--- Day changed

--- Log opened Thu Jul 10 00:00:04 2003
00:03 -!- Pahan [~pahan@ca-manbch-cuda3-c2a-a-52.stmnca.adelphia.net] has quit [Quit: leaving]
00:15 -!- stratsi [~stratsi@80.72.68.187] has quit [Ping timeout: 488 seconds]
00:33 -!- coryb [~cory@190.40.cm.sunflower.com] has joined #uml
00:47 < coryb> does patching 2.4.21 w/ 2.4.20-6um work okay? is anyone using it in production?
00:51 -!- ljlane [ljlane@66.93.83.133] has joined #uml
01:20 < david> coryb: you get a reject, but you can ignore it
01:20 < david> coryb: I have patches against 2.4.21 at kernel.usermodelinux.org
02:09 -!- stratsi [~stratsi@80.72.68.187] has joined #uml
02:10 -!- pahan [Blah@ca-manbch-cuda3-c2a-a-52.stmnca.adelphia.net] has joined #uml
02:10 -!- pahan is now known as Pahan
02:48 < coryb> yes, I looked at the reject, it seems that part of the uml patch was already in 2.4.21
02:48 < coryb> so it patched fine
04:30 -!- MicW [~chatzilla@pD952B701.dip.t-dialin.net] has joined #uml
04:30 < MicW> hi!
04:31 < MicW> i need some help with booting from hostfs:
04:31 < MicW> i can boot the redhad image without problmes, when is run "linux ubd0=root_fs".
04:32 < MicW> but when i mount it on the host with "mount root_fs /mnt -o loop" and run "linux ubd0=/mnt" it fails with "Kernel panic: VFS: Unable to mount root fs on 62:00"
04:33 < MicW> what i'm doing wrong? i think it's exact the way that the faq describes
04:59 -!- stratsi [~stratsi@80.72.68.187] has quit [Ping timeout: 488 seconds]
05:15 < rob> MicW: you dont mount the root_fs like that
05:16 < rob> the HOWTO describes being able to edit the root_fs by mounting it on the host
05:16 < rob> so on the host, you do
05:16 < rob> mount -o loop root_fs /mnt/root_fs
05:16 < rob> and you get the root_fs mounted on /mnt/root_fs so you can edit it without booting a UML
05:16 < MicW> rob: there's a section in the howto named "Host file access"
05:17 < MicW> "9.2 hostfs as the root filesystem"
05:17 < MicW> ind this says "It's possible to boot from a directory hierarchy on the host using hostfs rather than using the standard filesystem in a file."
05:17 < MicW> but it does'nt work
05:18 < MicW> i'd like to use a directory instead of an image file to boot the uml from
05:18 < rob> ah, I see
05:18 < rob> then you need to do..
05:19 < rob> mount root_fs uml_root_dir -o loop
05:19 < rob> and then
05:19 < rob> linux ubd0=/path/to/uml/root/directory
05:20 < MicW> yes. but it fails always
05:20 < MicW> the image works the directory not
05:20 < rob> does the user have read/write permission to the directory?
05:21 < MicW> yes. the user is root :-)
05:21 < rob> hmm
05:21 < rob> tell you what
05:21 < rob> let me have a go
05:22 < MicW> sorry. my english is not that goot. that meens, you'll try it?
05:22 < rob> yes, I'll see what happens with this one..
05:22 < rob> wait a moment, I dont have host_fs compiled into my UML kernel, do you?
05:23 < rob> and also, have you changed /etc/fstab so that the UML knows that the root fs is hostfs?
05:24 < MicW> 1. yes. 2. let me ckeck...
05:30 < MicW> i tried it now with the redhat image. i mounted it on the host, changed etc/fstab and runned "./bin/linux ubd0=/opt/usermodelinux/hosts/test/"
05:30 < MicW> there is a kernel message "unable to open /opt/usermodelinux/hosts/test/ for validation"
05:30 < MicW> and later "Kernel panic: VFS: Unable to mount root fs on 62:00"
05:30 < MicW> did this work for you?
05:31 < rob> nope, I dont have hostfs, so I get a kernel panic as well, 
05:32 < MicW> i have hostfs. when i boot and run "cat /proc/filesystems", it's there. and i complied it in
05:34 < MicW> do you know if a uml developer is here in the channel?
05:35 < rob> I'm not sure if there's any developers, there's david who's very very knowlegable, not sure if he's a developer though
05:37 < MicW> so i'll try to ask him :-)
05:39 < MicW> david, rob saiy, you'r very knowlegable :-) - do you know how i can boot a uml from a directory instaed of an image file? the way described in the howto does not work
05:40 -!- MicW [~chatzilla@pD952B701.dip.t-dialin.net] has quit [uranium.oftc.net quark.oftc.net]
05:40 -!- Pahan [Blah@ca-manbch-cuda3-c2a-a-52.stmnca.adelphia.net] has quit [uranium.oftc.net quark.oftc.net]
05:40 -!- ljlane [ljlane@66.93.83.133] has quit [uranium.oftc.net quark.oftc.net]
05:40 -!- coryb [~cory@190.40.cm.sunflower.com] has quit [uranium.oftc.net quark.oftc.net]
05:40 -!- mistral [mistral@jstevenson.plus.com] has quit [uranium.oftc.net quark.oftc.net]
05:40 -!- pirlouit [~peter@64.162.195.202] has quit [uranium.oftc.net quark.oftc.net]
05:40 -!- mdz [~mdz@216-15-124-77.c3-0.smr-ubr3.sbo-smr.ma.cable.rcn.com] has quit [uranium.oftc.net quark.oftc.net]
05:40 -!- Z-Wing [zedders@zanzar.net] has quit [uranium.oftc.net quark.oftc.net]
05:40 -!- albino [~albino@neuf.dorms.usu.edu] has quit [uranium.oftc.net quark.oftc.net]
05:40 -!- ichilton [~ian@pc3-stoc3-4-cust203.midd.cable.ntl.com] has quit [uranium.oftc.net quark.oftc.net]
05:40 -!- dilinger [irc@sloth.voxel.net] has quit [uranium.oftc.net quark.oftc.net]
05:40 -!- albino [~albino@129.123.227.104] has joined #uml
05:40 -!- Pahan [~pahan@68.68.228.52] has joined #uml
05:40 -!- ichilton [~ian@80.6.255.203] has joined #uml
05:40 -!- mistral [mistral@jstevenson.plus.com] has joined #uml
05:41 -!- mdz [~mdz@216.15.124.77] has joined #uml
05:41 -!- pirlouit [~peter@duvel.drunkcoders.com] has joined #uml
05:41 -!- dilinger [irc@207.99.115.73] has joined #uml
05:43 -!- coryb [~cory@24.124.40.190] has joined #uml
05:45 -!- coryb_ [~cory@24.124.40.190] has joined #uml
05:46 < lah> arch/i386/kernel/kernel.o(.text+0x6129):undefined reference to
05:46 -!- coryb_ [~cory@24.124.40.190] has left #uml []
05:46 < lah> 'proc_mm_get_mm'
05:46 < lah> hum.
05:47 < lah> ideas, anyone?
05:49 < rob> don't compile /proc/mm support in and see if it likes that?
05:50 < lah> hum. sounds reasonable, but don't I need /proc/mm for uml->skas ?
05:51 < rob> query on that one, not sure.
05:51 < rob> try it :)
05:53 -!- MicW [~chatzilla@pD952B701.dip.t-dialin.net] has joined #uml
05:53 < lah> will do
05:53 < MicW> re
05:53 < MicW> i had network problems :-/
05:53 < rob> :(
05:54 < MicW> was there an answer to my question?
05:54 < rob> no
05:55 -!- ljlane [ljlane@66.93.83.133] has joined #uml
05:58 < lah> rob, err.. it looks like /proc/mm was disabled (dunno how, I'm sure I had enabled it), so I'll retry with it enabled :)
05:59 < lah> WAY better.
05:59 < rob> :)
06:02 < MicW> is /proc/mm nessesary to run uml?
06:04 < coryb> only in skas mode
06:05 < MicW> hm. this may be th reason why my kernel build fails on mm when i apply the skas patch
06:06 < coryb> mrmmrmmm
06:06 < coryb> I"m way too tired to offer any kind of insight
06:06 < coryb> <- awake 38 of the past 39 hours
06:07 < lah> micq, what error msg?
06:09 < MicW> don't know. wait. i compile it again
06:10 < MicW> (maybe it conflicts with the ctx kernel patch that was also applied)
06:12 < lah> i'm trying grsecurity+skas ;)
06:12 < lah> hangs before running init
06:13 < MicW> is here anyone who can boot uml from a directory instead of an image?
06:31 < MicW> i'm now compiling a clean 2.4.21 kernel with skas patch (and without ctx patch). hope it compiles now :-)
06:31 < MicW> no. same error:
06:31 < MicW> arch/i386/kernel/kernel.o: In function `sys_ptrace':
06:31 < MicW> arch/i386/kernel/kernel.o(.text+0x6afc): undefined reference to `proc_mm_get_mm'
06:31 < MicW> make: *** [vmlinux] Error 1
06:32 < MicW> i have mm-1.2.1 on my system
06:32 < MicW> do you know what's the problem?
06:34 < lah> yeah
06:34 < lah> i got the same before
06:34 < lah> make sure you check /proc/mm support
06:34 < MicW> thanks
06:34 < MicW> recompiling... :-) once again
06:36 < MicW> btw: arp (with tuntap) seems not to work if the traffic_shaper device is compiled in the kernel: all arp entries are assigned to that device and the uml is unreachable from other hosts
06:43 < MicW> lah: thanks: it compiled
06:44 < lah> i had exactly the same problem before :)
06:45 < MicW> maybe the skas patch should force this option
06:45 < lah> rob mentioned that I should disable /proc/mm, but I already had it disabled.. so I thought I could try it with /proc/mm enabled
06:51 -!- stratsi [~stratsi@80.72.68.187] has joined #uml
06:55 < rob> lah: you're trying to get grsec and skas working?
06:55 < rob> I've had no success with that because both of the patches affect the mm stuff within the kernel so the failed hunks are a direct conflict with each other
06:56 < MicW> yeah. redhat is installing from the web
07:06 -!- MicW is now known as MicW_away
07:06 < lah> rob: yeah, from the rejects it looks like some do_mmap vs do_mmap_pgoff problem
07:07 < rob> yep, I've played with it for a week
07:07 < lah> but since I don't know anything about it I simply tried replacing that line, now it just stops before running init
07:07 < rob> and can't get it going
07:07 < rob> have you looked at the openwall patches?
07:07 < lah> with 2.2.x yes
07:11 < lah> i wonder if anyone got it working with grsecurity though
07:11 < lah> "pflanze" was looking for it, too
08:54 -!- MicW_away is now known as MicW
08:54 < MicW> re
08:58  * green finally got canadian visa. Phew, that was hard
09:09 -!- stratsi [~stratsi@80.72.68.187] has quit [uranium.oftc.net quark.oftc.net]
09:10 -!- stratsi [~stratsi@80.72.68.187] has joined #uml
09:28 < MicW> i try to install redhat on my uml machine. i
09:28 < MicW> i'm using the initrd from the bootdisk and an empty disk image.
09:29 < MicW> but when the "staage2.img" is mountet, the setup gets a sig11 and halts the system
09:29 < MicW> any idea?
09:29 < green> hae you seen http://linuxhacker.ru/uml already?
09:30 < green> sounds like you did not gave enough RAM to the uml
09:30 < MicW> hey. that's great ;-)
09:30 < MicW> looks like a step by step doc
09:31 < MicW> the uml has 256mb ram
09:31 < green> ok, then read the howto and try to urtilize the newly gained knowledge ;)
09:33 < MicW> i'm going to
09:34 < MicW> compiling uml again...
09:35 -!- jdike [~jdike@dhcp136.ISTS.dartmouth.edu] has joined #uml
09:35 < green> Hi Jeff!
09:35 < jdike> hi Oleg
09:35 < jdike> Just the guy whose brain I wanted to pick
09:36 < green> hehe
09:44  * jdike commences the brain-picking of green
09:45 < jdike> green: OK, my COW driver has a problem I don't know how to deal with
09:46 < jdike> green: When it gets a write request, it has to make a request (or two) to write out the COW bitmap if necessary
09:46 < jdike> green: err, hold on
09:47 < jdike> green: brainfart - I was thinking the bitmap request had to finish before the original write started, but it's the other way around
09:49 < green> yeh, this is very common thing - you start to describe your problem to someone and suddenly see the problem ;)
09:49 < Getty> brainfart?
09:50 < Getty> what a word :)
09:50 < jdike> green: if I have a make_request that just rewrites the request, how do I find out when the request has finished?
09:50 < jdike> green: hehe, yeah
09:50 < Getty> is that american slang?
09:50 < green> jdike: wait_on_buffer?
09:50 < jdike> green: I have a separate thread, so it can wait_on_buffer for it, and do the IO when it wakes up
09:51 < jdike> green: wait_on_buffer doesn't work within a make_request, I think :-)
09:51 < jdike> Getty: maybe it's US slang, I thought it was pretty well understood
09:52 < Getty> i never heard it, and even when i translate it, i can't see any german common part :)
09:52 < Getty> brainfart - hirnbratzer ;)
09:52 < Getty> or hirnfurz ;)
09:52 < Getty> oh man, thats funny :)
09:53 < jdike> Getty: do "thinko" or "braino" mean anything, they're the same thing
09:53 < Getty> braino?
09:53 < Getty> funny
09:53 < jdike> Getty: analogous to "typo"
09:53 < Getty> ah :)
09:54 < green> jdike: why do you need to wait for something anyway? Are you trying to achieve some kind of atomicity?
09:54 < jdike> green: the bitmap can't be updated until after the data write has finished
09:55 < jdike> green: coz if it's done the other way and there's a crash in between
09:55 < jdike> green: the bitmap will say there's data in the COW file when there isn't
09:56 < green> jdike: when io on some buffer head is finished, it's bh->b_end_io() method is called, does this helps you?
09:57 < jdike> green: don't think so, because I have a make_request that redirects the request to an underlying device
09:57 < jdike> green: so it will be that other driver that gets the b_end_io call, not the COW driver
10:00 < green> well, devices do not use this b_end_io
10:00 < green> ll_rw_blk sets it to simple completion handler tha tmarks page uptodate
10:01 < green> so if you do not use ll_rw_blk, you can easily get b_end_io to do something that you need. (and then call default halndler if necessary)
10:01 < jdike> green: ah, I mistook it for end_request, nevermind
10:02 < jdike> green: is that safe to do?  i.e. if another driver involved with the request uses it, will it chain my b_end_io?
10:03 < green> drivers have nothing to do with b_end_io, generally speaking.
10:03 < jdike> green: well, you're suggesting that mine do, so maybe others do, as well
10:03 < jdike> green: and they should play well together if so
10:04 < green> well, I assume you are going to wait on buffer that is yours, if not, then it is good idea to chain b_end_io calls of course
10:04 < jdike> green: another thing, I don't control whether the original request goes through ll_rw_block
10:05 < jdike> green: someone else submits it
10:05 < green> well, then you must remember old b_end_io value and call it from your b_end_io method
10:05 < jdike> green: so the buffer belongs to someone else, I just need to know when it finishes
10:09 < MicW> i modified uml and stage2.img as described in "http://linuxhacker.ru/uml/". but the system still crashes with int11 after mounting the image
10:10 < green> jdike: so I still think that hooking into b_end_io chain is one of the right ways for you.
10:10 < Getty> hey btw...
10:10 < green> MicW: are you using some modern kernel? are there any kernel messages?
10:10 < Getty> stupid question...
10:11 < Getty> do you think that it could be possible to remove parts from the host kernel (deeper then what you have in kernel config) to probably get more power to the UMls?
10:11 < jdike> green: can I submit IO from b_end_io?
10:11 < MicW> green nothing that tells me why it crashes (the setup crashes and shuts down the system, not the whole uml)
10:12 < MicW> it tries to insmod something before that
10:12 < jdike> Getty: example?
10:12 < Getty> jdike: i
10:12 < Getty> i'm no kernel coder, i can't exactly say what i could mean
10:13 < Getty> the idea is simple, you make the host kernel fast, cause of removing things which are only relevant to non-uml things
10:13 < jdike> MicW: if it's using modules, then they need to be UML modules
10:13 < Getty> some kind of uml-host-linux patch if you know what i mean
10:13 < Getty> but not to make the uml more powerful
10:13 < Getty> only making the host-kernel faster
10:13 < jdike> Getty: you could, but I don't think it would get you too much
10:13 < jdike> Getty: removing system calls that UML doesn't use, etc
10:13 < Getty> jdike: what is btw at the moment the percent you lose on the power of the UML?
10:14 < green> jdike: I think you can.
10:14 < MicW> jdike: i did the things described in http://linuxhacker.ru/uml/ step by step. there where nothing about the modules.
10:14 < jdike> Getty: it'll save some memory, but that's about it, I think
10:14 < jdike> Getty: depends
10:14 < Getty> jdike: from... till ... ?
10:14 < jdike> Getty: on CPU-bound stuff, you lose basically nothing
10:14 < green> MicW: Yes, but are there any suspicious messages in other windows (not only the main one with installer)
10:14 < Getty> 10-50% ? ;) 2-5%?
10:14 < Getty> jdike: oh really?
10:15 < jdike> Getty: with skas, on a kernel build, it's something like 25%
10:15 < jdike> err, more like 30%
10:15 < MicW> green: the messages where on one of these cosoles. now i modified the file "module-info" to make sure the modules are not the problem
10:15 < Getty> jdike: thats very high
10:15 < green> MicW: modules are not the problem in any case, as there are none ;)
10:15 < Getty> jdike: and you think i get no advance in my idea?
10:16 < green> MicW: but the messages might provide some info about why it crashed
10:16 < MicW> green: yes. but redhat tried to load some
10:16 < jdike> Getty: not as high as it used to be
10:16 < Getty> i don't wanna use skas :-/
10:16 < green> MicW: It tried, but this does not matter
10:16 < MicW> i restart and paste the messages here
10:16 < green> MicW: not all of them, just last 2-3
10:16 < jdike> Getty: on a kernel build, with tt mode, it's more like 100%
10:16 < MicW> maybe the problem is that it tried to insmod and my uml kernel is kompiled without module support
10:17 < green> MicW: No, I installed RH with monolithic kernel with zero problems
10:17 < MicW> here are the messahes:
10:17 < MicW> messages
10:17 < MicW> * modules to insert cramfs vfat sunrpc lockd nfs
10:17 < MicW> * module(s) cramfs vfat sunrpc lockd nfs not found
10:17 < MicW> * load module set done
10:17 < MicW> * modules to insert ide-cd
10:17 < MicW> * module(s) ide-cd not found
10:17 < MicW> * load module set done
10:17 < MicW> * modules to insert scsi_mod sd_mod sr_mod
10:17 < MicW> * module(s) scsi_mod sd_mod sr_mod not found
10:18 < MicW> * load module set done
10:18 < MicW> * looking for usb controllers
10:18 < MicW> * no usb controller found
10:18 < MicW> * no firewire controller found
10:18 < MicW> * probing for floppy devices
10:18 < MicW> * no floppy devices found
10:18 < MicW> * PIO_FONT failed: Invalid argument
10:18 < MicW> * PIO_UNIMAPCLR failed: Invalid argument
10:18 < MicW> * PIO_UNIMAP failed: Invalid argument
10:18 < MicW> * PIO_UNISCRNMAP failed: Invalid argument
10:18 < MicW> * starting to STEP_URL
10:18 < MicW> * going to do getNetConfig
10:18 < MicW> * reverse name lookup failed
10:18 < MicW> * going to do nfsGetSetup
10:18 < MicW> * mounting nfs path snoopy.lan:/store/redhat/8.0
10:18 < MicW> * mntloop loop0 on /mnt/runtime as /mnt/source/RedHat/base/stage2.img fd is 14
10:18 < MicW> * got url nfs://mnt/source/.
10:18 < MicW> * getting ready to spawn shell now
10:18 < MicW> * not spawning a shell
10:18 < MicW> * modules to insert ide-cd
10:18 < MicW> * module(s) ide-cd not found
10:18 < MicW> * load module set done
10:18 < MicW> * modules to insert scsi_mod sd_mod sr_mod
10:18 < MicW> * module(s) scsi_mod sd_mod sr_mod not found
10:18 < MicW> * load module set done
10:18 < MicW> and on the other shell:
10:18 < MicW> <4>nfs warning: mount version older than kernel
10:18 < MicW> <4>Unable to identify CD-ROM format.
10:18 < MicW> and on the main console:
10:18 < MicW> install exited abnormally -- received signal 11
10:18 < MicW> and some shutdown messages
10:19 < green> hm, "CD-ROM format"? 
10:19 < MicW> i don't know why that message appears
10:19 < Getty> jdike: 100% percent means you have no power on uml ;)
10:19 < Getty> jdike: so i should use skas patch?
10:19 < green> you have chosen installation from NFS, did you?
10:20 < jdike> Getty: depends how you interpret it
10:20 < jdike> Getty: with your interpretation, the slowdown with tt would be 50%
10:20 < Getty> tt is without skas, right?
10:20 < MicW> jdike: athre you _the_ jdike who wrote that great howto?
10:21 < MicW> green: yes. i tried to install from nfs
10:21 < jdike> Getty: and with skas, more like 25%
10:21 < MicW> i used the initrd from bootnet.img
10:22 < Getty> jdike: thats really much
10:22 < Getty> mhmhmh...
10:22 < jdike> MicW: I don't know how great it is, but I'm the one
10:22 < green> MicW: hm, and you said that your installation media is NFS, right?
10:23 < MicW> green: yes. there is only nfs, http, and ftp, no word about cdrom
10:23 < MicW> jdike: yes. ist' a very good start to uml
10:24 < green> MicW: Hm, you are installing what RH version?
10:25 < MicW> jdike: could you boot from an directory as described in "9.2 hostfs as the root filesystem"? that never worked for me
10:25 < MicW> green: 8.0
10:26 < jdike> MicW: it has worked for me, I haven't done it recently
10:27 < green> MicW: Hm. I cannot immediately see what's wrong, but simething is definitely wrong as I can install rh8 with no problems
10:28 < MicW> green: i'm going to try it with the uml config provided on that page
10:29 < green> good idea ;)
10:29 < MicW> jdike: setting a directory as argument for ub0=... seems not to work
10:29 < jdike> MicW: I got rid of that in recent UMLs
10:30 < jdike> MicW: turns out there's a better way to do it, which is documented
10:30 < MicW> jdike: any hint, how this could be done or where it's documented?
10:31 < jdike> http://user-mode-linux.sf.net/hostfs.html
10:32 < green> jdike: btw it seems I will appear at KS, got my canadian visa today. (it took 3 weeks, canadian embassy is worst embassy of all ones that I've seen)
10:32 < jdike> green: awesome
10:32 < coryb> KS?
10:32 < jdike> green: did Hans give you any grief about you going and not him?
10:33  * coryb is from kansas, can't help but perk up ;)
10:33 < jdike> coryb: Kernel Summit :-)
10:33 < MicW> jdike: great. i saw that page 100 times but i never saw, that tha last command line is different from the one in the howto ;-)
10:33 < jdike> coryb: Ottawa, some distance from Kansas
10:33 < coryb> hehe
10:33 < green> jdike: we had lots of fighting about that and he heavily pressed on me to not go, and then a week ago he suddenly changed his mind and even offered to reimburse me all travel costs
10:34 < jdike> MicW: hmmm, the howto is supposed to be generated from the same source and at the same time
10:34 < jdike> green: strange, I wonder why
10:35 < green> jdike: Same here. actually. He is strongly believing that people tried to underestimate his role by not inviting him.
10:35 < MicW> jdike: the howto (that one that is linked from user-mode-linux.sf.net) is different here
10:35 < green> jdike: anyway he is not pressuring me anymore and this is good
10:36 < jdike> green: yeah, I can believe that
10:36 < jdike> green: he's an odd combination of intelligence, arrogance, and paranoia
10:38 < green> hehe
10:39 < jdike> green: do you have your hotel set?
10:39 < green> no. I only got visa today, so I am planning to make reservation today. I am way too off for the official otel, though.
10:39 < jdike> green: and are you staying for OLS?
10:40 < green> No, I am not staying for OLS, nobody going to reimburse me for that ;)
10:40 < green> and I am too cheap to afford that myself ;)
10:40 < jdike> green: I've got a spare bed in my room in Les Suites
10:40 < jdike> green: spare room, actually, for KS
10:40 < MicW> green: the different uml kernel seems to work
10:40 < jdike> green: you're welcome to it
10:41 < green> jdike: First I will try to get something in nearby hotels. I hate to disturb people. Also I arrive July 19th, and everybody seems to come on 20th.
10:42 < green> but thatnk yo for the offer anyway.
10:42 < jdike> green: I'm arriving on the 20th
10:43 < green> it's funny how planes from russia to canada arrive on the random dates but the ones that are suitable for me ;)
10:43 < jdike> green: the offer stands, let me know if you change your mind
10:43 < jdike> green: hehe
10:43 < green> MicW: now look at difference in configs and may be it will give you some ideas, if you are interested to find the reason
10:44 < MicW> green: yes. but first i start the installation
10:44 < MicW> :-D
10:44 < green> ;)
10:44 < MicW> do you know if a can later (when the stage2 is mountet from nfs) can switch to ftp install?
10:44 < MicW> it's my first redhat install...
10:45 < green> MicW: no, you cannot. You can start ftp install immediately, only it will hang later one because it seems to take probe binary from other place
10:46 < MicW> so i have to download all rpms to my nfs location?
10:46 < green> MicW: yes. Or you can use this userspace stuff, that allows you to mount remote ftp sites. I forgot the name, though. something like userfs
10:47 < MicW> i tried this once. it was very unstable
10:47 < MicW> are there some public nfs mirrors?
10:47 < green> I do not know.
10:48 < BB> nfs mirrors of what?
10:48 < green> I always amke RH CDs, so it was not a problem for me to setup local NFS repository for install
10:48 < MicW> hm. my virtual harddisk is shown with 13 mb :-(
10:48 < MicW> not my day today
10:49 < green> create a bigger one? ;)
10:49 < MicW> i createt a 500 mb file
10:50 < MicW> fdisk shows only 7 cylinders
10:51 < green> hm, weird.
10:51 < MicW> it was my fault
10:52 < MicW> i used the wrong file as ubd0
10:52 < MicW> i should have a break
10:52 < MicW> i fdisked my test root_fs
10:54 < MicW> is there a way to install rh directly into a hostfs directory?
10:54 < MicW> or should i copy it after installing?
10:55 < green> probably this (native installation to hostfs) won;t be too easy. But it may be more easy if you'd use umlbuilder
10:56 < jdike> MicW: my build procedure wasn't keeping the HOWTO up to date
10:56 < jdike> MicW: I just fixed that
10:56 < MicW> jdike: thanks
10:57 < MicW> jdike: maybe you could write to the tuntap section that the ip that is assigned to the tap device should be the same as the hosts ip. i misunderstood this, asigned the guest ip and searched hours for the bug ;-)
10:58 < jdike> MicW: it doesn't need to be
10:58 < jdike> MicW: I typically don't
10:58 < MicW> jdike: it needed this to reach the uml from other hosts in my network. but  it may have been an arp problem
10:58  * green usually does not have any extra IPs just for tuntap interfaces ;)
10:59 < MicW> jdike: or say that the ip _must not_ be the same as the one assigned to the guest
11:01 < MicW> when the ip is the same as the guests, the host hanled all connections to that ip locally and nothing seems to be forwarded. an nmap to that ip shows the host's services
11:01 < jdike> MicW: "Note that the IP address you assign to the host end of the tap device must be different than the IP you assign to the eth device inside UML."
11:02 < jdike> MicW: followed immediately by "If you are short on IPs and don't want to comsume two per UML, then you can reuse the host's eth IP address for the host ends of the tap devices. Internally, the UMLs must still get unique IPs for their eth devices."
11:03 < MicW> wow. when i disable everything, rh installs 500 mb (fits exactly on my virtual disk)
11:04 < green> there is special "router" setup, but I do not know if you can activate that if not using kickstart
11:04 < MicW> jdike: i thougth this was for ethertap when i read this. after ethertap not worked, i switched to tuntap and forgot that i read this
11:05 < jdike> MicW: also, UML complains if you assign the same IP to both ends of the tap device
11:06 < MicW> jdike: i didnt see this
11:06 < MicW> but now it works :-)
11:06 < david> morning
11:07 < jdike> hi david
11:07 < jdike> david: BTW, I noticed you revived the original color scheme on uml.org
11:07 < jdike> david: it's a lot more pleasant than the white-on-white one
11:10 < david> jdike: yep, it's much better
11:14 < green> jdike: had you a chance to look at this page-dirtier-makes-gdb-cpuhungry-in-skas-mode problem?
11:15 < jdike> green: no
11:15 < jdike> green: I barely have time to go to the bathroom
11:16 < green> ah
11:17 < BB> maybe you should setup a paypal fund for a comode ;)
11:17 < jdike> BB: hehe
11:17 < coryb> herman miller aeron toilet
11:18 < green> jdike: btw it works for namesys, we charge something like $25 for various stuff done on higher priority and this gives us some (relatively small) amount of money. Hans claims this stream of money is enough to have 1/4 of developer ;)
11:19 < MicW> do you think it's better to use a hostfs or an image as root?
11:20 < jdike> green: reiserfs has a somewhat high profile than UML
11:20 < jdike> s/high/higher
11:21 < green> jdike: UML is quickly gaining popularity at various virtual host providers
11:22 < green> jdike: and reiserfs is lacking major bugs these days which keeps our income reduced... (/me makes notes in the evil todo list ;) )
11:22 < jdike> green: it is, and that may help in the future
11:22 < MicW> green: yes. you can set up a small server for just a few $/¬ per month
11:22 < jdike> green: hehe, funny how that works
11:22 < jdike> green: the better your code, the smaller your support revenue
11:23 < green> jdike> green: hehe, funny how that works
11:23 < green> oops
11:23 < green> jdike: Actually there are random feature requests too
11:23 < green> and even support contracts if you are lucky ;)
11:23 < jdike> green: so far, they are small outfits with no spare cash
11:23 < jdike> green: that will probably change
11:23 < green> yes, I hope so.
11:24 < david> jdike: once they make money using UML they are more likely to spend it to support the project
11:24 < green> jdike: uml works pretty well for my virtual hosting needs ;)
11:24 < jdike> green: an amusing thing - sw-soft seems to be advertising on google searches for UML
11:24 < jdike> green: also virtuozzi (sp?)
11:24 < green> jdike: hehe.
11:25 < david> jdike: doesn't sw-soft make that?
11:25 < green> they are using this crap "shared kernel idea"
11:25 < jdike> green: so they seem to be feeling some heat
11:25 < jdike> david: no, virtuozzi is done by some .ca company
11:25  * green hates how swsoft hides virtuozzo kernel kode. Too bad I do not have thei binary kernel to get GPL enforced on them
11:25 < jdike> david: sw-soft is Russian
11:26 < david> jdike: http://www.sw-soft.com/en/products/virtuozzo/ve/
11:26 < jdike> david: hmmmm, maybe I have some wires crossed
11:26 < green> swsoft makes "virtuozzo" it is mostly located in Singapur, though they have large development facility in Moscow, too.
11:27 < jdike> green: there's another shared-hosting thing
11:27  * jdike totally blanks on the name
11:27 < david> ENsim?
11:27 < green> jdike: there are lots of those things
11:27 < green> jdike: e.g. VmWare ;)
11:27 < jdike> david: no, that's BSD last time I checked
11:28 < david> oh, ok
11:29  * MicW is downloading rh8 package by package :-(
11:29 < david> MicW: mget
11:30 < green> MicW: don't download from ftp.redhat.com, they have weird bandwidth limits, use some mirror
11:30 < MicW> david: the line does only 80kbit/sec. so i download the packages the install searches. and yes: i'm using a near mirror
11:30 < green> MicW: also you can get rh9 while you are at it. I like rh9 much more
11:30 < MicW> green: rh9 is currently english only
11:31 < MicW> my language is german
11:31 < MicW> if it where english, i'd take debian ;-)
11:31 < green> redhat does not produce russian distros which never stopped me from using it ;)
11:31 < MicW> normaly i use suse because it's a german distri.
11:32 < MicW> but there where no way to make it run on uml
11:32 < green> there is
11:32 < jdike> MicW: the installer can be made to work
11:32 < green> the mailing list have several links to some "install-uml" script
11:32 < green> or something like that
11:32 < jdike> MicW: also maybe umlbuilder
11:32 < MicW> working with the latest suse?
11:32  * green runs SuSE's rescue disk in uml as a test fs for various debugging stuff
11:33 < MicW> i got it run today. but "yast" did not work. it hangs on loading it's modules
11:34 < MicW> i had to change many scripts to make it boot
11:34 < green> people report that they ca install suse, so I guess it works.
11:34 < MicW> and now i saw, that there's redhat which 1) runs nearly out of the box ( ;-) ) and 2) is localized to germany
11:34  * green tired of SuSE on my WS hard enough to not use it anywhere else if possible ;)
11:35 < MicW> i like suse on my workstation
11:36 < MicW> jdike: how does umlbuilder work?
11:36  * green found suse 8.x to be way too fragile
11:37 < MicW> i'm running 8.1 on my workstation since this version is out. and it works fine
11:37 < jdike> MicW: haven't used it lately
11:38  * green skipped 8.1, 8.2 have broken compiler which is pain in the ass by itself already
11:39 < MicW> what's broken there? here every thing compiles fine (also uml :-D )
11:39 < green> some of our code was miscompiled.
11:40 < green> BTW gcc in suse 8.1 is also broken, it segfaults when you compile recent 2.5 kernel
11:40 < MicW> never tried this. i always use stable kernels :-)
11:41 < green> we found than 8.2's gcc eats stack for no good reason randomly (like 5k at a time) which is not very good idea for kernel code
11:41 < green> also it generated rather poor code
11:41 < MicW> so maybe i should wait for 8.3
11:41 < MicW> or using rh8. have you experience with this on desktop?
11:42 < green> yes, I like rh9 more, but rh8 was also usable.
11:43 < MicW> hm. or should i wait for rh9 german?
11:43 < green> they now use unicode (in rh8 and rh9) by default, but rh8 was less unicode-ready. Probably not critical for you, but it is critical for me
11:43 < MicW> i see
11:44 < MicW> but thats more important for uor servers since there are some multilingual webaps
11:45 < MicW> now i'm on letetr "m" with downloading/installing packages manually :-)
11:48 < MicW> how would you set up swap when using hostfs as root?
11:51 < green> swap t oa file, I guess. of swap into separate ubd
11:51 < MicW> which is faster? a file on hotfs or an ubd?
11:51 < MicW> or gife the machine much memory and let the host swap?
11:51 < green> I do not know. ubd is slow, and it involves double copy
11:52 < MicW> uh. it's package 28 of 250 :-(
11:53 < MicW> i decided to download all and set it up tomorrow :-)
11:53 < green> ;)
11:54 < MicW> i hope, the result was all the work worth...
11:54 < MicW> currently i have the virtual server running with ctx linux
11:54 < MicW> that's a patch that enables multiple contexts on one kernel
11:54 < green> multiple contexts?
11:55 < MicW> yes. you can run any software in a context. the root context is 0 which is the "normal" system
11:55 < MicW> wait. i have the url...
11:55 < green> ah, the flawed "shared kernel" approach to virtual hosts...
11:56 < MicW> yes. the processes shared the kernel
11:57 < MicW> the good thing is that every virtual context runs at full speed and cann access all ressources
11:57 < MicW> http://www.solucorp.qc.ca/miscprj/s_context.hc
11:57 < green> yeah, this is very useful for people fho are trying to break out of their context ;)
11:58 < MicW> i'm using it to have a development copy of our productive machines without the need of mor hardware
11:58 < MicW> and therefore it's fine
11:58 < green> ah.
11:59  * green prefers to have more hardware whenever possible ;)
11:59 < MicW> but it consists of many scripts and when i want to move it to a different machine, it's nearly as much work as the satup was
11:59 < MicW> the uml way is much cleaner
12:00 < MicW> and i can take the whole package and start it anywhere
12:00 < MicW> so i can put it on every development machine and the first thah is bootet will start the test server
12:01  * green decides to go home
12:01 < MicW> god idea
12:01 < MicW> my download will take some hours
12:02 < MicW> and i was working for 10 ours today (one again...)
12:02 < MicW> s/(ours)/h$1/
12:03 < MicW> thanks a lot for your help
12:03 < MicW> cu
12:04 -!- MicW [~chatzilla@pD952B701.dip.t-dialin.net] has quit []
13:32 -!- kayl [~jada@6532176hfc248.tampabay.rr.com] has joined #uml
13:44 < kayl> Hi, trying to recompile a uml and the system ends up in single user mode for maintenance telling me that it can not find /dev/ubd0 and that the root partition is corrupted. After looking, turns out there is a /dev/ubd/0 but no /deb/ubd0... Would it be a bad option in the .configure (really I don't see why but...) or something else? Any suggestion is more than welcomed :)
13:45 < BB> more like your using devfs and it didnt expect you to possibly...
13:47 < kayl> let me check about this, I never looked into devfs and just expected it to designate the handling of the /dev ... 
13:52 < kayl> mmmm why is the /dev FS support rated experimental, I thought it was part of the basics for any linux... I think I misunderstand what the support for devfs means... Thanks for the tip BB I'm recompiling trying this disabled
13:53 < caker> kayl: you can just launch with devfs=nomount on the command line to turn it off
13:53 < jdike> caker types faster than me :-0
13:53 < kayl> caker: oops ok I take note! 
13:53 < kayl> jdike: :)
14:01 < kayl> it worked, thansk for the advices ! 
14:02 < kayl> that was definitively it, gonna read up a little on devfs I think I'm off the road on what this is ;P
14:02 < jdike> kayl: the device names in your fstab and inittab were non-devfs names
14:05 < kayl> jdike: ok does this mean that linux has a "raw" syntax in the form of /dev/ubd/0 and, when devfs is used, a syntax like /dev/ubd0 ? A while ago, I looked at the solaris paths to designate a scsi cd rom and it was a lot more complicated than the linux /dev interface I was used to ... Is devfs responsible then for the pretty convenient organization of /dev?
14:06 < caker> kayl: subfolders indicate devfs
14:06 < jdike> kayl: it's responsible for a different organization
14:06 < kayl> caker, jdike: ok, something I completely overlooked 
14:06 < jdike> kayl: ubd0 is the standard syntax, ubd/0 is devfs, from your point of view, it's just different names
14:12 -!- idcmp [idcmp@69.41.229.181] has joined #uml
14:20 < idcmp> any life?
14:20 < caker> depends on definition
14:21 < jdike> zzzzzz *snort* zZZzzZ
14:23 < jdike> oops
14:23  * jdike forgot about the logger
14:23 < idcmp> heh
14:23 < idcmp> is the best way to backup a uml instance still to freeze/sync/copy cow/thaw ?
14:24 < caker> Doesn't LVM have a snapshot feature?
14:24 < caker> (related topic)
14:24 < idcmp> yup
14:27 < idcmp> i wonder if that would all work together..
14:34 -!- davidc [~david@207.166.203.132] has joined #uml
14:34 < davidc> hey
14:41 -!- davidc [~david@207.166.203.132] has quit [Quit: davidc has no reason]
14:41 -!- davidc [~david@tailtiu.davidcoulson.net] has joined #uml
14:41 < davidc> is this thing on?
14:42 < caker> Hey
14:42 < davidc> jdike: i'm not home right now
14:42 < jdike> yup
14:42 < davidc> jdike: david is my laptop
14:42 < jdike> OK
15:01  * davidc is away: (Auto-Away after 10 mins) [BX-MsgLog On]
15:13 < kayl> weird, I'm working through ssh with x11 forwarding and the xterm console pops up only to disappear immediately 
15:13 < kayl> xterm console of the uml, I meant 
15:13 < kayl> xterm support is on and con=xterm is used at cmd line 
15:14 < jdike> kayl: what does dmesg say?
15:14 < kayl> I can get it to work with a con=pty con1=fd:0,fd:1 as indicated on the uml site though 
15:14 < kayl> jdike: on the uml or the host?
15:15 < jdike> kayl: uml
15:17 < kayl> jdike: the uml hangs after setting the system clock nothing is outputed to indicate the xterm-console closed. I tried a ./linux con1=fd:0,fd:1 con2=xterm to be able to access a prompt on the uml to dig in logs but it didn't quite work 
15:19 < jdike> kayl: this is after the xterm pops up and disappears?
15:20 < kayl> nope, the clock msg is on console, then the xterm pops up and disappear immediately. No more logging comes on the console (console = my ssh connection to the host of the uml)
15:21 < jdike> kayl: dmesg should tell you what happened
15:22 < kayl> now I'm on the uml box with pty instead of xterm and looking in /var/log if there is indication of what happened at previous boot (where I tried the xterm consoles instead)
15:28 < jdike> kayl: make sure that you have the utilities installed
15:28 < jdike> kayl: port-helper is the one that's needed
15:29 < kayl> the default linux executable coming with the install of uml works fine with xterms, my recompiled version is more of a problem so I suspect I messed up while configuring the kernel but I went over the options with no idea on what could be responsible
15:30 < kayl> mmm would I have to recompile the utilities if I recompile a uml linux ...? (didn't think so) 
15:30 < jdike> kayl: strange, I can't think offhand of any options which would cause that
15:30 < jdike> kayl: about rebuilding utils, no
15:30 < kayl> jdike: ok I didn't thought it would help but better make sure 
15:31 -!- coryb [~cory@24.124.40.190] has quit [Quit: leaving]
15:31 < kayl> jdike: yeah pretty weird and I checked the x11 forwarding is not the problem
15:32 < jdike> kayl: if you got a window at all, X forwarding is fine
15:32 < caker> try appending console=tty1
15:38 < kayl> caker: weird! with console=fd:0,fd:1 I still don't get any more ouput in my ssh connection terminal once the drivers kick in. with console=tty1 I get a Seg fault in signals directly 
15:39 < jdike> kayl: send that one to uml-devel, along with how to reproduce it
15:39 < caker> kayl: here's the old stand-by:  ./linux con=null con0=fd:0,fd:1       and, make sure you've got a getty listening on tty0 in the uml's etc/inittab
15:40 < kayl> jdike: tomorrow I'll test this out again on the host directly (no remote connection) just to make sure and write it up to email if I didn't find out what's going on
15:40 < kayl> caker: checking it 
15:41 < jdike> caker: the real standby is just 'single' :-)
15:41 < caker> or init=/bin/bash
15:41 < jdike> caker: you'll get a shell prompt that doesn't rely on anything
15:41 < caker> jdike: good to know, thanks
15:42 < jdike> caker: that's the moral equivalent of init=/bin/bash
15:42 < jdike> caker: just slightly heavier
15:43 < jdike> caker: init=/bin/bash will get you a shell in some situations where 'single' won't
15:43 < caker> ./linux single whouldn't still use the defaults compiled into UML for console?
15:43 < jdike> caker: not really
15:43 < caker> xterms, for instance on a machine not running x?
15:43 < jdike> caker: hrm, let me think about that
15:43 < jdike> caker: yeah, you're right
15:44 < kayl> caker: nothing listening on tt0 in /etc/inittab... adding it
15:54 < davidc> hi
15:54 < caker> yo
15:55 < caker> davidc: you're not "at work" are you? :-)
15:55 < davidc> caker: no, I'm at my colo
15:55 < davidc> caker: but I got my EAD today :-)
15:55 < davidc> so if I was at 'work', it'd be legal
15:56 < caker> davidc: is that the thing you needed?
15:56 < caker> davidc: Congrats!!!
15:56 < davidc> caker: yep
15:56  * caker thinks it's time for david to get paid
15:57 < davidc> caker: indeed
15:57 < davidc> caker: btw, I think it's perfectly possible to run 96 UML on one 3k box
15:57 < davidc> caker: mucho revenue
15:58 < caker> davidc: up to 96, now? :-)
15:58 < davidc> caker: why not?
15:59 < davidc> caker: My box swaps everything useless out when updatedb is run
15:59 < caker> davidc: last night it was 80 :-)
15:59 < davidc> caker: I forgot about updatedb then though
16:01 < davidc> caker: 48 Linux instances per U is pretty nice :-)
16:01 < caker> davidc: heh
16:01 < jdike> davidc: how many U's in a rack?
16:01 < caker> How many versions of Linux can run per square foot?
16:01 < caker> :-p
16:01 < davidc> jdike: 42, but you'd leave a couple for switch and UPS
16:02 < davidc> jdike: basically 48*40 :-)
16:02 < davidc> 1920
16:02 < jdike> caker: that would be 1920 UMLs per rack footprint
16:02 < kayl> davidc: 48 uml on a 1u :) what kind of u is that? (cpu, ghz) 
16:02 < caker> hehe 19" * 1.75" * how deep? 20" ?
16:02 < davidc> kayl: er, rack unit
16:03 < davidc> caker: most racks are around 2.5ft square
16:03 < jdike> caker: == ~6 sq ft?
16:03 < davidc> 123 per sq ft
16:03 < caker> lol
16:03 < davidc> oh, wait
16:03 < davidc> 309
16:03 < kayl> davidc: ok but I thought that each vendor of rack unit had various configs. I'm curious to have an idea of what kind of cpu / ram / ghz can run 48 uml :)
16:04 < green> well, idle umls do not eat much resources ;)
16:04 < davidc> kayl; Dual 2.8Ghz Xeon with 4Gb of RAM and 8 120Gb drives in RAID-5 with a hot-spare
16:04 < kayl> green: true :)
16:04 < davidc> kayl: That'll run 96ish
16:04 < davidc> kayl: throw that in a 2U box, and you're set
16:04  * kayl drools
16:04 < davidc> You could give each UML 8Gb of space
16:04 < davidc> without overselling
16:05 < davidc> kayl: most of the time, the UMLs won't be doing much of anything
16:05 < BB> lol u tart beer at home aint same as beer in pub
16:05 < BB> doh
16:05 < kayl> davidc: stills makes uml a pretty good choice for hosting :)
16:06 < green> there is some room for improvements
16:06 < davidc> kayl: it's about as dense as you can get
16:06  * green can enumerate at least two places for optimisations ;)
16:06 < davidc> kayl: comparing 40boxes per rack to 1920 UMLs per rack
16:06 < davidc> green: where?
16:06 < caker> green: where?
16:07 < caker> oops
16:08 < green> davidc: remove two extra context switches from syscall overhead ; mmap ubd files, to avoid extra copying around and save ram
16:08 < green> davidc: will also save some IO
16:08 < jdike> yup, those are the big ones right now
16:08 < davidc> green: oh, right - Yes, they would improve matters significant
16:08 < davidc> ly
16:08 < caker> nice
16:10 < davidc> caker: you just need the bandwidth to handle 2000 UMLs :-)
16:10 < caker> davidc: no kidding!
16:11 < jdike> davidc: they're not all going to be running at once
16:11 < jdike> davidc: if they are, then the hosts have a major problem
16:12 < caker> Booting 100 UMLs would take an age
16:12 < green> jdike: one just need 100 cpu box ;) well, even 50 cpu one will be sufficient
16:13 < silug> jdike: do you know if anyone has tried to run plain old x86 uml on an opteron (running a 64-bit kernel)?
16:14 < jdike> silug: yes, that works
16:14 < silug> i was thinking that might be cool...  the current motherboards will go to either 12 or 16gb (if 2gb ddr is ever available)
16:14 < silug> jdike: neat
16:14 < davidc> silug: you can get Xeon boards which will do 12Gb of PC2100
16:14 < davidc> god knows how much that RAM will cost
16:14 < BB> yeah we did a dual zeon took 12GB
16:14 < jdike> silug: the guy who did it pointed out some 64-bit problems in the mconsole protocol
16:14 < silug> davidc: yeah, but >2gb on x86 sucks
16:14 < BB> xeon even
16:15 < silug> and the 1.4ghz opterons are actually reasonably priced compared to xeons
16:15 < green> are they available in stores already?
16:16 < silug> they're readily available...  i haven't sold any yet, but i've had several customers interested.
16:16 < green> hm, interesting
16:16 < BB> we havnt sold any either, but have the suppliers ready incase we need the,
16:16 < green> 1.4Ghz sound slow, though ;)
16:16 < BB> ,=m
16:17 < silug> 1.4ghz opteron is supposedly around the same speed as an athlon 2600+ or 2800+
16:17 < silug> give or take
16:17 < green> how so?
16:17 < BB> hmm 1.4GHz with a bus as wide as nikki's arse ;) - just dont quote me on that or I'll get busted <g>
16:17 < silug> that's what the reviews say
16:18 < silug> clock speed != performance, as amd has been trying to say for years now.  :)
16:18 < BB> but playing the cyrix trick really doesnt impress me
16:19 < green> silug: well, depends on what clock speed do you mean ;) memory clock speed is pretty important ;)
16:46 < QSECOFR> know of anyone willing to do a UML port to Opteron if hardware access is provided?
16:50 < davidc> hello
16:52 < kayl> ok guys the uml's xterms are not yet finding their way to my screen but I'll have a chance tomorrow to check onsite if this is somehow related to the specifics of the remote access... see you later and lots of thanks for the time you all spent helping me thought the cascading problems ;P
16:52 -!- kayl [~jada@6532176hfc248.tampabay.rr.com] has quit []
16:53 < QSECOFR> anyone else had lockups with br0'ing and RH80?.. or is it something specific to my box..any sucesses with rh8?
16:54 < jdike> QSECOFR: I would, if I had any time
16:54 < jdike> QSECOFR: I'd love to have a working second port of UML
16:55 < QSECOFR> well if you hve the time, i have the hardware
16:56 < QSECOFR> i dont have the time, but i dont think i have the skills either..:)
17:29 -!- ticallion [~ticallion@213.175.160.103] has joined #uml
17:30 < caker> Linus just released the last 2.5 kernel
17:35 < jdike> caker: the pre-6 series can still be a long one
17:41 -!- jdike [~jdike@dhcp136.ISTS.dartmouth.edu] has quit [Quit: Leaving]
18:06 < dilinger> has anyone been trying to get marcelo to include the skas host patch in 2.4.22?
18:25 -!- ticallion [~ticallion@213.175.160.103] has quit [Quit: Client exiting]
18:59 -!- kayl [~jada@65.32.176.248] has joined #uml
18:59 < kayl> hi again 
19:22 -!- wepy [~wepy@bgp546250bgs.ewndsr01.nj.comcast.net] has joined #uml
19:22 < wepy> hi
19:22 -!- wepy [~wepy@bgp546250bgs.ewndsr01.nj.comcast.net] has quit [Client Quit]
20:38 -!- mistral [mistral@jstevenson.plus.com] has quit [Ping timeout: 488 seconds]
20:47 -!- ticallion [~ticallion@213.175.160.210] has joined #uml
20:50 -!- ticallion is now known as ticallion[jh]
21:13 < david> hello
21:13 -!- davidc [~david@tailtiu.davidcoulson.net] has quit [Quit: davidc has no reason]
21:13 < ticallion[jh]> howdies
21:13 -!- caker [~null@pcp507591pcs.nash01.tn.comcast.net] has quit [Ping timeout: 488 seconds]

--- Log opened Thu Jul 10 22:57:39 2003
23:02 -!- caker [~null@pcp507591pcs.nash01.tn.comcast.net] has joined #uml
23:02 < david> caker: hey
23:36 < caker> david: what do I need to do to set up a mirror for uml?
23:39 < david> caker: mget everything from sourceforge.net and setup a vhost for it
23:40 < caker> david: isn't rsync involved some place?
23:41 < david> caker: I've got an rsync system setup, but Jeff isn't using me as the primary mirror right now
23:41 < caker> david: so what is the solution for updates?
23:41 < david> caker: rsync, but I need to poke jeff abt it
23:42 < david> caker: let me put rsync up on my firewall so you can rsync from me with a cron
23:42 < david> uh
23:42 < david> I can't remember which box I put rsync on
23:43 < caker> can't remember which one out of 1920? :-)
23:43 < david> ok
23:44 < david> rsync -avc ssh-gate.davidcoulson.net::umlmirror
23:44 < caker> working .. -- how big is this? :-)
23:47 < david> 3.6Gb
23:47 < caker> ok .. still receiving file list ...
23:47 < david> yeah, 434 files
23:49 < david> wow
23:49 < david> rock
23:49 < caker> What's that?
23:49 < david> UML 0.4.0-devel actually works
23:49 < caker> heh
23:49 < david> uml boot default /bin/sh
23:49 < david> uh
23:49 < david> uml <umid> boot default /bin/sh
23:49 < david> brings up the uml with /bin/sh as init
23:56 < caker> cool
23:56 < caker> hey, been 10 minutes, still on file list
23:57 < david> yeah, it checksums everything I think
--- Log closed Fri Jul 11 00:00:00 2003