--- Day changed

--- Log opened Sat Jul 26 00:00:02 2003
03:36 -!- paq_ [~paq@202.155.9.188] has quit [Ping timeout: 501 seconds]
03:47 -!- MikeH [~mike@62.56.2.163] has quit [Ping timeout: 501 seconds]
05:43 -!- AquaJo [~torbofh@pD9559BC1.dip.t-dialin.net] has joined #uml
06:32 -!- MikeH [~mike@62.56.2.207] has joined #uml
08:04 -!- mistral [mistral@jstevenson.plus.com] has joined #uml
09:40 -!- Dorian [~dorian@host81-132-153-10.in-addr.btopenworld.com] has joined #uml
09:40 -!- Dorian [~dorian@host81-132-153-10.in-addr.btopenworld.com] has left #uml []
09:50 -!- MikeH [~mike@62.56.2.207] has quit [Ping timeout: 488 seconds]
10:02 -!- MikeH [~mike@62.56.2.154] has joined #uml
10:15 -!- ticallion [~ticallion@213.175.160.244] has joined #uml
10:27 -!- ticallion [~ticallion@213.175.160.244] has quit [Quit: Client exiting]
10:28 -!- jdike [~jdike@216.208.38.106] has joined #uml
10:32 < AquaJo> hi jeff
10:33 < jdike> Hi Johannes
10:42 < AquaJo> how are you?
10:44 < jdike> AquaJo tired
10:45 < jdike> AquaJo: KS and OLS have worn me out
10:45 < AquaJo> KS and OLS?
10:45 < jdike> AquaJo: KS == Kernel Summit
10:45 < jdike> AquaJo: OLS == Ottwawa Linux Symposium
10:46 -!- MikeH [~mike@62.56.2.154] has quit [Ping timeout: 490 seconds]
10:46 < AquaJo> so you'd like to sleep for the next 2 Weeks?
10:46 < AquaJo> *g*
10:46 < jdike> AquaJo: yeah, that would do nicely :-)
10:50 < AquaJo> <- I'm setting up the next server, and I'm waiting for 2 others
11:28 < jdike> 2.6.0-test1 compiles
11:29 < jdike> Linux usermode 2.6.0-test1-1um #2 SMP Sat Jul 26 11:29:42 EDT 2003 i686 unknown
11:29 < jdike> it also boots
11:32 < AquaJo> nice
11:55 < jdike> it passes the kernel build test...
11:56  * jdike makes the patch
12:10 -!- AquaJo [~torbofh@pD9559BC1.dip.t-dialin.net] has quit [Ping timeout: 490 seconds]
12:13 -!- AquaJo [~torbofh@pD9E4CCA2.dip.t-dialin.net] has joined #uml
12:34 -!- caker [~null@pcp507591pcs.nash01.tn.comcast.net] has joined #uml
12:35 < jdike> Hi Chris
12:35 < caker> Hey!
12:35 < caker> had my home PC's motherboard die yesterday :(
12:36 < caker> what a hassle
12:47 < AquaJo> I thought my switch died
12:47 < AquaJo> but it was only the cable to the router which wasn't pluged in anymore
12:49 < Lathiat> heh
13:01 -!- mkli [~mkli@211.26.118.204] has joined #uml
13:39 -!- MikeH [~mike@62.56.2.149] has joined #uml
13:50 -!- MikeH [~mike@62.56.2.149] has quit [Ping timeout: 490 seconds]
13:55 -!- MikeH [~mike@62.56.3.197] has joined #uml
13:56 -!- MikeH [~mike@62.56.3.197] has quit [Read error: Connection reset by peer]
13:56 -!- MikeH [~mike@62.56.3.197] has joined #uml
13:56 < MikeH> anyone know where i can order a .ez domain (on an english site) ?
13:59 -!- mkli [~mkli@211.26.118.204] has quit [Quit: Lost terminal]
14:14 -!- wna [~jan@D2d19.pppool.de] has joined #uml
14:19 -!- AquaJo [~torbofh@pD9E4CCA2.dip.t-dialin.net] has quit [Quit: Ein Künstler schafft keine Schönheit, er führt sie uns vor Augen.]
15:06 -!- wna [~jan@D2d19.pppool.de] has quit [Read error: Connection reset by peer]
15:07 -!- cray [cray@130.238.8.202] has joined #uml
15:10 < cray> I want to automatically setup a bunch of uml nodes, say 6 different nodes with small changes. Can I automate these COW-files in any easy way? So I only need a "base rootfs" and then can generate the COW-files for the 6 nodes...
15:16 < mistik1> hmm
15:17 < mistik1> You may be after what I do with my UMs
15:17 < cray> what are you doing?
15:19 < mistik1> COW files are auto generated if they dont exist anyway so for my startup script I use a variable like this
15:19 < mistik1> fs=" ubd0=$HOME/uml/$name/rootfs.cow,/home/uml/$name/root_fs root=/dev/ubd0 \
15:19 < mistik1>         ubd1=$HOME/uml/$name/usrfs.cow,/home/uml/$name/usr_fs \
15:19 < mistik1>                 ubd2=$HOME/uml/$name/varfs.cow,/home/uml/$name/var_fs"
15:19 < mistik1> the $name variable there is the umid
15:20 < mistik1> so in the case of the rootfs I get /home/mistik1/uml/$umid/rootfs.cow
15:20 < mistik1> it is created if it dont exist
15:21 < mistik1> of couse in my case I also create the directories first if they dont exist
15:21 < cray> yes, but I want to make the changes to all nodes rootfs before hand...
15:22 < mistik1> elaborate
15:22 < cray> like, changing the /etc structure for node 5
15:22 < mistik1> um, that is my reason and prolly the main reason for using COW in the first place
15:23 < cray> But if you don't have any COW file, you don't have any changes to the "base rootfs", right?
15:24 < mistik1> what I do is send certain variables on the kernel command line and then use scripts to make changes to the /etc structure from these
15:24 < mistik1> a rootfs is changable only before you have a COW made from it
15:24 < mistik1> unless you remerge
15:25 < mistik1> I'm not sure you understand how COW works.
15:25 < mistik1> shall I explain a little?
15:25 < cray> sure..
15:25 < mistik1> ok
15:25 < mistik1> you have ROOT_FS which is whatever OS is there
15:26 < mistik1> and settings
15:26 < mistik1> say you start uml like this, linux ubd0=ROOTFS.cow,ROOT_FS
15:27 < mistik1> you can then make whatever changes to anything in ROOTFS.cow and it does NOT change the ROOT_FS
15:27 < mistik1> then say the next user comes along
15:27 < mistik1> they start uml like this, linux ubd0=MYROOTFS.cow,ROOT_FS
15:28 < mistik1> his COW is still a mirror of the original ROOT_FS
15:28 < mistik1> the changes made to ROOTFS.cow will NOT be there
15:28 < mistik1> do you understand?
15:29 < cray> yepp.
15:29 < cray> My question: if you want to make a big file available for the user ROOTFS.cow and not change the ROOT_FS, how do you do it?
15:30 < cray> the user which uses COW files ROOTFS.cow, I mean..
15:31 < mistik1> then just add the file to his COW
15:31 < cray> How can I do this before-hand, without booting uml with his COW?
15:31 < cray> Which format is COW files stored in?
15:31 < mistik1> hmm
15:31 < mistik1> same as the fs
15:32 < mistik1> I see what you're getting at but you cant do anything to the cow before its created
15:33 < mistik1> I'm not sure if the utils can create a cow file at the cli
15:33 < mistik1> i've never tried
15:33 < cray> Can't I manually create the COW file? and chose what to store there?
15:33 < cray> ok..
15:33 < mistik1> there is always a way to get things done though
15:33 < mistik1> If you have networking in your UMs  then its even easier
15:34 < mistik1> just need a little scritping
15:34 < cray> Ehh... Tell me more.
15:34 < mistik1> like adding something to the rc.local
15:34 < cray> NFS mount?
15:34 < mistik1> sure
15:34 < mistik1> dont even need NFS
15:34 < mistik1> ftp or http will do
15:35 < mistik1> lets say for example you needed to get bigfiles.tgz over to the new node and exploded somewhere
15:36 < mistik1> you could write a little script either have it run at boot or when they login that could just wget the file from somewhere and explode it ready for use
15:36 < mistik1> but only do this if a certain condition exist
15:36 < cray> Hmm.. yes..
15:36 < mistik1> such as the existence of a .file or the directory in question
15:36 < cray> not sure whether I like this solution or not.. let me think..
15:37 < mistik1> well its just an example ;-)
15:37 < cray> Yeah, and I appreciate it... gives me some ideas how it could be solved
15:37 < mistik1> that is the joy of unix, six million ways to die, choose one ;-)
15:38 < cray> What I think I would like, is to build a rootfs. Make a copy of it, modify it for a certain node. Then make a COW file of the diff between the filesystems...
15:39 < mistik1> is that practical though?
15:40 < mistik1> I mean that would include a lot of duplication right
15:40 < mistik1> what COW should prevent in the first place
15:40 < mistik1> I have another idea too
15:40 < cray> go ahead
15:40 < mistik1> how about creating a filesystem with all you need on it
15:40 < mistik1> then handle it with your startup script
15:41 < mistik1> you can easily have your UML startup script test if the user in question has already loaded his/her node
15:41 < mistik1> if not
15:42 < mistik1> then append ubd5=/path/to/my/archive_fs
15:42 < mistik1> then in the init scripts inside your nodes you could hardcode something to look for the existense of a fs on /dev/ubd5
15:43 < mistik1> or /dev/ubd/5 as it were
15:43 < mistik1> if this has a fs mount the SOB somewhere and run what would be a install script on the ubd5 filesystem
15:43 < mistik1> again six million ways to die ;)
15:44 < cray> hmm.. well.. not sure if that is applicable for me.
15:44  * cray thinks for a minute
15:44 < mistik1> what exactly are you trying to distribute to the nodes anyway?
15:45 < cray> basically network configs, configuration files for variuos programs, like apache and named
15:45 < mistik1> really
15:45 < mistik1> shit
15:45 < cray> I'm building a lab system for data communication labs...
15:46 < cray> want the students to run 16 nodes on their on machine and watch RIP die... or something like that...
15:46 < cray> They will get all of the rootfs:s for all nodes preconfigured..
15:47 < cray> And it's my job to automatically generate these rootfs, so they are ready to go.
15:47 < mistik1> well this still imho comes down to a little clever scripting
15:49 < cray> I've done this for ordinary NFS mounted linux, by changing files like in /etc for each netbooted computer...
15:49 < cray> Now I want to run UML inside these NFS mounted linux:s...
15:50 < mistik1> what means did you use with the netboots?
15:50 < mistik1> to change /etc i mean'
15:50 < cray> bash, perl
15:51 < mistik1> heh
15:51 < cray> The students will potentially be root on each machine, both physical and UML:s..
15:51 < cray> I want to do everything I can on the server, where they hopefully isn't root...
15:52 < cray> :)
15:52 < mistik1> well as far as i'm concerned your answer is still bash/perl
15:52 < cray> and use the /etc/rc.local solution?
15:52 -!- caker [~null@pcp507591pcs.nash01.tn.comcast.net] has quit [Read error: Connection reset by peer]
15:53 < cray> or something like that, maybe prior in the boot sequence...
15:53 < mistik1> you really dont even need that
15:53 < cray> ehh?
15:53 < cray> Now I'm lost again
15:53 < mistik1> if you script well enough you can pass all sorts of things to your UM via the kernel command line
15:54 < mistik1> lemme show you the trick I mean
15:54 < mistik1> say you start like this: linux [options] MYIP=192.168.1.100 MYHOST=umnode2000
15:55 < mistik1> you now would have access to those two variables in /proc/cmdline
15:55 < mistik1> so some nice cat/sed/awk/grep gets them for you
15:56 < mistik1> they mean nothing to the kernel of course
15:56 < mistik1> but its just a means of getting them into the UML's environment
15:56 < mistik1> see what I mean?
15:57 < mistik1> just make SURE NOT to use any variable that is actually a kernel command options
15:57 < cray> yes, but you still have to have a script which takes care of these, parses them, etc.
15:57 < mistik1> like mistake and use something like 'IP'
15:57 < mistik1> yes exactly
15:57 < cray> Okey...
15:58 < cray> that's what I meant 10 line above.. but.. wtf..
15:58 < mistik1> this is why I said now cat/sed/awk/grep :)
15:58 < mistik1> I got you:)
15:59 < mistik1> right now I set a few things with this last method I mentioned
15:59 < cray> ok... Thanks man, for the six million solutions.
15:59 < mistik1> btw, I'm an ideas person so I can prolly spit them out all day long 
16:00 < mistik1> no problem at all
16:17 -!- MikeH [~mike@62.56.3.197] has quit [Ping timeout: 490 seconds]
16:32 -!- MikeH [~mike@62.56.2.222] has joined #uml
16:36 -!- skimpIzu [hperes@24.175.39.66] has joined #uml
16:36 < skimpIzu> does uml make things slower ?
17:06 -!- cray [cray@130.238.8.202] has quit [Quit: ircII EPIC4-1.1.11 -- Are we there yet?]
17:06 < green> yes it does
17:28 -!- caker [~null@pcp507591pcs.nash01.tn.comcast.net] has joined #uml
17:50 -!- MikeH [~mike@62.56.2.222] has quit [Ping timeout: 501 seconds]
17:54 -!- kaber [~kaber@port-212-202-53-133.reverse.qsc.de] has joined #uml
17:55 < kaber> hi .. can someone please tell me how to setup the symbolic links in arch/um and include/asm-um with 2.6 ? make setup doesn't exist, make prepare doesn't work ..
17:56 < Lathiat> make config
17:56 < Lathiat> make modules
17:56 < Lathiat> make bzImage
17:56 < Lathiat> works fine for me>?
17:56 < kaber> with 2.6 ?
17:57 < Lathiat> yeh
17:58 < kaber> doesn't work for me .. i'm going to try a fresh tree
17:58 < Lathiat> did you make config?
17:58 < Lathiat> or menuconfig/xconfig
17:58 < kaber> oldconfig
17:59 < Lathiat> hrm
18:00 -!- MikeH [~mike@62.56.3.135] has joined #uml
18:00 < Lathiat> nwo the magic question is
18:00 < Lathiat> kaber: did you put ARCH=um
18:01 < kaber> sure ;)
18:01 < Lathiat> on all steps?
18:01 < kaber> yes 
18:02 < kaber> i'm just trying with a new tree one sec
18:03 < kaber> ok it set up some links but asm/cpufeature.h is missing
18:07 < Lathiat> argh i doubt mounted a fs
18:10 < kaber> with cpufeature.h linked to arch/cpufeature.h i get compile errors in include/linux/seqlock.h:35, spinlock_t is unknown but spinlock.h is included
18:12 < Lathiat> sounds liek something is bogus
18:12 < Lathiat> try making a new config
18:15 < Lathiat> fresh
18:16 < kaber> i just saw jeff's latest 2.6.0-test1 patch would create a new cpufeatures.h .. i'm trying that first
18:19 < kaber> looks good 
18:37 -!- caker [~null@pcp507591pcs.nash01.tn.comcast.net] has quit [Read error: No route to host]
18:45 < skimpIzu> whats the process of making an image of a system and that is bootable as non uml AND as UML ?
18:49 < Lathiat> well thats complicated
18:49 < Lathiat> theres quite a few differences
18:49 < Lathiat> ud need to use fake IDE etc
19:37 < skimpIzu> Lathiat: ???
19:42 -!- kaber [~kaber@port-212-202-53-133.reverse.qsc.de] has left #uml [Client exiting]
20:14 -!- AquaJo [~torbofh@pD9E4CCA2.dip.t-dialin.net] has joined #uml
20:53 -!- mistral [mistral@jstevenson.plus.com] has quit [Remote host closed the connection]
21:00 -!- AquaJo [~torbofh@pD9E4CCA2.dip.t-dialin.net] has quit [Quit: Jedenfalls ist es besser, ein eckiges Etwas zu sein als ein rundes Nichts - Friedrich Hebbel]
22:03 < BB> rar one raq meeting down ... I am well wasted lol
22:24 -!- Lathiat [~Lathiat@2001:388:7094:4080::7] has quit [Quit: Lost terminal]
22:24 -!- Lathiat [~Lathiat@2001:388:7094:4080::7] has joined #uml
22:45 -!- ElectricElf [~david@elf.noc.oftc.net] has quit [Ping timeout: 244 seconds]
22:52 -!- ElectricElf [~david@elf.noc.oftc.net] has joined #uml
23:24 -!- ElectricElf [~david@elf.noc.oftc.net] has quit []
23:26 -!- ElectricElf [~david@elf.noc.oftc.net] has joined #uml
--- Log closed Sun Jul 27 00:00:00 2003