thpo (~thpo@dotrc.de) left irc: Ping timeout: 480 seconds
thpo (~thpo@dotrc.de) joined #fai.
fai-guy (~fai-guy@p4FC2357E.dip0.t-ipconnect.de) joined #fai.
stockholm (~andreas@marta.schuldei.org) left irc: Read error: Connection reset by peer
stockholm (~andreas@marta.schuldei.org) joined #fai.
stockholm (~andreas@marta.schuldei.org) left irc: Remote host closed the connection
stockholm (~andreas@marta.schuldei.org) joined #fai.
09:43 stockholmis it possible that fcopy copies in the config files before postgresql's postinst is run and created its own config?
09:44 stockholmbecause postgresql's postinst fails if (among other things) the path exists
09:45 stockholmit can fail for other reasons, too
09:48 h01gerno. fcopy is run after apt/aptitude
11:14 fai-guybut i think you can run fcopy in a hook before apt/aptitude
fai-guy (~fai-guy@p4FC2357E.dip0.t-ipconnect.de) left irc: Quit: SIGTERM
12:04 stockholmi understand the problem now, but i dont know how to solve it
12:05 stockholmpostgresql's config files are in the SPOTIFY class
12:06 stockholmbut the package that depends on postgresql is now not installed at INSTALL time but during the softupdate that follows at first boot
12:07 stockholmfor that reason postgres is not installed during INSTALL either
12:07 stockholmbut only during the softupdate
12:08 stockholmbut that that time postgresqls config files are there already and postgresqls postinst exists and postgresql cant be configured and started
12:08 stockholmhow to solve this?
12:08 stockholmit would be best if postgresql was installed at INSTALL time
12:09 stockholmbut for that to happen i would need a POSTGRESQL class
12:09 stockholmthat is not a problem, i could create one.
12:10 stockholmbut i dont want to manually keep track of what packages and classes need to be defined in order to define the POSTGRESQL class
12:10 stockholman automatic dependency oriented class definition would be welcome
12:11 juri_put a script in your system that prints a bright ugly warning when POSTGRESQL is defined, but not its dependancies.
12:11 h01gerthere might be a bug with patch about class dependencies
12:11 juri_that would be a start.
12:11 stockholmh01ger: do you have a hint what to look for in the bts?
12:12 h01gerclass+depend
12:12 stockholmwhy was THAT patch not accepted? :-)
12:13 h01gersee the archived bugs too (or where wontfix is)
12:13 h01gerstockholm, "_might_"
12:13 stockholmyeah
12:13 stockholmthat would be nice...
12:14 stockholmwhat package would that be filed against?
12:18 h01gersrc:fai
12:18 stockholmhttp://bugs.debian.org/cgi-bin/bugreport.cgi?bug=353536
12:18 stockholmsome holger character filed that one
12:19 stockholmbut that is not it yet
12:20 h01gerna, its wontfix, nobody needs dependencies in classes..
12:20 stockholmh01ger: what that what you meant?
12:20 h01gerthose would actually simplify my configspace, esp. the class/ subdir
12:21 stockholmits the only bug that google finds
12:21 h01gerstockholm, it should be fairly straightforward to implement #353536 in a hook. but as thomas said, you can work around that by designing the classes carefully
12:21 h01geryes, thats the one
12:22 h01gerthe patch is left out as an excercize to the reader :-D
12:22 stockholmso how could i design the classes carefully so that my above problem would not occure and i remain modular?
12:23 h01geryou cant. then you need dependencies
12:23 stockholmgahgahgah
Action: stockholm bites the table
12:24 stockholmhttp://bugs.debian.org/cgi-bin/bugreport.cgi?bug=468352
12:24 stockholmmore of the same
12:25 h01geractually you can implement #353536 even without a hook but with a .sh file in class: ifclass FOO && [ -f $FAI_CONFIGDIR_PATH/classes/FOO ] && cat $FAI_CONFIGDIR_PATH/classes/FOO
12:26 h01gernow you just need to make it generic, for all classes and be done. thats probably done by a "for i in $FAI_CLASSES" loop. (and then replacing FOO with $i in the above line)
leehartley (~lyric@209.182.184.2) joined #fai.
12:26 stockholmi dont understand that yet.
leehartley (~lyric@209.182.184.2) left irc:
leehartley (~lyric@209.182.184.2) joined #fai.
12:27 h01gerstockholm, forget it then, the other bug has a patch
12:27 h01gerand got marked "wontfix" thru my forcemerge :-(
12:28 stockholmyes, i read it just now
12:28 stockholmi would want to define classes based on package dependencies
12:28 h01geri wouldnt
12:29 stockholmif an installed package depends on postgesql, POSTGRESQL is defined
12:29 stockholmwhy not?
12:30 h01geri agree with thomas that one should plan a bit. but i'm in favor of being able to define dependencies for classes. probably i would also be in favor of your stuff, but i wouldnt recommend to use it
12:30 h01gerthe instaled package comes from a class too
12:30 h01gerso there you choose
12:31 h01gerand there you should better choose FUBARCLASS depends on POSTGRESQL than letting _-package_depends handle a _class_-depends
12:31 stockholmbut isnt that recreating dependency information manually?
12:32 h01gerno
12:32 h01gerthats not needed. apt does that just fine :)
12:33 h01geryou want more, but not really based on apts criteria. so you shouldnt do that. i might become painful. it might also work sometimes.
12:33 stockholmbut you do just that if you let FUBARCLASS depend on POSTGRESQL
12:33 h01gers/more/something else/
12:33 h01gerstockholm, you only need POSTGRESQL because the normal apt mechanisms dont work
12:33 h01gerotherwise you could just put it into FUBARCLASS
12:34 stockholmright
12:35 h01ger#468352 really looks good
12:35 stockholmyes, i like it
12:36 stockholmits not in the config space, unfortunatly
12:36 stockholmbut what the hack
12:42 stockholmshell scares me
pwr_sneak (~janp@e177235132.adsl.alicedsl.de) joined #fai.
13:34 pwr_sneakhi, hab ein problem mit setup-storage: "can't locate Linux/LVM.pm in @inc", fai-server 3.2.8 auf etch. brandneues nfsroot (auch etch)
13:35 pwr_sneakhat das schonmal jemand gesehen?
13:36 pwr_sneakmy problem during install: "can't locate Linux/LVM.pm in @inc", fai-server 3.2.8 on etch, brand-new nfsroot (also etch)
13:37 pwr_sneakright after : "partitioning local harddisks using setup-storage"
13:37 pwr_sneakdoes someone knows this problem?
13:38 h01geryou miss a perl library
13:38 pwr_sneakok, and why is that?
13:38 pwr_sneakand do i have to install it in nfsroot?
13:39 h01geryes
13:39 h01gerand because you didnt install the recommends of fai-client, for whatever reason
13:40 h01ger liblinux-lvm-perl it is, FWIW
13:41 pwr_sneakok, thanks
pwr_sneak (~janp@e177235132.adsl.alicedsl.de) left irc: Quit: Ex-Chat
snoop (~sagem@194.3.40.246) joined #fai.
therziger (~therziger@195.122.130.2) joined #fai.
15:32 therzigerhallihallo! ich hab schin wieder das problem, dass mein neuer fai server die nfsroot nicht mounten will. ich musste einen aktuellen Kernel bauen, damit der RAID Controller funktioniert. Ich habe unionfs in diesen gepatcht. kernel steht in der /etc/fai/NFSROOT. modul ist in /srv/fai/nfsroot/.../initrd-tools/modules eingetragen. make-fai-nfsroot hab ich gemacht. Der zu installierende Server kommt aber nur bis Begin: Mounting root file system. ... Hiiilfe!  
15:34 therzigerfai version: 3.2.8
16:31 therzigerschade. keiner aktiv ;(
17:23 h01gereartoast_, you have mail :/
therziger_ (~therziger@195.122.130.2) joined #fai.
therziger (~therziger@195.122.130.2) got netsplit.
snoop (~sagem@194.3.40.246) got netsplit.
istaz (~olivier@2.225-242-81.adsl-dyn.isp.belgacom.be) left irc: Quit: Reconnecting
snoop (~sagem@194.3.40.246) returned to #fai.
istaz_ (~olivier@2.225-242-81.adsl-dyn.isp.belgacom.be) joined #fai.
therziger (~therziger@195.122.130.2) got lost in the net-split.
allee (~ach@allee.mpe.mpg.de) left irc: Remote host closed the connection
leehartley (~lyric@209.182.184.2) left irc: Quit: leehartley
--- Fri Jul 11 2008

Generated by irclog2html.pl 2.1_h01ger1 by Jeff Waugh - find it at freshmeat.net! Customized by Holger 'h01ger' Levsen