need to update lenny to a more newer release at some time... [08:28] \sh: shouldn't happen anyway :) is there anything we could against it? [08:37] <\sh> mikap, dunno..there are many solutions for this (pxelinux + ipappend 2/3 + bootif questioning inside initramfs/scripts/functions/configure_network() [08:37] \sh: hm, proper bootif support would be nice, right [08:40] verwilst (~verwilst@router.begen1.office.netnoc.eu) left irc: Quit: Ex-Chat [08:43] <\sh> mikap, it's in squeeze [08:43] \sh: speaking of the pxelinux part? [08:44] Kajeci (~Kajeci@pa1-84-91-3-196.netvisao.pt) joined #fai. [08:44] <\sh> mikap, pxelinux can do that with adding "IPAPPEND 2" to the pxelinux.cfg/ [08:44] <\sh> initramfs is the bugger regarding configure_network() [08:44] Kajeci (~Kajeci@pa1-84-91-3-196.netvisao.pt) left irc: [08:44] Kajeci (~Kajeci@pa1-84-91-3-196.netvisao.pt) joined #fai. [08:45] \sh: ah ok, yeah - the initramfs part is the one i'd like to take care of [08:48] <\sh> http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=535008 + http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=577149 (<- bugfix of 535008) [08:50] <\sh> I added this to configure_networking() but still crashing [08:51] hm [08:54] <\sh> giving it another try [08:55] \sh: which initramfs-tools + live-initramfs version are you using? [08:55] <\sh> and I wonder why _log_msg inside the configure_networking() function is not printing out anything on the console [08:55] <\sh> while _log_msg during beginning of init-premount does exactly the same and this works [08:56] \sh: does dropping from "quiet" (if present) or adding debug=1 help? [08:57] <\sh> I don't have quiet as kernel param ;) I'll give debug=1 a try...let's see [09:03] <\sh> mikap, http://www.ubuntu-server.eu/ipconfig_boot_fail_initramfs+bootif.png [09:04] \sh: thx, what's dpkg --list initramfs-tools live-initramfs? [09:05] <\sh> ebian-lenny-fai:/# dpkg -l|grep initramfs [09:05] <\sh> ii initramfs-tools 0.92o tools for generating an initramfs [09:05] <\sh> ii libklibc 1.5.12-2 minimal libc subset for use with initramfs [09:05] <\sh> ii live-initramfs 1.156.1+1.57.1-1 Debian Live initramfs hook [09:05] <\sh> it's plain lenny boot environment [09:06] \sh: could you please give it a shot using initramfs-tools from unstable and live-initramfs from fai/grml? [09:06] <\sh> but there wouldn't be a problem if the kernel would enumerate the NICs as they are enumerated as inside the bios...but here: HP NIC1 == Kernel NIC 3 [09:06] yeah, known problem :-/ [09:09] <\sh> oh fucking hell [09:09] -vv :) [09:10] <\sh> live initramfs brings it's own configure_network [09:10] <\sh> it doesn't use the func from scripts/functions [09:10] yeah [09:10] <\sh> scripts/live holds a func which is named do_netmount [09:11] <\sh> it repeats the whole logic from functions/configure_networking() what freaking crap is that [09:15] <\sh> there are at least two ways to get around this problem [09:15] <\sh> 1. use bootif + pxelinux: IPAPPEND 2 or 3 or [09:15] <\sh> 2. add some blackmagic script to determine the first interface with a carrier from /sys/class/net/*/carrier (remove lo from the list) [09:18] current l-i supports for device in /sys/class/net/*; do already [09:48] <\sh> ok...BOOTIF works [09:48] great [09:48] <\sh> now I'm running into problems with nfsroot .. [09:49] <\sh> when you set IPAPPEND 2 it just adds the mac address to BOOTIF and tries an dhcp [09:49] <\sh> when you set IPAPPEND 3 it sets the BOOTIF and the IP cli arguments to the kernel..and then somehow nfs doesn't work anymore [09:56] coda (~coda@dhcp186-33.satec.es) joined #fai. [09:56] hi there! [09:57] excuse me guys, I'm just trying to install a new client and the installation seems to fail with permission denied while trying to mount 192.168.1.1:/srv/fai/config.. would you guys know where could be the problem? [09:59] the point is that it leaves a root shell, and if I try to mount this partition by hand, it says that rpc.statd is not running, and then if I add the '-o nolock' option it mounts it without problems :S [10:03] \sh: you'll attend developer meeting in july, right? i'd like to discuss pxe booting (what we could do to improve the situation) [10:08] <\sh> mikap, not officially but when I have the time, I would like to join you guys while I'm visiting my sister in essen :) [10:08] <\sh> it's just around the corner [10:11] <\sh> mikap, actually I want to present this: http://www.ubuntu-server.eu/faimanager-faiconfig-defaultclasses-3.ogv ;) [10:16] \sh: ah ok :) looks promising! [10:18] <\sh> mikap, auto inventarisation of servers via pxe+tftp+fai, network config (including lo,eth,bond,vlan,and bond of bonds etc.) , fai class administration, deployment state handling (not triggered async, but sync via pxe+tftp), and the tftp is not tftpd-hpa but something pythonic ;) [10:19] \sh: cool! how do you do config management? puppet? [10:20] <\sh> yes [10:20] nice [10:20] <\sh> oh yes...complete integration of puppet via fai run...from 0 to 100 in less then 5 mins (on vmware) and max. 10 mins on bare metal (when everything is already in place and you planned your deployment before ;)) [10:23] very cool [10:33] I've forgot to say that the /srv/fai/nfsroot-amd64 is being mounted without problems :$ [10:58] <\sh> we do have arch detectable nfsroots in fai? [11:08] I've just made it by updating my pxelinux.cfg file :$ [13:22] <\sh> mikap, what's a bit strange is, that after ipconfig is started, the kernel just then enables the NIC and set it to "up"...is it possible to bring up the NIC before and doing then the DHCP request via klibc ipconfig? [13:23] \sh: good question, hm [13:24] <\sh> mikap, it looks like that the dhcp request will be send out before the actual NIC is up [13:25] <\sh> or at least "up for the kernel, means modules loaded" [13:26] interesting [13:46] <\sh> that gives me some problems, even with portfast config on our cisco switches...it just takes some time...and then ip config sometimes fails ... a timeout of 360 is too much, a timeout of 120 is too low [13:56] ouch [13:58] \sh: this is so bad. [13:59] I hate all this for a long time, nw. [13:59] s/nw/now/ [14:04] excuse me guys, would you know which script is executed when mounting the config space during a new installation? [14:04] arrghh sorry by my poor english.. [14:08] <\sh> oz_, you mean I'm not alone with this problem? ;) [14:08] no [14:08] network config is *hualp* [14:08] sorry. :) [14:09] coda: if you have problems with nfs mounts, have a look at the logs of the nfs server. [14:09] <\sh> the ltsp guys from ubuntu they changed from ipconfig to udhcpc or something like this...I'm trying to catch up with stgraber about their changes...eventually it's a good thing to try it out with a better dhcpc implementation [14:09] I'm pretty sure it's a problem there [14:09] <\sh> but ipconfig is really *censored* crap [14:09] vca (~yrmthr@gibson.lawngnome.org) joined #fai. [14:10] full *censored* ack. [14:11] vca_ (~yrmthr@gibson.lawngnome.org) got netsplit. [14:11] \sh: you may want to replace pxe and tftp with gpxe and http. Or gpxe chain loading, so you can use http for getting the pxelinux.cfg configuration, instead of using tftp and python [14:12] what is gpxe? [14:12] Mrfai: thanks! the problem is that the nfsroot dir is being mounted without problems, but the config dir fails with "Permission denied", and it leaves a root shell.. then if I try again by mounting it manually with the '-o nolock' option, it works without problems :S [14:12] <\sh> Mrfai, well, I'm writing the pxelinux.cfg stuff on the fly ;) [14:13] <\sh> Mrfai, anyways...during pxe/tftp there is no problem with the network...it's just during network configuration inside initramfs [14:13] <\sh> and this is done via ipconfig and every klibc dev knows ipconfig is bad, not well documented and totally stupid ;) [14:15] hm. what's the point with this avahi-stuff? [14:16] <\sh> Mrfai, the other question is, how do I replace pxe which is enabled on server side, with gpxe? [14:20] stockholm: opensource pxe replacement, that supports http. Very cool [14:21] \sh: you can flash your NIC ROM's or you can chainload gpxe from a normal PXE bios [14:21] flashing is cool but not always possible. chainloading is easy [14:22] vca_ (~yrmthr@gibson.lawngnome.org) got lost in the net-split. [14:22] using gpxe you can create you pxelinux.cfg on the fly but then using a http request [14:22] <\sh> Mrfai, flashing bios on HP servers...is not as cool as it sounds ;) [14:24] <\sh> uh...hp smartarray p410 does raid6 and raid6+0 only with a license key...WTF! [14:25] <\sh> what comes next...HP DL385 G666 enabling cpu cores via license key? [14:26] <\sh> Mrfai, hmm...I could just take my stuff from the tftpy hack into a simple web app...that's easy...I'll give it a try when I finished up here...sounds like a cool feature [14:32] <\sh> but really, the real problem is to get rid of ipconfig inside initramfs [15:05] fai: 3 lange committed revision 5796 to 7 fai: ainsl: exit with 0 if line already exists in the file [15:23] Mrfai: i was just wondering... do you think it would be possible to install a 32-bit system from a 64-bit nfsroot? (via basefiles/) [15:28] mgoetze: not sure if this works. if the fai process enters the chroot (in /target) /proc is still 64bit and maybe some parts has to be forced to think they are 32bit [15:42] pjustice (~pjustice@yagi.h-net.msu.edu) left irc: Ping timeout: 480 seconds [15:44] pjustice (~pjustice@yagi.h-net.msu.edu) joined #fai. [15:57] pjustice (~pjustice@yagi.h-net.msu.edu) left irc: Ping timeout: 480 seconds [15:59] pjustice (~pjustice@yagi.h-net.msu.edu) joined #fai. [16:08] hallo? [16:08] torge (torge@blfd-4db0e0e5.pool.mediaWays.net) left #fai. [16:17] siggg (~jean@tra78-3-82-244-85-196.fbx.proxad.net) left irc: Quit: Leaving. [16:33] Kajeci (~Kajeci@pa1-84-91-3-196.netvisao.pt) left irc: Quit: Quit [17:12] pjustice (~pjustice@yagi.h-net.msu.edu) left irc: Ping timeout: 480 seconds [17:12] pjustice (~pjustice@yagi.h-net.msu.edu) joined #fai. [17:27] verwilst (~verwilst@d54C6A330.access.telenet.be) joined #fai. [17:31] mikap: do i have to subscribe to the mailing list to be able to send to it? or does it just take a while before a message shows up in the archive? [17:32] i sent a message, but i have no indication that it has been received: no reply, and it's not in the mailing list archive [18:30] Sp4rKy (~Sp4rKy@rennes1.dunnewind.net) left irc: Remote host closed the connection [18:39] Sp4rKy (~Sp4rKy@rennes1.dunnewind.net) joined #fai. [18:54] siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) joined #fai. [19:04] Mrfai (~lange@kueppers.informatik.uni-koeln.de) left irc: Quit: leaving [19:35] verwilst (~verwilst@d54C6A330.access.telenet.be) left irc: Quit: Ex-Chat [20:21] fai: 3 lange committed revision 5797 to 7 fai: sources.list: fix security.geo.debian.org domain entries [20:58] siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) left irc: Quit: Leaving. [22:14] ErKa (~keryell@static-64-22-43-58.albyny.csvoip.net) joined #fai. [22:17] tuv: sorry, i'm not a mailing list admin, in case of questions mail thomas (lange (at) informatik.uni-koeln.de) [23:15] mikap: no problem. i found out that i had to subscribe to the list in order to be able to send to it. usually this type of information is mentioned on the mailing list's page [23:16] tuvwx_: ack, this should be fixed [23:19] Nick change: tuvwx_ -> tuvwx [00:00] --- Wed Jun 2 2010