TheGrebs.com:IRC Logs:#xen:2005:05:10

---	Log	opened Tue May 10 00:00:58 2005
01:22	---	<<-- lilo [~lilo@lilo.usercloak.oftc.net] has quit (Quit: bbiab)
01:29	---	---> andy [~andy@andy.netop.oftc.net] has joined #xen
01:34	---	---> lilo [~lilo@lilo.usercloak.oftc.net] has joined #xen
01:35	andy	does xen support freebsd for domain 0, yet?
01:40	---	---> Squink [~purple@S010600131077a67e.vc.shawcable.net] has joined #xen
01:52	---	<<-- rusty [~rusty@CPE-139-168-158-101.nsw.bigpond.net.au] has quit (Quit: Client exiting)
02:45	---	---> soffi [~soffi@proxy.du.vdsl.is] has joined #xen
02:46	mael	lo
02:57	---	<<-- soffi [~soffi@proxy.du.vdsl.is] has quit (Quit: This computer has gone to sleep)
03:12	---	User: *** heb_away is now known as heb
03:19	heb	hi
03:43	---	---> athomas [~athomas@ppp-0-19.lond-a-1.access.uk.tiscali.com] has joined #xen
03:50	---	---> drbyte [~byte@c210-49-121-44.eburwd3.vic.optusnet.com.au] has joined #xen
03:53	---	---> soffi [~soffi@proxy.du.vdsl.is] has joined #xen
04:05	---	<<-- soffi [~soffi@proxy.du.vdsl.is] has quit (Quit: This computer has gone to sleep)
04:27	---	---> soffi [~soffi@Tolvudeild-SC-gw.vv.simnet.is] has joined #xen
05:27	---	---> Beaky [~chatzilla@ncg-adsl.demon.co.uk] has joined #xen
06:27	---	<<-- soffi [~soffi@Tolvudeild-SC-gw.vv.simnet.is] has quit (Quit: Leaving)
06:40	---	---> Sting [~sting@mvil2.bb.netvision.net.il] has joined #xen
06:41	---	User: *** riel is now known as surriel
06:50	Sting	Newbie question:
06:50	Sting	I work with xen-2.0-testing source.
06:50	Sting	I tried to do a slight change under a file
06:50	Sting	under linux-2.6.11-xen0; now in order to build only the kernel of linux-2.6.11-xen0
06:50	Sting	what should I do ? I of course don't want to do the whole process of "make world".
06:50	Sting	should I do make arch=xen from under linux-2.6.11-xen0 ? thanks
06:53	surriel	make ARCH=xen vmlinuz modules
06:53	Sting	thanks
06:57	knewt	surriel: a simple make ARCH=xen should do it nowadays shouldn't it
06:58	knewt	it certainly made both the vmlinuz and all the .ko files for me when i did that
07:03	---	<<-- Sting [~sting@mvil2.bb.netvision.net.il] has quit (Quit: Client Exiting)
07:04	---	---> sunny [sunny@opencurve.org] has joined #xen
07:10	surriel	knewt: cool
07:11	knewt	actually, since i had exported ARCH=xen into that shells environment to make life easier, it was just "make" :)
08:29	---	---> franR [~franr@inextensia.net] has joined #xen
08:30	franR	heya all
08:39	---	User: *** unriel is now known as riel
08:46	---	---> Sting [~sting@mvil2.bb.netvision.net.il] has joined #xen
08:47	---	<<-- Sting [~sting@mvil2.bb.netvision.net.il] has quit (Quit: )
08:47	---	---> Sting [~sting@mvil2.bb.netvision.net.il] has joined #xen
08:51	---	<<-- drbyte [~byte@c210-49-121-44.eburwd3.vic.optusnet.com.au] has quit (Ping timeout: 480 seconds)
08:55	---	<<-- katzj [~katzj@c-66-30-206-55.hsd1.ma.comcast.net] has quit (Ping timeout: 480 seconds)
09:06	---	---> drbyte [~byte@c210-49-121-44.eburwd3.vic.optusnet.com.au] has joined #xen
09:06	---	---> rharper [~rharper@pixpat.austin.ibm.com] has joined #xen
09:07	---	---> hollis [~hollis@user-0vvde2g.cable.mindspring.com] has joined #xen
09:20	---	<<-- homebaum [~michael@wbar1.sea1-4-5-031-104.sea1.dsl-verizon.net] has quit (Quit: Client exiting)
09:30	---	---> sdog [~sdog@62.58.98.210] has joined #xen
09:33	sdog	hmm.. does anyone know how to enable multiple xen-br(0..X) by default
09:33	sdog	at boottime ?
09:36	sdog	anyone here ?
09:36	mael	pong
09:36	knewt	xen-brN is/are just bridge(s), so either modify the xend network script to set more of them up, or just use your distro's method of setting up bridges at boot
09:37	sdog	knewt: no problem there .. but I was wondering wether there was some kind of default mechanism that I'm failing to find ...
09:37	sdog	doing things " the standard way " makes it easier for upgrades/other collegues ..
09:38	knewt	myself i just disable the xend network script and configure my bridge using /etc/network/interfaces (debian/ubuntu system)
09:39	---	---> katzj [~katzj@wlanconf-nat-pool-bos.redhat.com] has joined #xen
09:40	sdog	somehow somewher when you start xend it calls the network script with some parameters.. where does it get those ? :)
09:46	knewt	vif-bridge and vif-antispoof in /etc/xen/xend-config.sxp
09:47	knewt	they get passed as bridge='%s' and antispoof=yes or antispoof=no
09:49	sdog	knewt: yep but they only take 1 valuy :(
09:50	knewt	yep
09:51	Sting	Hello does any know a URL (if there is one) where logs of this irc channel are kept?
09:53		* knewt points to the /whois of VS_ChanLog
09:59	Sting	I am a newbie in this: should I login into http://www.linode.com/irc/
10:02	---	<<-- hollis [~hollis@user-0vvde2g.cable.mindspring.com] has quit (Quit: leaving)
10:05	heb	Sting: use web browser, follow link to #xen channel logs, click on log :-)
10:06	heb	Sting: i.e. cut and paste http://www.linode.com/irc/ into your browser bar.
10:11	Sting	thnks!
10:18	---	<<-- Sting [~sting@mvil2.bb.netvision.net.il] has quit (Quit: Client Exiting)
10:41	---	<<-- athomas [~athomas@ppp-0-19.lond-a-1.access.uk.tiscali.com] has quit (Read error: Connection reset by peer)
10:42	---	---> hollis [~hollis@pixpat.austin.ibm.com] has joined #xen
10:49	---	---> matta-lt [~matta@69.93.28.254] has joined #xen
10:51	---	---> MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen
10:51	sdog	right .. fixed it with my own scripts .. ugly but whatever :)
10:51	---	<<-- sdog [~sdog@62.58.98.210] has quit (Quit: Oh shit, cops at my door!1!!@!!1!1!@!@@1!!)
10:59	---	<<-- Beaky [~chatzilla@ncg-adsl.demon.co.uk] has quit (Remote host closed the connection)
11:06	---	---> athomas [~athomas@ppp-0-65.lond-b-4.access.uk.tiscali.com] has joined #xen
11:29	---	Netsplit oxygen.oftc.net <-> jupiter.oftc.net quits: DEac-, grifferz, rharper, knewt, Robot101, franR, yosh, Squink, athomas, paavon, (+23 more, use /NETSPLIT to show all of them)
11:29	---	Netsplit over, joins: caker, HoraPe, demon, riel, yosh
11:29	---	Netsplit over, joins: Hunger, Shaun, grifferz, Robot101, tab, wenchien, murble
11:29	---	Netsplit over, joins: lilo, drbyte, surriel, athomas, MarkW, hollis, katzj, rharper, franR, Squink (+2 more)
11:29	---	Netsplit over, joins: cartel, movement, knewt, matta, @Sir_Ahzz, eigood, paavon
11:30	---	<<-- franR [~franr@inextensia.net] has quit (Quit: Error inopin�e -10s, -9, -8... Unexpected error -10s, -9, -8,... Really ?)
11:30	---	Netsplit over, joins: DEac-, spike723
11:46	knewt	yaaay. sys-con has dumped mog
11:47	MarkW	knewt: nurrrgh?
11:48	knewt	mog == maureen o'gara
11:48	hollis	that must be British slang for "I don't understand" ;)
11:48	demon	hot
11:48	katzj	about time
11:49	MarkW	hollis: correctly translated :-)
11:49	MarkW	knewt: I see, Good!
11:50	knewt	and pj has had to ask people to stop commenting on the Intimidation article, as at 1188 comments geeklog is having issues ;)
11:51	hollis	I don't see why anybody cares. people write crazy stuff on the internet all the time
11:51	MarkW	knewt: wow!
11:52	knewt	it gathered comments at something like a post a minute from the moment it was posted
11:53	---	<<-- athomas [~athomas@ppp-0-65.lond-b-4.access.uk.tiscali.com] has quit (Quit: Leaving)
12:07	---	---> cfreak [cfreak@dsl-084-056-105-110.arcor-ip.net] has joined #xen
12:08	---	---> stekloff [~stekloff@bi01p1.co.us.ibm.com] has joined #xen
12:30	---	<<-- DEac- [~deac@xdsl-195-14-200-54.netcologne.de] has quit (Ping timeout: 480 seconds)
12:42	---	---> DEac- [~deac@xdsl-81-173-163-48.netcologne.de] has joined #xen
12:51	---	---> tierra [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has joined #xen
13:09	---	---> ap [~andre@adsl-63-201-59-77.dsl.snfc21.pacbell.net] has joined #xen
13:30	---	<--- ap [~andre@adsl-63-201-59-77.dsl.snfc21.pacbell.net] has left #xen ()
13:53	---	<<-- matta-lt [~matta@69.93.28.254] has quit (Quit: There are 3 kinds of people: those who can count & those who can't.)
14:30	MarkW	hmmm, weird
14:30	MarkW	System RAM doesn't show up on my Xen test box under /proc/iomem. I've got a feeling that's a bug!
14:31	MarkW	Or maybe it's a "feature". Hrrmmmmm.
14:31	---	---> aap [~andre@adsl-63-201-59-77.dsl.snfc21.pacbell.net] has joined #xen
14:33	hollis	hmm? RAM is not IO memory
14:33	MarkW	hollis: no but under Vanilla Linux, /proc/iomem lists all kinds of memory
14:34	---	---> jimix [~jimix@yktgi01e0-s4.watson.ibm.com] has joined #xen
14:34	hollis	so it does. how odd.
14:34	MarkW	ROMs, RAMs, as well as bona fide io memory
14:34	MarkW	unfortunately the Kexec tools parse that file, so I'd quite like it to look what they're expecting
14:35	MarkW	I will investigate...
14:36	---	<<-- DEac- [~deac@xdsl-81-173-163-48.netcologne.de] has quit (Quit: Verlassend)
14:37	---	---> DEac- [~deac@xdsl-81-173-163-48.netcologne.de] has joined #xen
15:11	---	---> jeroney [~jeroney@pixpat.austin.ibm.com] has joined #xen
15:23	---	<<-- cfreak [cfreak@dsl-084-056-105-110.arcor-ip.net] has quit (Quit: .)
15:27	---	<--- aap [~andre@adsl-63-201-59-77.dsl.snfc21.pacbell.net] has left #xen ()
15:29	---	<<-- jimix [~jimix@yktgi01e0-s4.watson.ibm.com] has quit (Quit: Leaving)
16:00	---	<--- tierra [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has left #xen (bug, n: A son of a glitch.)
16:14	Squink	If anyone can shine a light on this situation:
16:14	Squink	I'm getting this on bootup;
16:14	Squink	kernel /boot/xen.gz dom0_me200704 console=vga
16:14	Squink	Error 15: File not found..
16:14	Squink	(which then leads to grub)
16:14	Squink	however
16:14	Squink	the file IS there
16:15	Squink	and the grub conf is
16:15	Squink	title Xen 2.0 / XenLinux 2.6.9
16:15	Squink	kernel /boot/xen.gz dom0_mem=200704 console=vga
16:15	Squink	module /boot/vmlinuz-2.6.9-xen0 root=LABEL=/ ro console=tty0
16:16	caker	Squink: remove /boot from those lines
16:16		* Squink <3 KVM hookup
16:18	Squink	hmm
16:18	Squink	http://523.org.uk/~squink/images/snapshot.png
16:18	Squink	^^ that's what it's doing now
16:19	MarkW	Squink: please stick your whole grub.conf online so we can have a peek
16:19	Squink	alright, logging back in
16:21	Squink	http://pastebin.com/282364 <- grub.conf
16:21	---	---> mael_ [~Mael@kali.tirnamban.org] has joined #xen
16:21	mael_	hi
16:21	mael_	MarkW: are you there?
16:21	MarkW	mael_: yup
16:22	mael_	do you remember the qemu/kqemu problem I told you about yesterday? I tested without kqemu and it worked
16:23	mael_	I posted on qemu-devel and will provide info to Fabrice Bellard about it
16:23	knewt	does kqemu itself run /under/ xen btw?
16:23	MarkW	mael_: thanks for that. I saw the post, wondered if it was you
16:23	mael_	it was
16:23	MarkW	knewt: I doubt it
16:24	mael_	knewt: well Jim Provan of win4Lin seem to be willing to make that work
16:24	MarkW	knewt: it uses cli / sti directly, although emulating those works in the unstable tree
16:24	mael_	qemu/kqemu under domO
16:24	MarkW	knewt: porting qvm86 (the open source accelerator) should work, though
16:24	mael_	and he'd also like to make qemu/kqemu work under a domU
16:24	knewt	i had a look at the sourcecode of vmmon. nasty assembly language, staying well away
16:25	MarkW	mael_: now that would be really cool!
16:25	MarkW	win4lin pro runs under a domU anyhow
16:26	mael_	btw MarkW, Jim told me that his firm was "currently working with XenSource and [that he] can get answers for you"
16:26	MarkW	Moshe has a video on his website. It'd be much nicer to use the qemu accelerator.
16:26	knewt	Squink: include the root (hd0,0) line as part of the Xen section
16:27	mael_	well I think that dom0 or domU qemu/kqemu support will be needed for non-VT arch where you want to run a windows desktop
16:27	knewt	Squink: also, unless you use an initrd, the LABEL syntax won't work. probably best to just change that to the actual device node being used
16:27	mael_	corporate env for example
16:27	mael_	I spoke with James and Keir about that a bit when I went to Cambridge in April
16:27	knewt	i'd like to be able to run a windows vm on my current systems, certainly
16:27	Squink	knewt: how would I find out the device node?
16:28	MarkW	mael_: I agree
16:28	Squink	actually im sure i saw a .img file for xen in the boot dir
16:28	MarkW	mael_: I didn't know there had been contact with xensource bit if so then all the better!
16:29	knewt	Squink: if you want to try that, supply it as a 2nd module line, after the kernel
16:29	Squink	ah, nope, no .img file.
16:29	mael_	MarkW: from what I heard, having windows running on a VT system won't be a piece of cake
16:29	mael_	it seems this won't be possible until post Xen 3.X
16:29	MarkW	mael_: there's a lot of stuff to implement to make it work.
16:30	MarkW	Yes, I'd heard that, although I think I also saw it as a "possible" for 3.x
16:30	Squink	/dev/hda1 on /boot type ext3 (rw)
16:30	Squink	I assume I should set grub to /dev/hda1, then
16:30	mael_	MarkW: you're the guy in charge with hardware drivers domain stuff, right? (among other things)
16:31	mael_	what is needed to get AGP working on a domU?
16:31	knewt	Squink: no, that's your /boot, you want the dev node for your /
16:32	MarkW	mael_: I worked on the driver domains stuff when it was first developed but I'm on other stuff at the moment.
16:32	mael_	XenFS I guess
16:32	MarkW	mael_: Yeah, although this week I'm looking at kexec instead.
16:32	MarkW	(got sick of the Linux VFS)
16:33	knewt	Squink: try root=/dev/hda3
16:33	mael_	to boot kernels from domU? (for virtual dedicated server environment)
16:33	mael_	or is is something else?
16:34		* Squink reboots again
16:34	MarkW	mael_: yes, primarily for untrusted domain boot loaders
16:34	Squink	knewt: http://523.org.uk/~squink/images/snapshot.png <-- still does this
16:35	MarkW	but also potentially if people want to run their own domU-internal crashdump
16:35	MarkW	At some stage I think it'd be worth having kexec functionality in Xen itself but I'm not going to do that initially.
16:35	mael_	MarkW: will this feature be needed if you want to run a windows system on a VT cpu also?
16:35	mael_	(to run the ntloader stuff)
16:36	MarkW	mael_: no, for VT guests the native bootloader will (eventually) work fine
16:36	knewt	Squink: you've still got root=LABEL=/ in there
16:36	mael_	hum ok
16:36	MarkW	mael_: but for paravirtualised guests, we need an untrusted path for the domain to say "now boot this kernel"
16:36	MarkW	mael_: so mounting the domain filesystem in dom0 isn't acceptable
16:36	mael_	I still have troubles figuring what will be needed for virtualized VT guests
16:36	knewt	Squink: am i guessing right that you also didn't add the "root (hd0,0)" line above the kernel line in the Xen section?
16:37	mael_	maybe only hardware emulation
16:37	MarkW	mael_: initially just hardware emulation
16:37	mael_	MarkW: ok I see the picture for the booting stuff
16:37	Squink	knewt: nah i did add that, but the screenshot was otherwise identical
16:37	MarkW	mael_: later on, some extra paravirtualised hardware
16:37	Squink	knewt: so i just used the previous one
16:37	mael_	MarkW: what for? to get better perfs or for concurrent access sake?
16:38	MarkW	Squink: are you sure that vmlinux is there (and not, for instance a slightly different version number)
16:38	MarkW	mael_: perfomance
16:38	---	<<-- hollis [~hollis@pixpat.austin.ibm.com] has quit (Quit: leaving)
16:38	MarkW	mael_: flexibility
16:38	mael_	and will you be able to do pure hardware emulation or provide direct access to the underlying hardware?
16:39	mael_	I mean if you have a modern machine with a high perf 3D graphic card (lets say its an AGP one as it seems to be more problematic than PCI E), you expect a VT guest Windows system to give you full perfs
16:40	Squink	MarkW: good call :)
16:40	Squink	MarkW: lemme check some stuff.
16:40	mael_	and not the perfs of a GD5446
16:41	MarkW	mael_: i doubt the emulated hardware will be very fast
16:41	mael_	I fear so
16:41	MarkW	we won't initially support 3D graphics hardware virtualisation at all
16:41	mael_	and can't you dedicate hardware to a VM?
16:42	MarkW	as an interim measure, a virtual FB has been discussed, which would give reasonably decent performance for 2D
16:42	MarkW	yes, dedicating graphics cards to a VM is probably easier
16:42	mael_	for instance the graphic card might not be needed for a server running "in the background" on a workstation
16:42	MarkW	mael_: VT allows a guest to "poke through" to selected hardware devices if you set appropriate state
16:43	mael_	so you could give full access to that device to a VM either virtualize or para-
16:43	mael_	hum I'm not sure to understand what "poke through" stand for there :)
16:43	MarkW	mael_: potentially. The code isn't there for driver-domain style device access under VT but it appears to be possible.
16:43	MarkW	:-)
16:44	MarkW	"poke through", as in, reach through the virtualisation layer to the real machine
16:44	mael_	ok
16:44	mael_	that's great news
16:44	mael_	without any overhead?
16:44	mael_	(or a near 0 one)
16:45	MarkW	I imagine the overhead would be pretty low but I haven't checked out the details.
16:45	mael_	ok that would be great
16:45	MarkW	it would be way cool :-)
16:45		* mael_ is still waiting for VT system to be available
16:45	mael_	that's why I'm fumbling around with qemu/kqemu
16:45	MarkW	sharing one screen with 3D graphics is much harder than giving a second graphics card to a guest
16:46	MarkW	there's even one or two people trying to dedicate keyboard, mouse, gfx card to a Linux driver domain right now
16:46	mael_	I tried to build a test system with a dom0 running windows on a qemu/kqemu running on fbcon using sdl
16:46	mael_	with no luck
16:47	mael_	I tried on a normal system also with no luck either
16:47	mael_	(normal mean non-Xen)
16:47	MarkW	hmmm. hopefully if XenSource / Win4Lin are looking at this, it'll happen at some stage.
16:47	mael_	seems I have fbcon access troubles
16:48	mael_	I couldn't go any further as my server's power source burnt
16:48	mael_	so now I was trying to get a xen system working on qemu :)
16:48	Squink	hmm.
16:48	mael_	hopefully I'll find a working combo somewhere
16:50	mael_	MarkW: is there more devs working for XenSource than you guys in Cambridge?
16:50	mael_	(not to mention ppl from IBM/RH/HP/Intel/etc)
16:52	---	---> yarihm [~yarihm@80-218-7-109.dclient.hispeed.ch] has joined #xen
16:59	---	<<-- katzj [~katzj@wlanconf-nat-pool-bos.redhat.com] has quit (Read error: Connection reset by peer)
17:00	demon	er... dom0 running windows?
17:01	mael_	eurk, did I said that?
17:01	---	---> hollis [~hollis@user-0vvde2g.cable.mindspring.com] has joined #xen
17:01	demon	<mael_:#xen> I tried to build a test system with a dom0 running windows on a
17:01	demon	+qemu/kqemu running on fbcon using sdl
17:02	mael_	hum
17:02	demon	oh, I see
17:02	mael_	yeah
17:02	demon	running windows under qemu
17:02	riel	kqemu, even
17:02	mael_	it's "sandboxed" in qemu
17:03	MarkW	mael_: in cambridge there will be few non-university developers who are paid primarily by XenSource
17:04	MarkW	mael_: currently there's only one (tab / vincent)
17:04	mael_	ok
17:04	MarkW	mael_: I don't know of XenSource developers anywhere else but they may exist also :-)
17:05		* mael_ has trouble to understand the nebulous world of Xen
17:05		* MarkW also has trouble understanding it
17:05	mael_	:)
17:05	MarkW	mael_: seriously, I try to keep on top of it but I keep getting surprised by developers I didn't know existed
17:05	mael_	Keir and James explained some stuff I had trouble to figure out, but still...
17:06	mael_	hehe
17:06	mael_	really, this project in moving fast
17:06	mael_	I don't know if someone really has a big picture of it
17:06	mael_	Ian maybe
17:07	MarkW	Yeah, Ian probably has the best idea what's going on.
17:07	mael_	someone should do a "Kernel Traffic" kind of stuff for Xen dev ML
17:08	mael_	it's becoming to hard to follow if you're not an expert
17:08	hollis	mael_: I think I just heard you volunteer :)
17:08	mael_	hum I wish I could
17:08	MarkW	It'd be good to have something like that.
17:09	mael_	but I'm not dev and I have only a very high level view of how a CPU from the real world really work
17:09	MarkW	We need to find a victim^W volunteer
17:10	mael_	I was graduated in Biology before I had a Master Degree in Computer Sciences...
17:10	mael_	There's still stuff missing
17:10	---	User: *** riel is now known as unriel
17:11	mael_	but if hollis or riel work on a Xen dev digest, I'll be happy to do the french translation :)
17:11	mael_	+ML
17:12	MarkW	:-)
17:12		* MarkW puts on his invisibility hat
17:13	MarkW	If only we had a volunteer...
17:13	MarkW	;-)
17:13	mael_	I think the perfect candidate would be someone from Xensource :)
17:15	hollis	yeah, they're probably not doing anything else anyways
17:15		* knewt surrounds himself with an SEP-field
17:16	mael_	I recon knewt would be the perfect candidate
17:17	mael_	this would of course mean that xensource hire you :)
17:20	mael_	time to go
17:20	MarkW	mael_: see ya
17:20	mael_	bed is calling in my timezone :)
17:21	MarkW	mael_: many many hours away in my timezone
17:21	knewt	feck. i just ran out of disk space. this means that the entire ghc build process from bootstrap -> stage1 -> stage2 will run again. aarrrggghhh.
17:21	MarkW	knewt: gak!
17:22	knewt	it was a good way through stage2 already as well :(
17:22	---	<<-- mael_ [~Mael@kali.tirnamban.org] has quit (Quit: letting mael eavesdropping while I'm sleeping)
17:52	Squink	hmm
17:52	Squink	I can get into it now, but I'm losing my swap and stuff
17:52	Squink	which im guessing it because of my lack of initrd
17:53	Squink	anyone know what im supposed to change in order to get swap etc back?
17:57	knewt	i'd expect swap to configured in fstab, so i'm surprised that's not working
17:59	Squink	well it is fedora...
18:00	---	<<-- stekloff [~stekloff@bi01p1.co.us.ibm.com] has quit (Quit: Leaving)
18:01	---	<<-- Arnald [~Arnald@81-86-116-102.dsl.pipex.com] has quit (Remote host closed the connection)
18:06	---	<<-- rharper [~rharper@pixpat.austin.ibm.com] has quit (Quit: Leaving)
18:10	Squink	ahh fixed it.
18:10	Squink	made a new initrd file.
18:15	---	<<-- Robot101 [robot101@light.bluelinux.co.uk] has quit (Ping timeout: 480 seconds)
18:23	---	---> Robot101 [robot101@light.bluelinux.co.uk] has joined #xen
18:34	---	<<-- yarihm [~yarihm@80-218-7-109.dclient.hispeed.ch] has quit (Quit: Leaving)
18:48	Squink	hmm
18:48	---	<<-- lilo [~lilo@lilo.usercloak.oftc.net] has quit (Quit: bbiab)
18:48	Squink	I'm getting this when I try to boot a xend domain:
18:48	Squink	Root-NFS: No NFS server available, giving up.
18:48	Squink	VFS: Unable to mount root fs via NFS, trying floppy.
18:48	---	---> rusty [~rusty@CPE-139-168-158-101.nsw.bigpond.net.au] has joined #xen
18:48	Squink	and then it just sits, obviously.
18:48	Squink	So what am I missing?
18:54	---	---> lilo [~lilo@lilo.usercloak.oftc.net] has joined #xen
18:59	---	---> dwh [~sc@bgp01386991bgs.brodwy01.nm.comcast.net] has joined #xen
19:01	---	<<-- hollis [~hollis@user-0vvde2g.cable.mindspring.com] has quit (Quit: leaving)
19:02	dwh	Is Daniel Stekloff in here, by any chance?
19:03	dwh	I'm a summer intern... just wanted to take a moment to say 'hello' before going back to work on school-related projects.
19:03	dwh	er, summer intern for IBM in Beaverton, that is.
19:04	eigood	but it's not summer
19:05	dwh	Meh, any time that does not involve waking up in the morning to go to classes I usually dislike is close enough to summer for me :)
19:05	dwh	Even if it's winter break.
19:16	MarkW	dwh: I remember waking up in the morning once
19:16	MarkW	I didn't like it!
19:16	MarkW	Squink: it's either that the domain can't find its root device or that it doesn't support the filesystem on it, can't remember which.=
19:16	MarkW	Where are you putting the rootfs and what filesystem is it?
19:17	Squink	MarkW: fixed it btw ;) My issue now is that it's saying the IP im trying to use is already in use!
19:17	MarkW	Squink: ah, is it?
19:17	Squink	it shouldnt be..
19:17	MarkW	Squink: remember it's probably a bridged configuration
19:17	Squink	It's the same one i was using for UML
19:18	MarkW	so addresses won't be private to the machine unless you configure it like that
19:18	Squink	well it's supposed to be attaching to 66.36.255.63
19:22	---	User: *** andy is now known as troll
19:22	---	User: *** troll is now known as andy
19:30	---	<<-- dwh [~sc@bgp01386991bgs.brodwy01.nm.comcast.net] has quit (Quit: Leaving)
19:34	MarkW	Squink: you sure it's not used by any other systems at your hosting provider?
19:34	MarkW	out of interest, can you ping that IP? does there seem to be anything there?
19:42	Squink	Well. I've just done ifconfig eth0:2 del 66.36.255.63
19:43	Squink	of course, now the new problem is it's freaking out about something and not syncing, some im just rebooting the host
19:43	Squink	I'm trying to figure out how to make the server internet-accessable, because as yet, i've not been able to connect to it.
19:44	MarkW	not even dom0?
19:44	Squink	dom0?
19:44	MarkW	that's the virtual machine you first boot into, the one which controls the screen
19:44	MarkW	you can get into that, right?
19:44	Squink	you mean when you reboot, then it goes through xen, then boots into the original kernel
19:44	Squink	?
19:45	MarkW	when you boot on a Xen enabled system, _everything_ is a virtual machine ("domain")
19:45	Squink	right
19:45	Squink	I'm still trying to get around the differences between xen and uml ;)
19:45	MarkW	the Linux that comes up first is called "domain 0"
19:46	MarkW	its virtual machine is privileged, so it gets to grab the screen, PCI devices, etc. but fundamentally it's no different to the other virtual machines you run.
19:46	Squink	clever :)
19:47	MarkW	(side effect being that it's also possible to give other virtual machines control over individual real devices)
19:48	Squink	MarkW: http://pastebin.com/282440 <- here's what it's doing/.
19:48	MarkW	so, that being the case: can you get to dom0 over the network?
19:48	MarkW	looking...
19:49	Squink	yeah, dom0 is working perfectly
19:49	Squink	which is nice :)
19:49	MarkW	OK, it's not finding the root filesystem, so that suggests something wronge with your domain config
19:50	MarkW	or more accurately, it's finding a filesystem, just not finding init in there...
19:50	MarkW	Other than that it looks fine. You can press ^] to exit a domain console by the way
19:51	Squink	Config: http://pastebin.com/282443
19:51	Squink	that's useful to know
19:52	MarkW	the console behaves like a serial line, so you can always reconnect to it later
19:52	MarkW	well the config looks ok
19:53	Squink	Right. What should I look for in the .disk?
19:53	MarkW	/sbin/init
19:53	Squink	oh
19:53	Squink	looks like the filesystem has been corrupted.
19:54		* Squink copies a new .disk over
19:54	Squink	2 secs
20:00	Squink	alright
20:00	Squink	I guess it should work this time.
20:00		* MarkW crosses his fingers
20:01	Squink	fsck.ext3Possibly non-existent or swap device?
20:01	Squink	: No such device or address while trying to open /dev/ubd0
20:01	Squink	[FAILED]
20:01	Squink	*** An error occurred during the file system check.
20:01	Squink	...
20:01	MarkW	better
20:01	MarkW	ish
20:01	MarkW	the /dev/ubd0 error is OK and probably a good sign.
20:02	MarkW	you'll need to change references to UML block devices to match with some hda or sda nodes
20:02	MarkW	not sure what's giving you the filesystem error tho...
20:03	Squink	right. Where should I change them?
20:03	Squink	heh
20:04	Squink	cause I've no idea where they might be :)
20:04	Squink	oh wait
20:04	Squink	there it is.
20:06	MarkW	they'd be in the /etc/fstab
20:06	MarkW	or at least that's all you'd need to change to boot properly
20:06	Squink	right
20:06	Squink	it's booting, hanging on atd init, which im guessing is due to lack of net access
20:07	Squink	still erroring on the IP,which is a bit odd since i just free'd that up
20:07		* Squink tries again
20:08	Squink	Bringing up interface eth0: [ OK ]
20:08	Squink	yay.
20:08	MarkW	woohoo!
20:08	Squink	sorry not, atd, sendmail even.
20:09	Squink	INIT: no more processes left in this runlevel
20:09	Squink	woo.
20:09	MarkW	nice
20:09	Squink	now... how do i log in? heh
20:10		* Squink scratches head
20:11	MarkW	you should have a terminal...
20:11	MarkW	unless inittab is set up for some kind of weirdass UML terminal!
20:11	MarkW	can you get in the filesystem and check /etc/inittab, see what it says?
20:11	MarkW	(shutdown the domain first or you'll trash the filesystem, though)
20:13	Squink	inittab > http://pastebin.com/282453
20:14	MarkW	the first mingetty line looks weird to me
20:14	MarkW	can you change it to say tty1 instead of tty0 and see if that works?
20:17	Squink	eep
20:18	surriel	yeah, tty0 is wrong
20:18	Squink	http://pastebin.com/282456.
20:18	Squink	http://pastebin.com/282456
20:18	Squink	even
20:19	MarkW	errrr that's a scary boot output...
20:19	Squink	surely # root = "/dev/sda1 ro" # should be w?
20:19	MarkW	that's what most distros do, it should be fine
20:19	surriel	nope, the initscripts should remount it rw after fsck
20:20	Squink	ah
20:20	Squink	well
20:20	Squink	something went wrong.
20:20	Squink	I guess I'll try just letting it go.
20:20	MarkW	you weren't getting that before though, were you?
20:20	Squink	nope
20:20	Squink	I did forget to umount, though.
20:20	MarkW	Squink: that'll be it
20:21	MarkW	Linux really doesn't like shared mountable writes and it's quite easy to trash data that way
20:21	Squink	http://pastebin.com/282457
20:21	MarkW	The tools check for things like this but not when you're using file-backed VBDs :-(
20:22	Squink	looks like i trashed it again
20:22	MarkW	It seems likely there actually isn't an inittab anymore
20:22	MarkW	Yeah, I'd start with a clean disk image.
20:22		* Squink copies a new disk image across
20:23	Squink	alright so; 1. edit fstab, 2. change tty1 to tty0 ?
20:23	surriel	yeah
20:24	MarkW	change tty0 to tty1 :-p
20:25	Squink	right :)
20:27	Squink	Fedora Core release 1 (Yarrow)
20:27	Squink	Kernel 2.6.10-xenU on an i686
20:27	Squink	irreality login:
20:27	Squink	\o/
20:27	MarkW	yay!
20:27	Squink	alright
20:27	Squink	now we need to get it hooked up to the internet.
20:28	Squink	also
20:28	Squink	I got a lot of this:
20:28	Squink	Finding module dependencies: WARNING: Couldn't open directory /lib/modules/2.6.10-xenU: No such file or directory
20:28	Squink	FATAL: Could not open /lib/modules/2.6.10-xenU/modules.dep.temp for writing: No such file or directory
20:28	Squink	[FAILED]
20:28	Squink	is that ignorable?
20:28	MarkW	yep
20:28	Squink	good.
20:28	MarkW	if you install the modules in the domU it'll go away but you probably don't need them anyway
20:29	Squink	indeed.
20:31	Squink	Alright, here's what I was using for networking with UML; http://pastebin.com/282462
20:31	Squink	that'd be on dom0
20:31	Squink	but i noticed Xen has already set up some bridging and stuff... soo.
20:34	MarkW	do you want to bridge guests directly onto the physical network?
20:34	Squink	yep
20:34	MarkW	ok, throw that script away ;-) and set the IP address in the domain config file
20:35	Squink	alright.
20:36	Squink	uh. what's the var for IP?
20:36	MarkW	ummmm. isn't there one in the example config?
20:36	Squink	nope
20:37	MarkW	it's probably ip="123.123.123.123", plus gateway="123.123.123.123" plus netmask="123.123.123.123"
20:37	MarkW	check the manual, that lists all config file options
20:38	Squink	heh
20:38	Squink	it's not in the manual either =)
20:38	Squink	http://www.cl.cam.ac.uk/Research/SRG/netos/xen/readmes/user/user.html#SECTION03210000000000000000
20:38	Squink	the ip is already set in the .disk, though, if that helps
20:39	MarkW	oh, weird
20:40	MarkW	yeah, you can ifup from inside the domain if you want
20:40	MarkW	try that. make sure the device in the domU is "eth0" and not some UML device, tho
20:41	Squink	domU being the guest?
20:41	MarkW	yup
20:41	Squink	yep thats eth0
20:42	MarkW	does it work?
20:43	Squink	nope.
20:43	Squink	From 66.36.255.63 icmp_seq=0 Destination Host Unreachable
20:44	MarkW	what were you pinging to?
20:44	Squink	205.209.129.150
20:44	MarkW	have you configured a gateway in the domain?
20:44	Squink	in the config
20:45	Squink	but, I'm gonna do this anyway:
20:45	Squink	ifconfig eth0 66.36.255.63 broadcast 66.36.253.255 netmask 255.255.0.0 up
20:45	Squink	route add default gw 66.36.230.1
20:45	Squink	still doesnt work :)
20:46	MarkW	can you ping the gateway now?
20:47	Squink	nope
20:47	MarkW	errrr, IP = 66.36.255.63 and gateway = 66.36.230.1
20:47	MarkW	are you sure?
20:47	MarkW	they're not on the same subnet...
20:49	Squink	yep.
20:50	Squink	which is why the netmask is 255.255.0.0.
20:50	Squink	caker and I ran into this problem when I was trying to get uml to run originally
20:51	MarkW	oh, sorry I didn't read that properly ;-)
20:51	MarkW	Well first thing to do is to figure out why it can't ping your gateway... What is the gateway? Another machine? Or dom0 itself?
20:53	Squink	some other machine
20:53	Squink	dom0 is 66.36.253.97 / 66.36.231.201
20:53	---	---> katzj [~katzj@c-66-30-206-55.hsd1.ma.comcast.net] has joined #xen
20:54	MarkW	can you ping dom0?
20:54	MarkW	can dom0 ping the domU?
20:54	Squink	not with the netmask as it is
20:55	Squink	See the thing i dont get it why that script worked with UML and this isnt..
20:58	MarkW	eh? what's the netmask in dom0? I thought it had an IP in the same range as the domU
20:58	MarkW	?
20:58	MarkW	you could try and arping things too
20:58	Squink	wtb
20:59	Squink	there's something new running on dom0
20:59	Squink	vif9.0 Link encap:Ethernet HWaddr FE:FF:FF:FF:FF:FF
20:59	Squink	UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
20:59	Squink	never seen /that/ before.
20:59	MarkW	that's the virtual device at the other end Xen's virtual crossover cable
21:00	MarkW	we don't use tun / tap. it's just a virtual device, with a fast point to point link to the domU's eth0
21:00	Squink	right
21:00	Squink	k, im gonna paste something in pastebin.
21:02	Squink	MarkW: http://pastebin.com/282471
21:03	MarkW	ok...
21:05	MarkW	nothing obviously wrong
21:05	MarkW	I still don't get why you can't ping from the domU to dom0's address 66.36.231.201, tho
21:05	MarkW	or vice versa
21:06	Squink	Ah. me either really. I remember caker explaining it, but I've sort of forgotton now. I'm 95% sure it's something to do with the netmask
21:06	MarkW	can you try it anyway? it looked like it should work...
21:07	Squink	--- 66.36.231.201 ping statistics ---
21:07	Squink	6 packets transmitted, 0 received, 100% packet loss, time 5010ms
21:07	Squink	--- 66.36.253.97 ping statistics ---
21:07	Squink	3 packets transmitted, 0 received, 100% packet loss, time 2011ms
21:08	Squink	and from dom0;
21:08	Squink	--- 66.36.255.63 ping statistics ---
21:08	Squink	3 packets transmitted, 0 received, 100% packet loss, time 2013ms
21:08	Squink	=/
21:09	MarkW	oh dear :-(
21:11	Squink	MarkW: http://pastebin.com/282474
21:11	Squink	that's the original script
21:13	---	---> aliguori [~anthony@cpe-70-112-81-91.austin.res.rr.com] has joined #xen
21:13	MarkW	Squink: the end result looks kinda the same
21:14	Squink	yeah, I was just about to say that actually =/
21:14	MarkW	it's not obvious to me what's significantly different
21:14	MarkW	and I really don't understand your subnet thing
21:15	Squink	alright, what about arp?
21:15	Squink	cause that's in the script;
21:15	Squink	/sbin/ifconfig eth0 0.0.0.0 arp promisc up
21:16	MarkW	just enables use of arp on that interface
21:16	MarkW	i'd hawe thought that was the default anyhow
21:18	Squink	hum
21:18	Squink	there must be a way to see what's going on
21:19	MarkW	tcpdump is your friend
21:19	MarkW	or ethereal if you prefer
21:20	MarkW	run tcpdump on the vif device in dom0 and try to access the network from the guest
21:22	Squink	[root@irreality home]# ping 205.209.129.150
21:22	Squink	PING 205.209.129.150 (205.209.129.150) 56(84) bytes of data.
21:22	Squink	From 66.36.255.63 icmp_seq=0 Destination Host Unreachable
21:22	Squink	tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
21:22	Squink	listening on vif9.0, link-type EN10MB (Ethernet), capture size 96 bytes
21:22	Squink	19:22:11.362470 arp who-has c-vl101-gw.acc.dca2.hopone.net tell 66.36.255.63
21:22	Squink	19:22:12.362452 arp who-has c-vl101-gw.acc.dca2.hopone.net tell 66.36.255.63
21:22	MarkW	that's the router it's arping for, right?
21:24	Squink	looks like it
21:24	MarkW	try tcpdump-ing xen-br0 and see if the traffic makes it onto there
21:24	MarkW	then dump eth0 and see if it sees the traffic
21:25	MarkW	we'll get there in the end!
21:25	Squink	haha, we better otherwise my partner will kill me ;)
21:26	Squink	--- 205.209.129.150 ping statistics ---
21:26	Squink	2 packets transmitted, 0 received, 100% packet loss, time 1013ms
21:26	Squink	tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
21:26	Squink	listening on xen-br0, link-type EN10MB (Ethernet), capture size 96 bytes
21:26	Squink	19:26:24.942465 arp who-has c-vl101-gw.acc.dca2.hopone.net tell 66.36.255.63
21:26	Squink	19:26:25.942449 arp who-has c-vl101-gw.acc.dca2.hopone.net tell 66.36.255.63
21:27	Squink	hmm
21:28	Squink	well, this is beyond me.
21:29	MarkW	try dumping eth0, that should tell us whether the arps are going out on the wire
21:30	Squink	uh
21:30	Squink	that could be a problem
21:31	Squink	from 3 seconds of tcpdump;
21:31	Squink	1705 packets captured
21:31	Squink	1706 packets received by filter
21:32	MarkW	pull up the man page for tcpdump and take a look at the filtering commands
21:32	MarkW	you should be able to narrow it down to only packets involving the domU
21:32	MarkW	or just grep the output from tcpdump ;-)
21:34	Squink	root@quartz [~]# tcpdump -i eth0 \| grep 66.36.255.63
21:34	Squink	look about right to you?
21:34	MarkW	seems reasonable
21:35	MarkW	for sanity check, also try it against xen-br0 to make sure it does pick up the packets when they are there
21:35	Squink	in that case, no, nothing is getting to eth0 :)
21:35	Squink	k
21:35	Squink	ok
21:35	Squink	grep isnt working.
21:37	MarkW	ok, you'll have to debug grep first, or use tcpdump's filtering
21:40	Squink	root@quartz [~]# tcpdump -i xen-br0 host 66.36.255.63
21:40	Squink	tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
21:40	Squink	listening on xen-br0, link-type EN10MB (Ethernet), capture size 96 bytes
21:40	Squink	alright
21:40	Squink	it's sane, and nothing is getting to eth0
21:41	---	---> tierra [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has joined #xen
21:42	MarkW	but you are still seeing packets output from tcpdump for the bridge and the vif?
21:42	Squink	yep
21:43	MarkW	well then i suppose it's a matter of debugging what the linux bridging code thinks it's doing
21:45	---	---> tierra_ [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has joined #xen
21:46	Squink	hm
21:46	Squink	alright, so packets are going from vif9.0 >> xen-br0
21:47	Squink	and from there they're supposed to be going to eth0, right?
21:47	Squink	so we could assume that it's a bridging issue rather than a xen issue
21:47	MarkW	that's where they need to go and where I expect them to go
21:47	MarkW	but apparently the bridge code thinks otherwise
21:48	Squink	does xen have a bridge script somewhere?
21:48	MarkW	Yes, this may be a purely bridge-related issue
21:48	MarkW	Look it /etc/xen/scripts for network scripts
21:51	---	<<-- tierra_ [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has quit (Quit: bug, n: A son of a glitch.)
21:52	---	<<-- tierra [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has quit (Read error: Operation timed out)
21:53	Squink	well the scripts seem normal.
21:53	MarkW	heh, right now I can't even start a domU on my system - your streets ahead of me
21:54	aliguori	uh oh, is -unstable still hosed?
21:54	MarkW	(nb: I'm not quite as incompetent as I sound)
21:54	Squink	lol
21:54	MarkW	aliguori: hosed?
21:54	aliguori	MarkW: broken
21:54	MarkW	no, I mean hosed how?
21:55	MarkW	Is there some well-known breakage that I've missed...?
21:55	Squink	oh hey.
21:55	Squink	bridge name bridge id STP enabled interfaces
21:55	Squink	xen-br0 8000.feffffffffff no vif9.0
21:55	Squink	shouldnt that have eth0 in it?
21:55	MarkW	errr yep
21:55	aliguori	oh, yesterday's build was panic'ing before dom0 startup... someone else was also experiencing the same
21:55	MarkW	nope, not the problem I'm seeing
21:56	aliguori	i'm about to update to the latest BK and try to recreate (now that I have access to a serial port)
21:56	MarkW	it's domUs that are unhappy. it seems rather like my install is inconsistent somehow but I've no idea where
21:56	aliguori	MarkW: what makes the domUs unhappy? do they just need a friend :-)
21:56	Squink	brctl addif xen-br0 eth0
21:57	Squink	who thinks this will lock me out? :)
21:57	MarkW	Squink: i was just going to ask you to do that :-)
21:57	MarkW	It might do...
21:57	Squink	heh
21:57	Squink	nothing a reboot can't solve
21:57	MarkW	linux bridging is weird
21:57	MarkW	aliguori: I think Xen is bullying them
21:57		* Squink gets ready to contact the reboot guys
21:57	Squink	bridge name bridge id STP enabled interfaces
21:57	Squink	xen-br0 8000.000d619d364a no vif9.0
21:57	Squink	eth0
21:57	Squink	that's better.