[07:32] MT (~MT@ppp-82-135-67-120.dynamic.mnet-online.de) joined #fai. [07:35] fai-guy (~fai-guy@p57B9BC3A.dip0.t-ipconnect.de) joined #fai. [07:57] Mrfai (~lange@kueppers.informatik.uni-koeln.de) joined #fai. [08:00] MT (~MT@ppp-82-135-67-120.dynamic.mnet-online.de) left irc: Ping timeout: 480 seconds [08:05] torkel_ (torkel@monsun.hpc2n.umu.se) left irc: Quit: leaving [08:05] torkel (torkel@ip64.degernas.se) joined #fai. [08:07] torkel (torkel@ip64.degernas.se) left irc: [08:07] torkel (torkel@ip64.degernas.se) joined #fai. [08:08] torkel (torkel@ip64.degernas.se) left irc: [08:10] torkel (torkel@ip64.degernas.se) joined #fai. [08:13] torkel (torkel@ip64.degernas.se) left irc: [08:13] torkel (torkel@ip64.degernas.se) joined #fai. [08:24] torkel (torkel@ip64.degernas.se) left irc: Quit: leaving [08:24] torkel (torkel@ip64.degernas.se) joined #fai. [08:36] itais (~itais@62.81.119.138.static.user.ono.com) joined #fai. [08:47] ErKa (~keryell@keryell.pck.nerim.net) joined #fai. [08:59] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [09:54] Mrfai (~lange@kueppers.informatik.uni-koeln.de) left irc: Quit: leaving [09:54] Mrfai (~lange@suenner.informatik.uni-koeln.de) joined #fai. [10:12] MT (~MT@dove.informatik.tu-muenchen.de) left irc: Ping timeout: 480 seconds [11:01] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [11:22] allee (~ach@allee.mpe.mpg.de) left irc: Read error: Connection reset by peer [11:23] allee (~ach@allee.mpe.mpg.de) joined #fai. [11:36] crg (crg@lagoon.freebsd.lublin.pl) left irc: Read error: No route to host [11:36] crg (crg@lagoon.freebsd.lublin.pl) joined #fai. [11:38] maswan (maswan@kennedy.acc.umu.se) got netsplit. [11:38] oz_ (~oz@anubis.coresystems.de) got netsplit. [11:38] maswan (maswan@kennedy.acc.umu.se) returned to #fai. [11:38] oz_ (~oz@anubis.coresystems.de) returned to #fai. [12:18] meandtheshell (~sa@85-127-137-224.dynamic.adsl-line.inode.at) joined #fai. [12:18] meandtheshell (~sa@85-127-137-224.dynamic.adsl-line.inode.at) left irc: Read error: Connection reset by peer [12:19] meandtheshell (~sa@85-127-137-224.dynamic.adsl-line.inode.at) joined #fai. [12:19] meandtheshell (~sa@85-127-137-224.dynamic.adsl-line.inode.at) left irc: Read error: Connection reset by peer [12:31] ErKa (~keryell@keryell.pck.nerim.net) left irc: Ping timeout: 480 seconds [13:11] ErKa (~keryell@keryell.pck.nerim.net) joined #fai. [13:16] nocturn (~nocturn@d51A4378A.access.telenet.be) joined #fai. [13:17] Hi guys [13:17] I'm trying to deploy FAI for a BEOWULF cluster (on Debian Etch) [13:17] But I have two kinds of nodes, one 64-bit, the other 32-bit. How do I handle this best with FAI? [13:22] have two classes? [13:23] stockholm: With two NFSRoots? [13:23] nocturn: yes, i would think so [13:28] alexanderwz (~alexander@karuna.med.harvard.edu) got netsplit. [13:29] alexanderwz (~alexander@karuna.med.harvard.edu) returned to #fai. [13:35] How do I make the 64-bit NFS-root and kernel image on an 32-bit FAI server? [13:36] this is not possible [13:36] but you can make a 32bit nfsroot on a 64bit server [13:37] I think there's something written in the faiwiki on that topic [13:40] I can't find an article on this... [13:40] I know it can be done, though it's possible that make-nfsroot was run on an 64-bit machine over NFS [13:41] I have an existing cluster here deployed that way, though I do not know how it was set up. [13:42] itais (~itais@62.81.119.138.static.user.ono.com) left irc: Quit: See you [13:46] running make-fai-nfsroot on a 64bit machine and using a directory that is NFS mounted from a 32bit machine is also a good solution. [13:47] Thanks, we'll go that route I guess [14:04] Mrfai: yes, maxpackages does it... any reason why this isn't used by default? [14:05] IMO it takes much longer when using MAXPACKAGES=1. Can you please compare the time of calling fai-mirror with and without setting it. [14:06] Mrfai, didn't we have this setting at some point? [14:06] after all, it is fine to have confilicting packages in the mirror... [14:06] i didn't measure this run I did now, but from the feeling it's not much more - in both cases about 5-10 minutes, while the real reporistory is on the local machine [14:07] or maybe 5-15 [14:07] for a mirror that is finally about 1GB [14:08] (running fai-mirror with time now...) [14:12] BTW another thing I realized is, fai-mirror will not work well if I have an infrastructure where different systems (e.g. test and production, etch and sid) have different sources.lists [14:14] so currently this can only be solved by having multiple fai configdirs and point fai-mirror to that [14:15] I think, it's acceptable to say people need to build different cd's for systems with different sources.list's, but fai-mirror should also accept an extra sources.list as parameter. [14:15] or is that a much more compley topic? [14:18] done with the fai-mirror with maxpackages [14:18] time: 9 minutes for a mirror of 1.6 GB [14:19] I'm not going to measure the cal wihtpout again, I am sure it was not much faster - even if it only had 1.1 GB, probably because of the errors [14:32] MT (~MT@dove.informatik.tu-muenchen.de) left irc: Remote host closed the connection [14:32] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [14:44] lazyb0y: fai-mirror for the simple examples. Using my local mirror. this creates a partitial mirror of 454MB. [14:44] MAXPACKAGES=1 => 2min 26 [14:44] MAXPACKAGES=9999 => 1m22 [14:45] and if you specify -v you will get a lot more output when setting the variable to 1. [14:45] and, well, is the mirror the same size? [14:45] yep. Same size in MB, I did not checked all packages [14:46] ok, so there is some (obvious ...) overhead [14:46] but the MAXPACKAGES=1 mirror is surely complete, the other may be broken [14:46] if your internet connextion is slow(er) then it does not count, because downloading takes more time [14:48] Here, I work with a bigger mirror, and then, the difference is not as much as there [14:48] and my connecttion here is very fast, it's the same machine [14:52] measuring the normal call now [14:52] difference in packages with simple example is only one package: inetutils-ping. It's missing when using MAXP...=9999 [14:53] anyway, the questions is: is this increase in runtime worth the fact that users who run into this problem will not easily find the problem and the solution? [14:53] Mrfai: you mean, even with the simple examples your mirror is incomplete? [14:54] BTW, my guess was wrong - it takes only 2m23 [14:54] still, creating something broken fast does not really count as "better" :) [14:55] (and there are 500MB of packages missing!) [14:58] ok, I will set MAXPACKAGES=1 in fai-mirror. [15:44] nocturn (~nocturn@d51A4378A.access.telenet.be) left irc: Quit: leaving [15:44] fai-guy (~fai-guy@p57B9BC3A.dip0.t-ipconnect.de) left irc: Quit: leaving [16:17] ErKa (~keryell@keryell.pck.nerim.net) left irc: Ping timeout: 480 seconds [16:41] glance (glance@shaka.acc.umu.se) left irc: Ping timeout: 480 seconds [16:50] Mrfai (~lange@suenner.informatik.uni-koeln.de) left irc: Quit: leaving [17:36] MT (~MT@dove.informatik.tu-muenchen.de) left irc: Remote host closed the connection [17:36] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [17:50] MT (~MT@dove.informatik.tu-muenchen.de) left irc: Ping timeout: 480 seconds [18:51] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [19:03] MT (~MT@dove.informatik.tu-muenchen.de) left irc: Ping timeout: 480 seconds [19:11] fai-guy (~fai-guy@p54BD7E34.dip.t-dialin.net) joined #fai. [19:11] Warum geht install_packages nicht, wenn es chrooten soll, aber läuft, wenn ichs per Hand im chroot starte? [19:11] Das Problem scheint zu seien, dass er im Chroot keine "Debian Package Index" findet sondern nur dpkg-foo [19:12] mein nfsroot ist 32bit und das chroot amd64 [19:13] Mrfai (~lange@xdsl-87-78-72-162.netcologne.de) joined #fai. [19:13] sepski (~sep@217.17.211.51) joined #fai. [19:14] Ah der Chef ist da [19:14] Warum geht install_packages nicht, wenn es chrooten soll, aber läuft, wenn ichs per Hand im chroot starte? [19:14] Das Problem scheint zu seien, dass er im Chroot keine "Debian Package Index" findet sondern nur dpkg-foo [19:14] mein nfsroot ist 32bit und das chroot amd64 [19:16] sepski (~sep@217.17.211.51) left irc: Remote host closed the connection [19:24] Barbarossa: versuches mit der "Mrfai:" notation. [19:24] vielleicht liest er es dann [19:25] Mrfai: *stubbs* [19:27] stockholm: Hast Du schonmal mit dem AptPkg::* Kram gearbeitet? [19:31] Barbarossa: nee [19:31] wozu braucht man das? [19:31] Scheiss amd64 Karren [19:31] Das benutzt FAI um zu checken ob die Pakete, die in package_config/* angegeben wurden da sind. [19:32] Wobei ich mich frage waurm man nicht einfach apt-get aufruft und guckt obs knallt. [19:32] Dann hätte ich dieses Problem nicht [19:33] Die alles entscheidende Frage ist, warum ich keinen 'Debian Package Index' habe [19:33] apt-get im chroot ausführen? [19:33] apt-get update [19:33] Das macht das Ding ja [19:33] Nur leider funktioniert das nicht [19:34] Nehmen wir mal die Checks raus. [19:34] Die Pakete gibts ja schliesslich [19:34] :-) [19:35] schaunwamal [19:36] verstehe das Problem nicht. Package Index klingt nach Packages.gz. Vielleicht ist die sources.list falsch. [19:36] nein [19:36] Wenn ich per Hand nen chroot mache und dann apt-get isntall list_der_pakete mache tut alles [19:37] in install_packages guckst Du ja im apt-cache nach ob das Pakete da ist. [19:37] kannst du ins generierte chroot reinchrooten und da apt-get machen? [19:37] ja [19:37] sag ich ja [19:37] Und Du machst sowas: return 1 if ($fitem->{File}{IndexType} eq "Debian Package Index"); [19:37] Allerdings steht da bei nicht Debian Package Index dron sondern was mit dpkg [19:37] wenn du aptitude statt apt-get nimmst (im package_config) dann wird nicht ueberprueft ob das Paket existiert. [19:38] Ich habs jett auskommentiert... [19:38] Barbarossa: nimm einfach aptitude statt install hinter PACKAGES [19:38] in der package_config? [19:38] ok das ist natürlich noch einfacher [19:38] probier ich gleich mal [19:40] ok nen 64bit kernel wäre jetzt noch gut gewesen ;) [19:46] Warum hat der lilo nicht installiert? *wunder* [19:55] Lin_ (~igor@200.179.57.57) joined #fai. [19:56] Lin_ (~igor@200.179.57.57) left irc: Remote host closed the connection [20:01] logfiles lesen :-) [20:03] Jo. Is schon doof, wenn man nen '}' zuviel auskomentiert... [20:03] fai-guy_ (~fai-guy@p54BD7DE3.dip.t-dialin.net) joined #fai. [20:06] Hmm. Ne Idee, warum ich nach der Installation kein Netz mehr bekomme? [20:07] komische hardware [20:07] gibt viele Gruende. [20:07] da sind eth0 und eth1 vertauscht :-( [20:07] firewire? kann zu eth0 werden [20:08] ne is nen Dell-Blade [20:08] eth0 und eth1 sind ja da. Nur falsch rum ... [20:08] ahh, ätzend. [20:08] Ich *HASSE* udev [20:08] Barbarossa: am einfachsten ist umstöpseln [20:08] liegt daran das der kernel eine andere reihenfolge hat als PXE boot. [20:08] stockholm: geht nicht. [20:08] Das ist nen Blade. Das kann ich einfahc nur 'reinstecken' [20:08] Barbarossa: kannst du eine im bios abstellen? [20:09] Will ich nicht. [20:09] hilft nur bei PXE, der kernel wird imme rbeide sehen AFAIK [20:09] kannst du das netzwerkbooten bei der einen abstellen? [20:09] beim booten ists ja richtig [20:09] ah [20:09] FAI nimmt auch die richtige [20:09] Die anderen ist nichtmal konfiguriet [20:09] fai-guy (~fai-guy@p54BD7E34.dip.t-dialin.net) left irc: Ping timeout: 480 seconds [20:11] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [20:35] Lin (~igor@200.179.57.57) left irc: Quit: Ex-Chat [20:45] fai-guy_ (~fai-guy@p54BD7DE3.dip.t-dialin.net) left irc: Quit: leaving [21:12] Mrfai (~lange@xdsl-87-78-72-162.netcologne.de) left irc: Quit: leaving [21:29] Nick change: MT -> Guest273 [21:29] MT (~MT@ppp-82-135-6-74.dynamic.mnet-online.de) joined #fai. [21:30] glance (glance@shaka.acc.umu.se) joined #fai. [21:35] Guest273 (~MT@dove.informatik.tu-muenchen.de) left irc: Ping timeout: 480 seconds [21:45] MT (~MT@ppp-82-135-6-74.dynamic.mnet-online.de) left irc: Ping timeout: 480 seconds [21:55] ErKa (keryell@keryell.pck.nerim.net) joined #fai. [23:00] MT (~MT@dove.informatik.tu-muenchen.de) joined #fai. [23:45] ErKa (keryell@keryell.pck.nerim.net) left irc: Ping timeout: 480 seconds [00:00] --- Wed Dec 5 2007