--- Day changed --- Log opened Wed May 04 23:59:03 2005 00:23 -!- lilo [~lilo@lilo.usercloak.oftc.net] has joined #xen 00:36 -!- TheQ [~TheQ@dial4a-65.btc-bci.com] has quit [Ping timeout: 480 seconds] 01:04 -!- superdave591 [~superdave@danders5.cs.uiuc.edu] has quit [Quit: Leaving] 01:32 -!- lilo [~lilo@lilo.usercloak.oftc.net] has quit [Quit: bbiab new kernel] 01:35 -!- lilo [~lilo@lilo.usercloak.oftc.net] has joined #xen 01:38 -!- lilo [~lilo@lilo.usercloak.oftc.net] has quit [Quit: ] 01:42 -!- lilo [~lilo@lilo.usercloak.oftc.net] has joined #xen 01:47 -!- lilo_ [~lilo@lilo.usercloak.oftc.net] has joined #xen 01:47 -!- lilo [~lilo@lilo.usercloak.oftc.net] has quit [Read error: Connection reset by peer] 01:58 < Squink> What's the difference between XenU and Xen0, anyone? 01:59 < rusty> Squink: Xen0 is the modified kernel which runs in domain 0 (speaks to real devices), and XenU is the modified kjernel which runs elsewhere. 01:59 < rusty> Squink: Xen0 is in charge. 02:00 < Squink> so Xen0 would be the one I'd want to boot into on the host box? 02:01 < Squink> (Since I'm guessing I can't just run an unmodified kernel, given the errors it just spat out at me when I ran xend start) 02:05 < rusty> Yes. 02:15 < cartel> hey 02:15 < cartel> isnt XenU the fictional entity in scientology manuscripts? 02:16 < cartel> and isnt lilo the guy that runs freenode? 02:16 < cartel> ITS ALL A HUGE CONSPIRACY!!!!! 02:16 < cartel> ARRGH!!!!!! 02:17 < Squink> lol 02:19 -!- tierra [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has quit [Quit: bug, n: A son of a glitch.] 02:28 < Squink> hrm. in the grub.conf, what should I set dom0_mem as? (I have 512mb) 02:29 < rusty> Squink: IIRC I used 128M 02:29 < Squink> I guess more to the point; is it going to matter what I set it as? :) 02:31 < Squink> Basically this if my first time playing with Xen. But the only access I have is via ssh, so I *Really* don't want to screw this up :) 02:34 < cartel> Squink: have you rebooted into your xen0 yet? 02:34 < Squink> not yet, that's what I'm mostly concerned about 02:34 < Squink> I'm just staring at my grub.conf at the moment. 02:35 -!- Tv [~Tv@hq.inoi.fi] has joined #xen 02:35 < Squink> And still trying to get my head around the concept of Xen. I was using UML, but I ended up here because of the overhead 02:35 < cartel> hrm\ 02:35 < cartel> be careful... it took me 3 or 4 goes to get my xen0 right 02:36 < Squink> well all the binaries are ready 02:36 < Squink> I just need to config grub and reboot.. 02:36 < cartel> but did you compile everything you need? 02:36 -!- betonamu [~Administr@dhcp-243-054.mag.keio.ac.jp] has left #xen [] 02:37 < Squink> urm. Well, I compiled the xen0 kernel, and also ran install.sh, and that said everything was ok 02:37 -!- betonamu [~Administr@dhcp-243-054.mag.keio.ac.jp] has joined #xen 02:37 < Squink> not sure what else I might need... 02:37 < cartel> drivers, filesystems.... 02:37 < cartel> bridge support... 02:37 < Squink> I have a filesystem ready to go which I was using in UML 02:38 < cartel> no, filesystem support in kernel.. enough to boot your box 02:38 < Squink> But otherwise I've just been following the HOWTO guides.. 02:38 * Squink stares blankly 02:39 < cartel> Squink: put it this way, if your box dies how long does it take to get someone there and reboot it or whatever? 02:39 < Squink> 10-30 mins, I guess 02:39 < Squink> I was sort of under the impression that I would just boot into xen0 kernel, and things would operate as they usually do on the host box 02:39 < Squink> but then, I'm prone to misunderstanding :) 02:39 < cartel> yep, thats pretty much how it goes 02:40 < Squink> so essentially I just need to whack my new grub entry into the top of grub.conf, reboot, and it'll work. theoretically. 02:40 < Squink> right? :) 02:41 < cartel> assuming u got it all configured right yeah :) 02:41 < Squink> heh. Any immediate things you can think of that would need configuring? 02:42 < cartel> just triple check your kernel config :) 02:42 < Squink> in grub or the make config? 02:42 < cartel> make menuconfig yea 02:42 < Squink> ah, then couldnt i use one of my current .config files? 02:43 < cartel> not without adding xen support 02:43 < cartel> but yes, you can do that 02:43 < Squink> sort of a make menuconfig oldconfig thing 02:43 < cartel> copy in your old config, make menuconfig on it.. then set ARCH=xen, set privileged guest etc 02:43 < betonamu> hi. anybody pls tell me if i can put those parameters in xen configuration file (of domU) on the commandline (after "xm create...") ? 02:44 < betonamu> or i must always put those params into a config file? 02:44 < cartel> betonamu: i think they have to be on config 02:45 < betonamu> hmm.. too bad 02:48 < Squink> cartel: hrm. it doesnt want to work. Where should I put .config if I want it to get picked up? I've got it in /xen-2.0-testing/linux-2.6.11-xen0/.config at the moment 02:51 -!- zsting [~sting@mvil2.bb.netvision.net.il] has joined #xen 02:52 < Squink> cartel: or thinking about it, would you recommend a Yum install of xen over anything I've already done? 02:54 -!- rusty [~rusty@bh02i525f01.au.ibm.com] has quit [Quit: Client exiting] 03:01 < Squink> nevermind, figured it out :) 03:05 < cartel> :) 03:23 < betonamu> in the Xen report found at the xensource, i see this line: 03:23 < betonamu> todo: Replace monolithic xend with tool suite communicating via The Registry 03:23 < betonamu> what is "the registry" in this case? 03:24 < cartel> i have no idea 03:25 < cartel> Linux Registry? 03:26 < betonamu> cartel: i dont know 03:26 < betonamu> dont understand at all 03:26 < betonamu> take a look: http://www.cl.cam.ac.uk/netos/papers/ian-status.ppt 03:28 < cartel> POWERPOINT!!!!!! No way! 03:28 * cartel barfs 03:28 < Squink> lol 03:29 * cartel writes a knote: 'Bash Ian for using powerpoint' 03:29 < betonamu> ask ian then ;) 03:29 -!- zsting [~sting@mvil2.bb.netvision.net.il] has quit [Quit: ] 03:30 < cartel> he should be using OOo Impress :) 03:30 < betonamu> that is not the only one: http://www.cl.cam.ac.uk/netos/papers/xen-lwe2005-short.ppt 03:30 < betonamu> :D 03:30 < betonamu> :-D 03:30 < cartel> xen also uses bitkeeper 03:30 < cartel> D: 03:39 < betonamu> i heard that they are rewriting xend? anybody knows about this? 04:26 -!- Nigelenki [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has quit [Read error: Operation timed out] 04:38 -!- athomas [~athomas@ppp-0-151.lond-a-3.access.uk.tiscali.com] has joined #xen 04:43 -!- athomas_ [~athomas@ppp-0-151.lond-a-3.access.uk.tiscali.com] has joined #xen 04:44 -!- athomas [~athomas@ppp-0-151.lond-a-3.access.uk.tiscali.com] has quit [Quit: Leaving] 05:03 -!- Bluefox [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has joined #xen 05:07 < Tv> I love xen & lvm. 05:10 -!- zsting [~sting@mvil2.bb.netvision.net.il] has joined #xen 05:13 -!- Bluefox [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has quit [Read error: Operation timed out] 05:16 -!- Bluefox [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has joined #xen 05:20 -!- Bluefox [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has quit [Read error: Operation timed out] 05:55 -!- Bluefox [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has joined #xen 06:26 -!- trt [~trt@dD5E04B7A.access.telenet.be] has joined #xen 07:12 -!- trt [~trt@dD5E04B7A.access.telenet.be] has left #xen [Leaving] 07:27 -!- Nigelenki [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has joined #xen 07:27 -!- Bluefox [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has quit [Read error: Connection reset by peer] 08:10 -!- cfreak [cfreak@dsl-084-056-114-005.arcor-ip.net] has joined #xen 08:12 -!- Nigelenki [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has quit [Read error: Operation timed out] 08:49 -!- katzj [~katzj@c-66-30-206-55.hsd1.ma.comcast.net] has quit [Read error: Operation timed out] 08:52 -!- hollis [~hollis@user-0vvde2g.cable.mindspring.com] has joined #xen 08:59 -!- shredda [~shredda@sential.de] has joined #xen 09:11 -!- Bluefox [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has joined #xen 09:50 -!- katzj [~katzj@wlanconf-nat-pool-bos.redhat.com] has joined #xen 10:37 -!- DEac- [~deac@xdsl-81-173-161-32.netcologne.de] has quit [Ping timeout: 480 seconds] 10:42 -!- hollis [~hollis@user-0vvde2g.cable.mindspring.com] has quit [Quit: leaving] 10:48 -!- heb [~harry@blueice4n1.uk.ibm.com] has quit [Quit: Leaving] 10:50 -!- DEac- [~deac@xdsl-84-44-215-218.netcologne.de] has joined #xen 10:53 -!- shredda [~shredda@sential.de] has quit [Quit: shredda] 10:54 -!- muli [~muli@alhambra.mulix.org] has quit [Read error: Operation timed out] 11:05 -!- hollis [~hollis@pixpat.austin.ibm.com] has joined #xen 11:15 -!- Squink [~purple@S010600131077a67e.vc.shawcable.net] has quit [Ping timeout: 480 seconds] 12:04 -!- athomas_ [~athomas@ppp-0-151.lond-a-3.access.uk.tiscali.com] has quit [Ping timeout: 480 seconds] 12:05 -!- athomas [~athomas@ppp-0-91.lond-b-1.access.uk.tiscali.com] has joined #xen 12:31 -!- muli [~muli@alhambra.mulix.org] has joined #xen 12:36 -!- stekloff [~stekloff@bi01p1.co.us.ibm.com] has joined #xen 12:55 -!- liberie [~root@dsl027-160-029.atl1.dsl.speakeasy.net] has joined #xen 13:10 -!- niv [~nivedita@bi01p1.co.us.ibm.com] has joined #xen 13:36 -!- athomas [~athomas@ppp-0-91.lond-b-1.access.uk.tiscali.com] has quit [Quit: Leaving] 13:44 -!- tierra [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has joined #xen 15:00 -!- superdave591 [~superdave@danders5.cs.uiuc.edu] has joined #xen 15:10 -!- _MarkW [~MarkWilli@hebble.cl.cam.ac.uk] has joined #xen 15:12 -!- _MarkW [~MarkWilli@hebble.cl.cam.ac.uk] has quit [Remote host closed the connection] --- Log closed Thu May 05 15:56:27 2005 --- Log opened Thu May 05 15:56:39 2005 --- Log closed Thu May 05 16:00:39 2005 --- Log opened Thu May 05 16:04:11 2005 --- Log closed Thu May 05 16:04:18 2005 --- Log opened Thu May 05 16:04:24 2005 --- Log closed Thu May 05 16:04:28 2005 --- Log opened Thu May 05 16:04:33 2005 --- Log closed Thu May 05 16:04:38 2005 --- Log opened Thu May 05 16:04:43 2005 --- Log closed Thu May 05 16:04:48 2005 --- Log opened Thu May 05 16:04:53 2005 --- Log closed Thu May 05 16:04:58 2005 --- Log opened Thu May 05 16:05:04 2005 --- Log closed Thu May 05 16:05:08 2005 --- Log opened Thu May 05 16:05:14 2005 --- Log closed Thu May 05 16:05:17 2005 --- Log opened Thu May 05 19:48:07 2005 19:49 < cartel> hmm 19:50 -!- niv [~nivedita@bi01p1.co.us.ibm.com] has quit [Quit: Quitting] 20:01 -!- hollis [~hollis@user-0vvde2g.cable.mindspring.com] has joined #xen 20:21 -!- hollis [~hollis@user-0vvde2g.cable.mindspring.com] has quit [Quit: Lost terminal] 20:33 < _MarkW> cartel: hmmm? 20:35 < knewt> _MarkW: i'm getting an error trying to build -unstable 20:36 < _MarkW> knewt: pulled when? what's the error? 20:37 < knewt> 20050505 175859 20:37 < knewt> 22:44 < knewt> arch/xen/i386/kernel/built-in.o(.init.text+0x3ec1): In function `simd_math_error': 20:37 < knewt> 22:44 < knewt> include/asm/bitops.h:252: undefined reference to `pmtmr_ioport' 20:37 < knewt> 22:44 < knewt> arch/xen/i386/kernel/built-in.o(.init.text+0x3ec6):include/asm/bitops.h:251: undefined reference to `pmtmr_ioport' 20:37 < _MarkW> ewww 20:37 < knewt> top of the changelog is: Clean up IO-APIC handling in domain0, and the hypercall interface exported by Xen (inc. removal of PCI-related calls). This hopefully fixes acpi=off. 20:38 < knewt> i got the same error with 20050505 102922 20:38 * _MarkW looks... 20:39 -!- katzj [~katzj@c-66-30-206-55.hsd1.ma.comcast.net] has joined #xen 20:39 < _MarkW> how did you actually download this? BK? 20:39 < knewt> BK using sfioball 20:39 < _MarkW> using what??? :-) 20:40 < knewt> the client LM released initially under the "No Whining Licence" 20:40 < _MarkW> Ah :-) 20:40 < knewt> all it can do is pull in a particular revision or the head 20:40 < knewt> and update an existing pull 20:40 < knewt> nothing else 20:40 < _MarkW> have you built previous checkouts in this tree? could be that a make clean is required... 20:41 < _MarkW> that tree built fine for me just now 20:41 < knewt> got the error with a fresh pull, and 2.6.11.8. i could email you the .config i'm using if you want? 20:42 < _MarkW> sure, i'll give it a go. or stick it online somewhere. 20:45 < knewt> test.pimb.org/unstable.config 20:46 < knewt> took longer than it could have there because i kept trying to send it to the wrong box, and the destination path didn't exist *g* 20:46 < _MarkW> :-) 20:46 * _MarkW clones a new tree... 20:47 < knewt> i've set my local xen area up with xen.unstable.[long revision code], then a symlink to that of the form xen.unstable.[date/time], then a symlink xen.unstable to the latest date/time 20:47 < knewt> same with -testing 20:48 < _MarkW> this will all be a good test for by new build system... 20:48 < _MarkW> what gcc are you on btw? 20:48 < knewt> i run the build directly from within the linux folder btw, i don't use any of the xen makefiles 20:48 < knewt> gcc.real (GCC) 3.3.5 (Debian 1:3.3.5-8ubuntu2) 20:49 < _MarkW> ok 20:53 < _MarkW> knewt: you build a lot of drivers ;-) 20:54 < knewt> i just took the existing distro config, modified it to xen, and removed some of the worst "never going to need them" things 20:55 < _MarkW> explains why it's taking a while! 20:55 < knewt> i built gtk2 last night. now /that/ took a while :) 20:56 < _MarkW> shouldn't be much longer though... i have 3 boxes doing compilation with distcc, plus all the stuff in the default Xen config should be in my ccache 20:59 < knewt> this is for my laptop you see, not a server 21:00 < knewt> my servers i generally build with only what i need, built in, and no module support 21:00 < _MarkW> ah, nice 21:00 < _MarkW> I get the same error :-) 21:00 < _MarkW> must be something weird in your config that's upsetting it... 21:02 < knewt> i actually turned off a couple of items previously that also caused build problems, but i could track them down easily and to stuff i knew i didn't actually need 21:02 < knewt> one of the irda net drivers gave errors for instance 21:02 < _MarkW> i think i'm getting close... 21:04 < _MarkW> knewt: go to the ACPI config and disable X86_PM_TIMER 21:04 < _MarkW> in your config 21:05 < _MarkW> probably worth also reporting it to xen-devel as something that breaks... 21:15 < _MarkW> knewt: it builds for me with that config, so hopefully it'll also fix it for you. 21:16 < _MarkW> i guess the acpi timer may not be a straightforward issue when running on Xen 21:18 < knewt> ok, yep. built ok right to the end 21:18 < _MarkW> now see if it boots :-) 21:18 < knewt> not going to do that tonight 21:18 < _MarkW> it is a bit late... 21:19 < knewt> i'm watching the election, and chatting on irc and what-not 21:19 -!- hollis [~hollis@user-0vvde2g.cable.mindspring.com] has joined #xen 21:19 < knewt> tony blair just got re-elected to his constituency *g* 21:19 < _MarkW> i'm monitoring the election on the bbc news website 21:20 < knewt> _MarkW: ok, could you try adding this item in: Device Drivers -> Networking -> IrDA -> Device Drivers -> Toshiba Type-O IR Port (TOSHIBA_FIR) 21:21 < knewt> i get this: 21:21 < knewt> include/asm-xen/xen-public/io/ring.h:143:1: warning: "RING_SIZE" redefined 21:21 < knewt> drivers/net/irda/donauboe.c:88:1: warning: this is the location of the previous definition 21:21 < knewt> drivers/net/irda/donauboe.c: In function `toshoboe_startchip': 21:21 < knewt> drivers/net/irda/donauboe.c:568: error: `RING_SIZE' undeclared (first use in this function) 21:21 < _MarkW> ah, yes, you would 21:21 < _MarkW> RING_SIZE is used by all sorts of Xen-related headers 21:22 < _MarkW> I don't need to test build that to tell you what the problem is ;-) 21:22 < knewt> i wonder if just changing RING_SIZE in the couple of places it appears to be in donauboe.c file to something else will fix it 21:22 < _MarkW> Could you post that error to the list too? 21:23 < knewt> prob would, but not the best fix 21:23 < _MarkW> knewt: certainly it will fix it 21:23 < _MarkW> but it probably means we need to either rename the macro (e.g. to XEN_RING_SIZE) 21:23 < _MarkW> or find some way of the driver not seeing its definition. 21:24 < _MarkW> you needn't even change the names in the driver file, just #undef RING_SIZE at the top 21:24 < _MarkW> that's a hack but it should work 21:27 < knewt> nope, because it's trying to redefine it in ring.h, not in donauboe.c 21:27 < _MarkW> oh well, renaming it is then :-( 21:27 < knewt> i'll try adding back the things i removed and figure out what the problems are, and then post to the list 21:27 -!- tierra [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has quit [Quit: bug, n: A son of a glitch.] 21:27 < _MarkW> it'd be really useful if you could do that! thanks. 21:28 -!- Icy [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has quit [Read error: Connection reset by peer] 21:30 -!- Bluefox [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has joined #xen 21:40 -!- monrad [~monrad@213083190130.sonofon.dk] has quit [Ping timeout: 480 seconds] 21:59 -!- Bluefox [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has quit [Quit: http://ubuntulinux.org/ | http://gentoo.org] 21:59 -!- Bluefox [~bluefox@pcp0012067827pcs.whtmrs01.md.comcast.net] has joined #xen 22:22 -!- hollis [~hollis@user-0vvde2g.cable.mindspring.com] has quit [Quit: leaving] 22:28 -!- spike723 [~dyeske@24.159.107.193] has joined #xen 22:29 < spike723> anyone home? 22:29 -!- tierra [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has joined #xen 22:31 < _MarkW> spike723: *ping* 22:31 < _MarkW> wassup? 22:32 < knewt> _MarkW: you're still up ?!?!?! 22:32 < _MarkW> knewt: You're still up!?!?!?!? :-p 22:32 < spike723> is it possible to run xen so that my other hosts on the network can access the ip addresses of a domain without having to set a route? 22:32 < _MarkW> spike723: yes, the bridged mode should do that for you 22:32 < Robot101> spike723: proxy ARP? 22:32 < cartel> spike723: in default configuration it does bridging 22:33 < spike723> well I'm running the demo cd 22:33 < knewt> well, with ENOJOB i don't particularly have any issues with staying up 22:33 < cartel> ENOJOB? 22:33 < _MarkW> knewt: same for me, sort of... 22:33 < cartel> is that a race condition? 22:33 < cartel> ;) 22:33 < _MarkW> spike723: I'm not so sure about the demo CD, haven't used it much 22:33 < spike723> it gave me an ip address 10.171.45.1 xen-br0 and 10.171.45.2 for my domain ip address 22:34 < knewt> cartel: i certainly appear to be in a deadlock 22:34 < spike723> in order for this to work the way I want, then xen-br0 can't have an ip address? 22:34 < cartel> knewt: took me 2 years to land my crummy job 22:34 < _MarkW> spike723: xen-br0 can have an ip address, that's ok 22:35 < _MarkW> and if eth0 is attached to the bridge xen-br0, then your domains will be bridged with the network 22:35 < spike723> how can I check for that? 22:37 < _MarkW> brctl will tell you, i can remember the command 22:37 < _MarkW> "brctl list", perhaps 22:37 < spike723> just a sec 22:43 < spike723> xendemo:~# brctl show 22:43 < spike723> bridge name bridge id STP enabled interfaces 22:43 < spike723> xen-br0 8000.feffffffffff no vif1.0 22:43 < _MarkW> ah, so eth0 isn't on it 22:44 < _MarkW> use "brctl add" 22:44 < spike723> hmm 22:45 < _MarkW> you'll need to bring up eth0 (e.g. using ifconfig) first 22:45 < spike723> I ran dhcpcd on eth0 22:45 < _MarkW> ok 22:45 < _MarkW> bad things will happen if you add an interface with an ip address to the bridge 22:45 < spike723> ok that just took my interface down 22:46 < _MarkW> sorry - too late :-( 22:46 < spike723> ok eth0 on my host needs to have an ip address 22:46 < spike723> and eth0 on my domain0 needs to have an ip 22:46 < _MarkW> spike723: don't give eth0 an ip address. add it to the bridge, then give the bridge an ip address. 22:46 < demon> you need to set its IP to 0.0.0.0, add eth0 to the bridge, then run the DHCP client on the bridge 22:46 < spike723> hmm just a sec 22:46 < _MarkW> don't ask me why the bridge people designed it like that :-) 22:47 < spike723> I'm a FreeBSD developer btw ;) 22:47 < _MarkW> we've been hoping someone will send a patch to the bridge people and make its behaviour more sane 22:47 < _MarkW> spike723: ah, i see. you know Kip? 22:48 < spike723> nah who is kip? 22:48 < _MarkW> Kip Macy, he did the FreeBSD port to Xen 22:48 < spike723> does he irc? 22:48 < _MarkW> I don't think so... I haven't seen him here, anyhow. 22:48 < spike723> I maintain about 45 FreeBSD ports, but I'm not a committer 22:48 < spike723> but I'm very interested in xen 22:48 < spike723> since i use netgraph and honeyd a lot 22:49 < _MarkW> ok 22:49 < _MarkW> hence your enthusiasm for making the network work :-) 22:50 * cartel installs kubuntu into a xenU 22:50 < demon> isn't kUbuntu just ubuntu + KDE? 22:50 < cartel> yep 22:51 < _MarkW> kubuntu can be installed as a separate entity though, even though they're basically the same 22:51 < spike723> ok check this out 22:51 < spike723> xendemo:~# brctl show 22:51 < spike723> bridge name bridge id STP enabled interfaces 22:51 < spike723> xen-br0 8000.000c297e61d4 no eth0 22:51 < spike723> vif1.0 22:51 * knewt considers attempting to try out Debian GNU/kFreeBSD in a domU 22:51 < spike723> I set eth0 to 0.0.0.0 22:52 < spike723> I ran dhcpcd on xen-br0 22:52 < _MarkW> the really neat thing is that "kubuntu" is also a word in some language - that must be the first time a KDE-relating thing had a proper name ;-) 22:52 < spike723> can I now run dhcpcd on eth0 for domain0? 22:52 < _MarkW> spike723: that sounds promising. 22:52 < _MarkW> you don't need to 22:52 < demon> spike723: you should be able to 22:52 < spike723> I don't need to? 22:52 < _MarkW> domain0 should route through the bridge device itself 22:52 < _MarkW> (you might have to add a route to tell it that, though) 22:52 < knewt> spike723: the bridge takes the place of eth0 for access to the host itself 22:53 < _MarkW> knewt: debian kFreeBSD has a Xen .deb 22:53 < _MarkW> and you can use "crosshurd" to install the GNU/FreeBSD userland into a chroot 22:53 < spike723> but domain0 has an ip address of 10.171.45.2 22:53 < _MarkW> knewt: don't ask me if anyone has ever done it though! 22:54 < _MarkW> spike723: sorry, i don't quite follow 22:54 < _MarkW> spike723: with linux bridging, once you've stuck eth0 on the bridge, you should use xen-br0 wherever you previously would have used eth0 22:54 < spike723> yeah I did that 22:54 < _MarkW> xen-br0 behaves like an ethernet device as well as a bridge 22:55 < spike723> I ran dhcpcd on xen-br0 and got a lease 22:55 < _MarkW> ok 22:55 < demon> then you're done with networking in dom0 22:55 < knewt> ok. so what's the problem? 22:55 < spike723> now I must run dhcpcd on eth0 on domain0? 22:55 < demon> no 22:55 < knewt> no 22:55 < knewt> why would you need to? 22:55 < demon> eth0's IP must remain 0.0.0.0 22:55 < spike723> ok lets get something strait 22:55 < demon> otherwise the bridging won't work 22:56 < knewt> once eth0 is part of the bridge, forget about it. just think about xen-br0 from then on 22:56 < spike723> eth0 on my host machine should be called what? 22:56 < cartel> demon: it can also have the same ip as the bridge and it will still work 22:56 < cartel> eth0 22:56 < demon> cartel: no, actually it won't 22:56 < spike723> how should I refer to eth0 on my domain0? 22:56 < _MarkW> spike723: are the terminologies confusing things? 22:56 < demon> because eth0 won't catch all packets normally 22:56 < _MarkW> domain0 == host 22:56 < demon> so you'll have networking troubles if you try to do it that way 22:56 < _MarkW> spike723: i.e. domain0 is where you configured the bridge 22:57 < demon> I've run into it myself when putting xen on a box 22:57 < cartel> demon: oh ok 22:57 < _MarkW> you can run dhcpd on *another* domain's eth0 22:57 < knewt> spike723: once you add eth0 to the bridge, you forget about eth0 from then on, and just think about the bridge 22:57 < demon> when the IP of an interface is 0.0.0.0, that makes it implicitly a promiscuous interface 22:57 < demon> i.e., it'll catch every packet on the wire 22:57 < demon> normally it'll just catch packets with its own MAC address 22:58 < spike723> yeah I'm talking about eth0 for domain0 though 22:59 < _MarkW> spike723: why do you want it to have an IP? 22:59 < _MarkW> it doesn't need one 22:59 < demon> spike: stop thinking of eth0 as a separate interface 22:59 < demon> the idea of the bridge device is to enslave multiple interfaces undera software bridge 22:59 * knewt notes that he's pointed twice now that you shouldn't think about eth0 once you've added it to the bridge 22:59 < knewt> s/twice/out twice/ 22:59 < spike723> how do network applications within domain0 use the network without an ip address? 22:59 < demon> you have one 23:00 < knewt> spike723: you have one. on the bridge. 23:00 < _MarkW> spike723: the bridge has an IP address 23:00 < demon> the one associated with xen-br0 23:00 < spike723> ok I'm not using the correct terminology 23:00 < demon> network applications don't care about the type of interface 23:00 < _MarkW> and the bridge *is* now dom0's ethernet device 23:00 < demon> they don't care, as long as there's a device that has an IP to act as a communication endpoint, that's all that matters 23:01 < _MarkW> A linux bridge behaves like a hardware switch *and* an ethernet card, all in one. 23:01 < spike723> I bridged eth0 and xen-br0, and then I ran dhcpcd on xen-br0 and got a lease 23:01 < demon> yes 23:01 < spike723> then I logged into domain0 23:01 < _MarkW> spike723: that's fine. 23:01 < knewt> ok, that's fine 23:01 < spike723> and I now must run dhcpcd within domain0 23:01 < demon> no 23:01 < demon> you've already run it on xen-br0 23:01 < demon> which is the bridge device in dom0 23:01 < _MarkW> spike723: i'm pretty sure you don't mean "domain0" 23:02 < knewt> spike723: eh? you had to already /be/ logged into domain 0 to run dhcpd on xen-br0 23:02 < _MarkW> knewt: exactly 23:02 < spike723> I need to have 2 ip addresses 23:02 < demon> you aren't going to get 2 IP addresses via DHCP 23:02 < knewt> spike723: are you sure you don't mean your guest domain? 23:03 < spike723> knewt netbsd would be domain0 or my guest domain? 23:03 < demon> unless you're talking in an unprivileged domain 23:03 < _MarkW> spike723: netbsd is a guest 23:03 < demon> that's an unprivileged domain 23:03 < demon> domU 23:03 < knewt> spike723: domain0 is whatever you boot into 23:03 < spike723> ok I was using the wrong terminology 23:03 < _MarkW> spike723: go ahead and run dhclient in the guest 23:03 < spike723> I have xen-br0 running dhcpcd on domain0 23:03 < demon> and yes, you can run dhcpcd/dhclient/whatever in there 23:03 < knewt> yes, run dhclient in the guest 23:04 < demon> and it should get an IP 23:04 < knewt> that's fine 23:04 < spike723> and I just ran dhcpcd on my guest 23:04 < spike723> I knew I was right :), I was just using the wrong terminology 23:05 < _MarkW> spike723: i take it that it all worked, then? 23:05 < spike723> what distro do yall suggest running for domain0? 23:05 < spike723> I'm testing it now 23:05 < demon> I personally prefer debian, but I'm sure there are people here who'll tell you some RH variant is best 23:05 < _MarkW> spike723: doesn't really matter - use whatever you like best. 23:05 < knewt> spike723: the domain you boot into is always domain 0. every time you boot a new guest domain after that (whether unprivileged or privileged) it gets a incrementing ID number (so the first you boot would be domain 1, then domain 2, and so on. the generic name for guest domains is domU) 23:06 < _MarkW> debian has Xen packages, SuSE 9.3 has packages, other distros you may need to compile Xen for yourself 23:06 < knewt> spike723: so when someone says dom[ain]0, they're always talking about the domain you booted the box into 23:06 < spike723> ok 23:06 < spike723> right now domain0 must be linux? 23:06 < knewt> netbsd is also available as dom0 23:06 < spike723> but domain1 can be netbsd, freebsd, or linux? 23:06 < _MarkW> spike723: on the democd it's linux 2.6 23:06 < spike723> whoah 23:06 < _MarkW> linux 2.4 and netbsd-current can run as dom0 23:06 < spike723> netbsd can run dom0? 23:07 < knewt> yep 23:07 < spike723> phat 23:07 < demon> oh? with xen 2.x, netbsd can be dom0? 23:07 < demon> nice. 23:07 < knewt> nowadays it can, yep 23:07 < demon> excellent 23:07 < knewt> not as tried and tested as linux yet of course 23:07 < demon> not that I'll be ethe one using it, but hey, why not :) 23:07 < knewt> but people are running it 23:08 < _MarkW> netbsd use Xen internally to manage their servers 23:08 < _MarkW> so they have a vested interest in making it good :-) 23:08 < demon> they were really stuck on xen 1.2 for awhile 23:08 < knewt> _MarkW: is that old Xen or have they migrated those servers to 2.x ? 23:08 < _MarkW> demon: that's because they didn't want to run a linux dom0, i guess :-) 23:09 < _MarkW> knewt: i don't know what they're running right now. i wouldn't be surprised if they were still on 1.2 23:09 < _MarkW> but NetBSD 3.0 is due in a couple of months 23:10 * knewt just wishes he'd have a job and so be able to afford a nice new VT system when the processors come out soon 23:11 < _MarkW> it's relatively unfortunate that netbsd 3.0 is coming out before xen 3.0 though 23:11 < _MarkW> i wonder what version they'll end up supporting... 23:12 < demon> what? netbsd 2.0 just came out not long ago... why the change in versioning? 23:12 < _MarkW> dunno, although everyone seems to be moving to shorter release cycles these days. 23:12 < _MarkW> freebsd (and arguably linux) also 23:13 < knewt> _MarkW: are these warnings important at all? 23:13 < knewt> *** Warning: "pm_power_off" [drivers/char/ipmi/ipmi_poweroff.ko] undefined! 23:13 < knewt> *** Warning: "sys_munlock" [arch/xen/i386/kernel/microcode.ko] undefined! 23:13 < knewt> *** Warning: "sys_mlock" [arch/xen/i386/kernel/microcode.ko] undefined! 23:13 < _MarkW> knewt: i can't say i particularly like them. 23:13 < demon> well, it didn't do so well for freebsd... 5.3 has given me fits 23:13 < _MarkW> knewt: i'm surprised about the second two, but not about the first one 23:14 < _MarkW> you don't need those modules, so you can rm them if you're worried 23:14 < knewt> my system is set up to load the latest microcode it's got into the processor at boot 23:14 < _MarkW> demon: 5 took them ages, they want 6 to come sooner with less upheaval 23:14 < _MarkW> knewt: alright then, you probably want that one then! 23:15 < _MarkW> I'm not sure why it's doing that. Does it work? 23:15 < knewt> won't know until i reboot tomorrow 23:15 < knewt> well, "later today" now :) 23:15 < demon> ... has there been any talk of getting dragonflyBSD under xen? 23:15 < _MarkW> demon: there's been talk 23:16 < _MarkW> kip macy started off doing just that 23:16 < demon> ah 23:16 < _MarkW> but then found that netapp would let him play with Xen/FreeBSD *at work* 23:16 < knewt> heh 23:16 < demon> ... you'd think they wouldn't mind dragonflyBSD, since it is a freeBSD offshoot... 23:16 < _MarkW> so he did that instead. there'll be common code, it just needs someone to step up and do it... 23:17 < demon> but oh well, I don't know how things work there 23:17 < demon> well, it's probably fairly different now 23:17 < _MarkW> demon: they probably use FreeBSD 5 anyhow, and want to stay the same 23:17 < knewt> _MarkW: how does vmware work, do you know? 23:17 < _MarkW> plus FBSD5 has diverged since dfly forked 23:17 < demon> since dragonflyBSD was based on 4.x 23:17 < _MarkW> knewt: something like this - 23:18 < _MarkW> for userspace, it executes code directly. vmware intercepts system calls. 23:18 < spike723> ok how do I get netbsd running as domain0? 23:18 < spike723> does it only work for netbsd i386? 23:18 < _MarkW> knewt: that's the easy part. for performance reasons, they want to execute kernel code directly. 23:19 < _MarkW> knewt: but because of x86 having various untrappable instructions that you really *want* to trap, they have to do lots of grim scanning and rewriting of the code 23:19 < spike723> do I need to build xen from cvs to get netbsd as domain0? 23:19 < _MarkW> knewt: that's where a lot of the real smarts are. if you look at QEmu with the accelerator module, it executes user code (more or less) directly but kernel code is entirely emulated 23:20 < _MarkW> knewt: Xen executes everything directly because we don't need to trap the weirdass instructions :-) 23:20 < knewt> it's a pity it's not possible to run vmware under xen 23:20 < _MarkW> spike723: there's a handy page for that... 23:20 < _MarkW> knewt: it *might* be possible technically to port vmware's kernel module to XenLinux 23:21 < spike723> is the repo only available via bk/ 23:21 < spike723> bk? 23:21 < _MarkW> but I dunno. 23:21 * spike723 notes subversion is nice 23:21 < _MarkW> spike723: you can get snapshots of the website if you don't want to use BK 23:21 < spike723> oh where is the "handy page"? 23:21 < _MarkW> or use the free BK client (knewt - what was it called) 23:21 < _MarkW> spike723: i'm looking for it now... 23:22 * spike723 starting to get a little excited about xen 23:22 < knewt> sfio/sfioball 23:22 < _MarkW> spike723: homepage http://www.netbsd.org/Ports/xen/ 23:22 < spike723> if I can run netbsd as domain0 and run freebsd as domain1 that would save me a lot of time 23:22 < _MarkW> there's a link to a howto. basically it's "install netbsd, install the xen packages, run stuff..." 23:22 < knewt> http://www.bitmover.com/bk-client.shar 23:23 < _MarkW> knewt: thanks. spike723 you might find that client useful if you don't want to use bk 23:23 < _MarkW> knewt: that microcode driver might work. it's calling system call functions from within the kernel, which seems a little strange. 23:24 < knewt> maybe if i build it in instead of as a module? 23:24 < spike723> so xen can only do i386 for domain0 and domain1 for now? 23:24 < _MarkW> knewt: i'm pretty sure that wouldn't complain. 23:24 < spike723> err 23:24 < _MarkW> i don't know if it'd work as a module, although i'd hawe thought it did. 23:24 < spike723> xen can only do i386 for domain0? 23:24 < _MarkW> spike723: yup 23:24 < knewt> spike723: x86-64 and ia64 are in development 23:24 < _MarkW> x86_64 coming with Xen 3.0 23:25 < _MarkW> ia64 and PPC coming with later releases 23:25 < spike723> xen is really its own arch though? 23:25 < spike723> but its close to i386? 23:25 < _MarkW> spike723: it's a lot like x86, without the inconvenient bits. 23:25 < spike723> anyone in here running netbsd as domain0? 23:25 < _MarkW> it's currently a separate arch directory on netbsd and linux (although that will change) 23:26 < _MarkW> spike723: not here, we do most stuff under Linux 2.6 in Cambridge. 23:26 < spike723> I use "jail" with FreeBSD 23:26 < spike723> but I think xen would be better in some ways 23:26 < spike723> ok how about porting apps to run on xen? 23:27 < _MarkW> spike723: they just run 23:27 < _MarkW> userspace looks exactly like x86 23:27 < spike723> you mean binary compat? 23:27 < _MarkW> yup 23:27 < spike723> oh phat 23:27 < spike723> i've also been playing with qemu 23:28 < spike723> anyone benchmarked vmware vs xen vs xemu? 23:28 < _MarkW> can't benchmark recent vmwares 23:28 < _MarkW> the license forbids it 23:28 < spike723> lame 23:28 < _MarkW> basically: for user space compute intensive, vmware and xen will be the same as native 23:29 < spike723> ah huh 23:29 < knewt> bah. labour has hit the 324 mark 23:29 < spike723> how are the elections going? 23:29 < _MarkW> but when you start doing lots of memory management, IO and other kernel stuff, Vmware will pay relatively a painful penalty 23:29 < spike723> I'm in USA and I'm pretty fed up with the government 23:29 < spike723> it's all corrupt 23:29 < spike723> democrats and republicans 23:29 < _MarkW> blair just won 23:29 < _MarkW> bummer 23:30 < knewt> spike723: labour has just got it's majority. it's not going to win by as much as it would have wanted though 23:30 < _MarkW> but his majority should be rather smaller - which is a good thing. 23:30 < cartel> _MarkW: ban on benchmarks only applies to !workstation 23:30 < _MarkW> cartel: I thought that recent workstations were also non-benchmarkable? 23:30 < cartel> really? 23:30 < spike723> I must admit I don't really like linux 23:31 < cartel> whatever it is, its a totally lame stanza 23:31 < _MarkW> anyhow, on the Xen homepage there are benchmarks against whatever version of VMWare is allowed 23:31 < cartel> spike723: thats ok, it doesnt really like you either 23:31 < spike723> heh 23:31 < spike723> I like NetBSD though 23:31 -!- DEac- [~deac@xdsl-84-44-215-218.netcologne.de] has quit [Read error: Operation timed out] 23:31 < _MarkW> spike723: linux always lets you know when it doesn't like you :-) 23:31 < cartel> i have a netbsd print server 23:31 < cartel> on a 486 mainboard in a shoebox in a wall 23:31 < _MarkW> cartel: In a wall?? 23:31 < cartel> 120mb hard drive 23:31 < cartel> yep 23:31 < cartel> 8mb ram 23:32 < cartel> running lpr 23:32 < _MarkW> right 23:33 < cartel> :) 23:33 < _MarkW> it's time for bed now. knewt should go to bed too, it's way past his bedtime :-) 23:33 < _MarkW> see yall later. maybe tomorrow. 23:33 -!- _MarkW [~MarkWilli@hebble.cl.cam.ac.uk] has quit [Quit: Connection interrupted by tinfoil hat] 23:46 -!- DEac- [~deac@xdsl-195-14-201-243.netcologne.de] has joined #xen --- Log closed Thu May 05 23:59:00 2005