--- Day changed --- Log opened Wed May 11 23:59:02 2005 00:01 -!- tierra [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has quit [Quit: bug, n: A son of a glitch.] 00:04 -!- linbot [~supybot@ns.theshore.net] has joined #xen 00:17 -!- soffi [~soffi@proxy.du.vdsl.is] has quit [Quit: This computer has gone to sleep] 02:06 -!- Zxcvb_ [asdf@12-215-141-137.client.mchsi.com] has joined #xen 02:07 < Zxcvb_> would there be any problem with using normal windows drivers with xen? 02:08 < Zxcvb_> specifically, if reactos is ported to xen would there be any issues with drivers for usb devices and such 02:28 -!- Squink [~purple@S010600131077a67e.vc.shawcable.net] has quit [Read error: Connection reset by peer] 02:49 -!- rusty [~rusty@bh02i525f01.au.ibm.com] has quit [Quit: Client exiting] 03:20 -!- franR [~franr@inextensia.net] has joined #xen 03:37 -!- cc [~cc2@c210-49-121-44.eburwd3.vic.optusnet.com.au] has quit [Quit: Leaving] 03:58 -!- andy [~andy@andy.netop.oftc.net] has quit [Remote host closed the connection] 03:59 -!- DEac- [~deac@xdsl-195-14-219-28.netcologne.de] has quit [Ping timeout: 480 seconds] 03:59 -!- andy [~andy@andy.netop.oftc.net] has joined #xen 04:13 -!- DEac- [~deac@xdsl-84-44-145-179.netcologne.de] has joined #xen 04:17 -!- athomas [~athomas@ppp-0-52.lond-a-1.access.uk.tiscali.com] has joined #xen 04:50 < heb> Zxcvb_: does reactos use unmodified windows drivers? 07:03 < Zxcvb_> it can 07:04 < Zxcvb_> more with each release 07:05 < Zxcvb_> but if a modified windows/reactos were running in xen, would installing a windows usb webcam (for example) driver cause problems? 07:34 < heb> I'm not familiar with the WIndows USB driver stack but assuming it's reasonably sane then there would be three options: 07:35 < heb> 1) map USB hardware and let reactos provide it to native windows host controller driver 07:36 < heb> 2) provide a paravirtualised host controller driver for reactos which connected to the existing windows usb stack. 07:37 < heb> 3) hook into the USB device emulation for unmodified operating systems (not sure if any exists yet). 07:38 < heb> In all cases, it ought to be possible for the device driver for a specific USB device to run unmodified since it should just hook into the USB driver stack as normal. 07:38 < Zxcvb_> yes, but drivers use ring 0 07:40 < Zxcvb_> as for the host controller, providing a generic ohci or uhci controller which windows/reactos uses sounds like it would work fine 07:40 < heb> Not sure I understand your point about ring 0. 07:40 < Zxcvb_> xen requires the OS to be modified right? 07:42 < heb> The paravirtualised guests are modified, there are plans for unmodified guests using the new CPU virtualisation features vanderpool and pacifica 07:42 < Zxcvb_> so presumably most drivers running under the guest would have to be modified as well 07:43 < heb> I was kind of assuming that you'd be able to modify the host controller aspects of the USB stack for reactos and then the native windows drivers wouldn't need modification because they were sufficiently high up the stack. 07:46 < Zxcvb_> That might work but a driver is a driver right? 07:46 < Zxcvb_> otherwise wine would be able to use many drivers 07:47 < heb> I think it depends on the API the driver is using. A USB device driver for a normal device probably doesn't need to talk directly to any hardware, it just connects into the stack which is a software API. If you provide the same API, it should work. 07:48 < Zxcvb_> I am refering to instructions, not hardware access 07:49 < Zxcvb_> for example, I doubt an unmodified safedisc/securrom driver would ever work no matter what you did with the OS 07:49 < heb> What specifically do you think would be a problem? 07:50 < Zxcvb_> some of the ring 0 instructions 07:51 < Zxcvb_> the only reason to modify the guest itself would be if certain cpu instructions could not be virtualized 07:52 < heb> I doubt a safedisc/securrom USB driver needs any ring 0 instructions. 07:52 < Zxcvb_> presumably xen is able to virtualize some ring 0 code or every driver would have to be reimplemented 07:54 < heb> Yes, most of the code that needs to be virtualised is not in the drivers, xen provides an alternative implementation and the drivers are compiled against that. 07:54 < Zxcvb_> is there a list of incompatible cpu instructions somewhere? 07:59 < heb> I think I'm a bit out of my depth here. Are you going for a paravirtualized implementation of reactos or an unmodified implementation? 08:03 < Zxcvb_> paravirtualized reactos with modified low level drivers (video/sound/cdrom/hard drive/usb stack/etc) but unmodified other drivers (safedisc/securrom/usb devices/printer devices through dlportio/etc) 08:08 < heb> So I would have thought that it would be possible to make that work. 08:10 < heb> BTW, when I said API above I should have said ABI. 08:11 < heb> The problem would be if there was any low-level stuff compiled into the drivers you wanted to reused. 08:12 < heb> For example from inline functions in header files which are no longer valid for the paravirtualised implementation. 08:13 < heb> With Linux, this is common but for windows I would expect it to be less common since they have to support binary drivers released by other companies and must therefore have a more stable ABI. 09:28 -!- katzj [~katzj@c-66-30-206-55.hsd1.ma.comcast.net] has quit [Ping timeout: 480 seconds] 09:48 -!- rh_out is now known as rharper 09:54 < Zxcvb_> heb: and I am talking about CPU instructions that won't work in xen 09:56 -!- hollis [~hollis@user-0vvde2g.cable.mindspring.com] has joined #xen 09:58 -!- Zxcvb_ [asdf@12-215-141-137.client.mchsi.com] has left #xen [] 09:59 -!- katzj [~katzj@wlanconf-nat-pool-bos.redhat.com] has joined #xen 10:11 -!- Tv [~Tv@hq.inoi.fi] has joined #xen 10:11 -!- Snake007uk [~Snake007u@194.159.31.243] has joined #xen 10:12 < Snake007uk> hello 10:12 < Snake007uk> silly question, how comes #xen isnt on freenode ? 10:13 < hollis> shrug, does it really matter? 10:13 < Snake007uk> nope just most opensoure oproject are there 10:13 < Snake007uk> anyway 10:13 < Snake007uk> i am here to ask about xen status, regarding fedroa 10:13 < Snake007uk> fedora 10:14 < Snake007uk> i manage to get it installed and i "think" i got a virtual system setup but when yuo boot into the system how do you detach from it and have it running in the background ? 10:15 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen 10:15 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has left #xen [] 10:15 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen 10:15 -!- mode/#xen [+o MarkW] by ChanServ 10:16 < riel> Snake007uk: ctrl+] 10:17 < Snake007uk> oh i see 10:17 < Snake007uk> hehe 10:17 < Snake007uk> ok 10:17 < Snake007uk> how long to a proper install process for vm is done ? 10:18 < riel> there's a dependency on device discovery (aka. xenbus) 10:19 < Snake007uk> lost me totally 10:19 < Snake007uk> riel lol i am talking to you in 2 channels :) 10:19 < riel> the installer needs to be able to detect what devices are present in the guest 10:19 < riel> new to irc ? ;) 10:20 < riel> in order to do that, we need a xen bus in sysfs 10:20 < Snake007uk> no not new, but i didnt realise 10:20 < Snake007uk> ah i see 10:20 < Snake007uk> sysfs is that new procfs ? 10:20 < riel> kind of ;) 10:20 < Snake007uk> do u mind explaining the difference, dont worry i aint that dumb so you can be straight forward 10:20 < Snake007uk> :) 10:21 < riel> sysfs is for device drivers 10:21 < riel> so userspace can find out what devices there are in the system 10:21 < riel> and what device drivers need to be loaded 10:21 < riel> an interface to configure device drivers 10:21 < riel> etc... 10:21 < Snake007uk> so does sysfs, connect to udev, etc... and procfs gets infro from udev ? 10:22 < MarkW> udev gets infor from sysfs 10:22 < riel> and procfs is completely separate 10:22 < Snake007uk> ok 10:22 < Snake007uk> i see 10:22 < MarkW> procfs contains information about processes, various kernel settings, etc 10:22 < Snake007uk> oh but 10:22 < Snake007uk> when you do do cat /proc/cpuinfo ? isnt it actually getting info from hardware ? 10:23 < riel> that's legacy ;) 10:23 < Snake007uk> oh so procfs is going also ? 10:23 < MarkW> no but sysfs is The New Way for storing hardware / device driver settings in a uniform way 10:24 < MarkW> avoiding the clutter that's in proc... 10:24 -!- matta [~matta@69.93.28.254] has quit [Ping timeout: 480 seconds] 10:24 < Snake007uk> oh ok, so sysfs will eventually replace proc 10:24 < MarkW> but procfs will stick around for compatibility reasons, and for stuff like info on running processes 10:24 < Snake007uk> ok 10:24 < Snake007uk> i see 10:25 < MarkW> (which was its original process before feature creep set in) 10:25 < Snake007uk> so before sysfs what did there used to be ? 10:25 < riel> how long before sysfs ? ;) 10:25 < Snake007uk> how long as sysfs been around ? 10:26 < MarkW> since 2.6 10:26 < Snake007uk> ah ok 10:26 < riel> in linux 1.2 there wasn't any good device discovery 10:26 < Snake007uk> ok bakc to the xen topic 10:26 < riel> in fact, modules were still new ;) 10:26 < Snake007uk> then 10:26 < Snake007uk> xenbus is at what stage 0.x 10:27 < riel> what do you mean with that question ? 10:27 -!- Tv [~Tv@hq.inoi.fi] has quit [Quit: Client exiting] 10:27 < MarkW> more like stage -1 10:27 < MarkW> there's a plan but no code yet (AFAIK) 10:28 < Snake007uk> oh ok, 10:28 < MarkW> it needs the registry stuff.... Rusty is working on a first cut of the registry 10:28 -!- Tv [~Tv@hq.inoi.fi] has joined #xen 10:28 < MarkW> XenBus will appear later 10:28 < Snake007uk> the reason i asked was, i was thinking tht xen 0.x would be out with fc4 10:28 < Snake007uk> and that it would be easy to install and setup VM 10:29 < MarkW> riel: Xen 3.0 (possibly a beta version) will be in FC4, right? 10:29 < Tv> p.x 10:29 < Tv> err 10:29 < riel> MarkW: yes 10:29 < riel> MarkW: but I guess xenbus won't be in time 10:29 < Snake007uk> ok xen 3.0, anyone got a link to feature it will have 10:29 < riel> MarkW: so the install will still be a bit hacky 10:29 < riel> Snake007uk: see wiki.xensource.com ;) 10:29 < MarkW> install into a directory? 10:29 < Snake007uk> i think xen is gona be the next big thing and its good to knwo something liek this early on 10:29 < riel> yeah, install using yum 10:30 < MarkW> Snake007uk: and see the roadmap on xen.sf.net 10:30 < MarkW> riel: OK. I've talked with soffi about writing a little "apt" style tool for installing VM images 10:30 < Snake007uk> wow 10:30 < Snake007uk> that would be cool 10:30 < riel> MarkW: how is that different from yum --installroot ? ;) 10:30 < Snake007uk> but would that be base images only 10:30 < MarkW> Won't be as good as a proper OS installer but it'll be distro independent 10:31 < riel> MarkW: or mach, for that matter 10:31 < MarkW> (grabbing the tarballs of his website) 10:31 < riel> ahhhh, I see 10:31 < riel> neat 10:31 < mael> hey! hello MarkW 10:31 < MarkW> riel:, Snake007uk: I had a little script that would download the tarballs and install them to phy, file or directory 10:32 < MarkW> but it died with my laptop... i imagine we'll put together something new 10:32 < MarkW> mael: hi 10:32 < mael> what happened? It seems an early time to see you here 10:32 < Snake007uk> riel so yum --installroot install a base fedora install then you can use group install to install other stuff like x and things ? 10:33 < MarkW> mael: I got up for pizza :-) 10:34 < riel> Snake007uk: please read the FedoraXenQuickstart page ;) 10:34 < MarkW> Snake007uk: btw, debootstrap / crosshurd are good for installing debian systems into VMs (even on a non-debian host) 10:34 < riel> Snake007uk: it's on the fedoraproject.org wiki 10:35 < MarkW> Snake007uk: and SuSE 9.3 bundles Xen 2.0 and has an "install into directory" wizard 10:35 < Snake007uk> yeah but i prefer doing by hand and they fiond out what each bit does :) 10:35 < Snake007uk> riel i did rush through that before 10:37 < MarkW> Snake007uk: ok. there are tarballs of various distros on soffi's website. it's linked from installation / getting started / something similar in the Xen wiki 10:37 < MarkW> you might find those nice to play with but you'll have to install manually for now 10:38 < Snake007uk> yeah i think i need to get the install process sussed out, at the moment i am very confused with it, i jsut followed the instructions and have no idea what just happend 10:44 < MarkW> Snake007uk: the Xen install process? 10:44 < Snake007uk> yep 10:45 < MarkW> what's the problem? 10:45 < Snake007uk> nothing just i rushed the install 10:45 < Snake007uk> i am gona read through what is being done 10:45 < Snake007uk> i know that 10:45 < Snake007uk> you make a file which is a file system for xen 10:46 < Snake007uk> this file is then mounted 10:46 < Snake007uk> and everything is installed inside there, but 10:46 < Snake007uk> what exactly is installed in there ? 10:46 < MarkW> You use disk files to install guest filesystem images. 10:46 < MarkW> Have you actually installed Xen itself yet? 10:47 < Snake007uk> yeah 10:47 < Snake007uk> i have done it on fc4 10:47 < MarkW> ah ok 10:48 < MarkW> so you need to use yum --installroot OR download one of soffi's tarballs and unpack it in the disk file 10:48 < riel> instead of a loopback file, you could also use a partition or an LVM volume 10:48 < MarkW> does booting into Xen / dom0 work OK? 10:48 < Snake007uk> i have done the yum --install roo 10:48 < Snake007uk> yes i am booting into it right now 10:48 < MarkW> OK, and the installroot worked ok? 10:49 < Snake007uk> yep 10:49 < Snake007uk> Linux skynet.eu.global.ad 2.6.11-1.1288_FC4xen0 #1 SMP Fri May 6 15:59:42 EDT 2005 i686 i686 i386 GNU/Linux 10:49 < Snake007uk> how do i check which dom i am in ? 10:49 < riel> MarkW: it's like debootstrap, except you need to create /etc/fstab /dev/null and /dev/console by hand inside the guest ;) 10:49 -!- hollis [~hollis@user-0vvde2g.cable.mindspring.com] has quit [Quit: leaving] 10:50 < MarkW> riel: OK, I had that impression but I've never used it myself. 10:50 < MarkW> Snake007uk: dom0 always owns the screen 10:50 < MarkW> Snake007uk: any other domains' terminals will appear in a window 10:50 < Snake007uk> xm list 10:50 < Snake007uk> Name Id Mem(MB) CPU State Time(s) Console 10:50 < Snake007uk> Domain-0 0 478 0 r---- 21.8 10:51 < Snake007uk> ok now i want to go into my vm 10:51 < Snake007uk> how do i do that 10:51 < MarkW> you need to create the device node riel mentioned in that domain's filesystem 10:51 < MarkW> otherwise it won't boot. 10:51 < Snake007uk> like this 10:51 < Snake007uk> do i have mount ht file first 10:52 < riel> yes, you need to mount it 10:52 -!- movement is now known as movewent 10:52 < MarkW> yes, you'll need to mount it again. the fedora wiki has the exact commands I think. 10:52 < riel> in the same way you did before you started the yum installroot 10:52 < riel> MarkW: yeah, it has all the commands ;) 10:52 < MarkW> I think the xen wiki's faq mentions it also. it's to avoid confusing udev ;-) 10:52 < riel> MarkW: I even went through them and corrected them before FC4 test3 came out 10:52 < MarkW> (well, strictly, it's to avoid confusing the kernel before udev has started...) 10:53 < MarkW> riel: cool! nice to see someone updates the wiki :-) 10:53 < riel> it's because normally early udev gets started from the initrd 10:53 < MarkW> riel: yup 10:53 < riel> but we don't do that yet for xenU 10:53 < Snake007uk> ok guy file mounted 10:53 < katzj> riel: "yet"... that should hopefully change soon :) and then we should probably just collapse down to kernel-xen 10:53 < Snake007uk> ok proc has everything inthere 10:54 < riel> katzj: yeah 10:54 < Snake007uk> what else should i be looking for 10:54 < Snake007uk> cat /mnt/etc/fstab 10:54 < Snake007uk> none /dev/pts devpts gid=5,mode=620 0 0 10:54 < Snake007uk> none /dev/shm tmpfs defaults 0 0 10:54 < Snake007uk> none /proc proc defaults 0 0 10:54 < Snake007uk> none /sys sysfs defaults 0 0 10:54 < MarkW> Snake007uk: you need to make a minimal set of device nodes 10:54 < riel> wait, no fstab entry for / ? 10:54 < MarkW> the command you need is on the faq, i can't remember the device numbers offhand 10:54 < MarkW> riel: good point! 10:55 < Snake007uk> i did that before 10:56 < Snake007uk> there is a entry for / 10:56 < Snake007uk> /dev/sda1 / ext3 defaults 1 1 10:56 < MarkW> ok, just the device nodes to do then 10:56 < Snake007uk> the device node are there i did that before when i first installed 10:56 < Snake007uk> i dont have to created again do i 10:56 < MarkW> ah, ok then 10:57 < Snake007uk> i sendt you out privatly MarkW 10:57 < MarkW> in that case you need to unmount the file 10:57 < MarkW> and write a config file 10:57 < Snake007uk> which config file ? 10:57 < MarkW> base it on one of the examples in /etc/xen/ 10:57 < MarkW> a config file for the domain 10:58 < Snake007uk> kernel ="/boot/vmlinuz-2.6.11-1.1286_FC4xenU" 10:58 < Snake007uk> memory = 64 10:58 < Snake007uk> name = "rawhide" 10:58 < Snake007uk> nics = 1 10:58 < Snake007uk> disk = ['file:/root/fedora.img,sda1,w'] 10:58 < Snake007uk> root = "/dev/sda1 ro" 10:58 < MarkW> good 10:58 < Snake007uk> i have updated 10:58 < MarkW> xm create -f /etc/xen/configfile -c 10:58 < Snake007uk> kernel ="/boot/vmlinuz-2.6.11-1.1288_FC4xenU" 10:58 < Snake007uk> new kernel 10:59 < Snake007uk> Using config file "/etc/xen/rawhide". 10:59 < Snake007uk> Error: Error creating domain: (12, 'Cannot allocate memory') 10:59 < Snake007uk> and what does that command do ? 10:59 < riel> ahhhhh 10:59 < riel> look at the last line of "xm info" output ;) 10:59 < Snake007uk> lol it says 5 10:59 < riel> then look at the FedoraXenQuickstart page, search for "balloon" 10:59 < Snake007uk> LOO 11:00 < MarkW> you can't do much with 5meg ;-) 11:00 < Snake007uk> iojk 11:00 < Snake007uk> its working 11:00 < MarkW> woo! 11:00 < Snake007uk> now its booking into tthe system 11:00 < Snake007uk> what does 11:00 < Snake007uk> xm create -f /etc/xen/configfile -c 11:01 < Snake007uk> do exactly 11:01 < Snake007uk> sldo 11:01 < Snake007uk> root = "/dev/sda1 ro" does that make the filesystem read only ? 11:01 < riel> it starts up the guest domain specified in /etc/xen/configfile, and attaches to the console 11:01 < MarkW> create a domain, using that config file, plus "-c" makes xm turn into the domain's console 11:01 < MarkW> rather than entering the console with a separate command 11:01 < riel> the initscripts will turn the mount read-write after fsck 11:01 < Snake007uk> ok selinux is activated 11:01 < MarkW> Snake007uk: no, that's just what linux distros do to start with 11:01 < Snake007uk> how do i turn it off 11:01 < riel> boot with enforcing=0 11:02 < riel> then edit /etc/selinux/config inside the guest 11:02 < Snake007uk> ok detached from the guest vm 11:02 < riel> make things permissive 11:02 < riel> or even disabled 11:02 < Snake007uk> how do i stop it 11:02 < Snake007uk> and then i can mount file again and edit it 11:02 < riel> xm destroy 11:02 < MarkW> xm shutdown or xm destroy 11:02 < MarkW> (xm destroy is like kill -9( 11:03 < Snake007uk> ok 11:04 < Snake007uk> xm create -f <--- whats teh -f do 11:04 < Snake007uk> so far so godo 11:04 < Snake007uk> booting into system 11:04 < Snake007uk> :) 11:04 < Snake007uk> ======= Memory map: ======== 11:04 < Snake007uk> frozen there 11:05 < Snake007uk> Enforcing mode requested but no policy loaded. Halting now. 11:05 < Snake007uk> *** glibc detected *** /sbin/init: free(): invalid pointer: 0x0804e534 *** 11:06 < MarkW> Snake007uk: -f == config file path 11:06 < MarkW> Snake007uk: errrm that's a bit weird. have you told it not to enforce? 11:07 < MarkW> (set enforce=0 in the extra variable in the config to add it to the cmdline) 11:07 < riel> alternatively, mount the filesystem from domain 0 11:07 < riel> and edit /mnt/etc/selinux/config ;) 11:07 < Snake007uk> i cahnged the selinux to =disabled 11:08 < riel> mmmm ok 11:08 < MarkW> but it's still trying to enforce stuff? 11:08 < MarkW> I'll defer to riel's expertise on this one... 11:08 < riel> is selinux disabled in domain 0 too ? 11:08 < Snake007uk> one sec 11:08 < Snake007uk> let me confirm 11:08 < Snake007uk> oh it didnt save :( 11:08 < Snake007uk> let me do this again 11:10 < knewt> MarkW: i have to admit i usually rely on the DWIMiness of xm create, and miss off the -f and the path to the config file :) 11:10 < riel> knewt: "xm create " 11:11 < riel> that's equivalent to "xm create -f /path/to/" 11:11 < Snake007uk> here we go ain 11:11 < MarkW> knewt: from python.superpowers import telepathy.* 11:11 < knewt> riel: exactly 11:11 < Snake007uk> Welcome to Fedora Core 11:11 < Snake007uk> Press 'I' to enter interactive startup. 11:11 < Snake007uk> Starting udev: [FAILED] 11:11 < Snake007uk> is udev failing normal 11:11 < riel> not normal 11:11 < Snake007uk> and i am in i got book screennn WOOOOOT 11:11 < Snake007uk> oh :* 11:11 < riel> but I have an explanation ;) 11:11 < riel> well, can you log in ? 11:11 < Snake007uk> yes 11:11 < Snake007uk> :) 11:12 < riel> rock 11:12 < Snake007uk> no password ? 11:12 < Snake007uk> but anyway 11:12 < riel> not yet, you need to set that yourself ;) 11:12 < Snake007uk> ok 11:12 < Snake007uk> but what baout udev issue 11:12 < Snake007uk> is that not normal then 11:12 < riel> not normal, but it doesn't matter 11:12 < riel> your system runs ;) 11:12 < Snake007uk> lol 11:12 < Snake007uk> i dont like broken stuff 11:12 < Snake007uk> what do you think the problem will be ? 11:12 < riel> what's in /dev ? 11:13 < Snake007uk> there is alot of stuff in there 11:13 < riel> then udev did its job 11:13 < Snake007uk> oh ok 11:13 < MarkW> Snake007uk: you created the full set of device nodes 11:14 < Snake007uk> let me install gpg keys for yum :) 11:14 < MarkW> riel: could that have upset udev? 11:14 < Snake007uk> when i detahc with crtl+] how do i go back i n? 11:14 < MarkW> Snake007uk: so actually you don't need udev 11:14 < MarkW> Snake007uk: "xm console " 11:14 < Snake007uk> ah ok 11:14 < riel> MarkW: not really 11:14 < riel> MarkW: udev simply mounts a tmpfs on top of /dev 11:15 < MarkW> riel: weird. it could be completely broken, given the nodes were there anyway 11:15 < riel> true, but udev can't do anything useful anyway 11:15 < riel> since xenbus doesn't exist yet 11:16 < Snake007uk> :) i got network up :) 11:16 < Snake007uk> its needs its own ip ofcourse 11:16 < Snake007uk> or can youbridge it >? 11:17 < riel> with bridging, it'll also need its own ip ;) 11:17 < MarkW> riel: well, it still creates device nodes for xen devices, i believe? 11:17 < Snake007uk> oh ok 11:17 < riel> MarkW: true, but without hotplug udev creates them at startup time 11:17 < riel> MarkW: not having it running later on doesn't matter 11:18 < MarkW> riel: ah, ok. 11:18 < knewt> riel: and what if you add extra devices during runtime? 11:18 < MarkW> riel: i thought hotplugging happened automagically for devices in sysfs? 11:18 < Snake007uk> just running yum update now :) 11:18 * riel never tried that ;) 11:18 < Snake007uk> then i can install x 11:18 < Snake007uk> :) 11:18 < Snake007uk> see how this is 11:18 < knewt> MarkW: it does i think 11:18 < MarkW> without explicit support? 11:18 < riel> Snake007uk: X inside a guest won't work - but VNC should work 11:19 < Snake007uk> yeah 11:19 < Snake007uk> thats what meant 11:19 < MarkW> Snake007uk: or you can use ssh -X to forward X apps to dom0 11:19 < knewt> MarkW: as long as the "dev" entry appears for the device the device node will be added 11:19 < Snake007uk> but xenus should a,llow X in vm shuold it 11:19 < MarkW> knewt: that's what i thought... 11:19 < MarkW> Snake007uk: can't run an X server in a VM, you'll have to use a network-based solution as if it was a separate machine 11:20 < riel> Snake007uk: nope, X wants direct hardware access 11:21 < MarkW> yey, the kexec tools now run under XenLinux! 11:21 < Snake007uk> ah ok, well 11:21 < MarkW> (they don't do anything, though ;-)) 11:21 < Snake007uk> i think xen is mainly for server market anyway rather then desktop :) 11:22 < Snake007uk> if the guest os crashes can its kernel make xen crash ? 11:22 < riel> it shouldn't be able to 11:23 < Snake007uk> ok 11:23 < knewt> although i've never tried it i'd have thought you could run an Xnest in the domU, pointing back out to dom0? 11:23 < MarkW> Snake007uk: cool graphics stuff will probably happen later on 11:23 < mael> MarkW: is it normal that I have trouble to exit a xm console using Ctrl+] ? 11:23 < Snake007uk> ethtool -s eth0 speed 100 autoneg off duplex full 11:23 < Snake007uk> Cannot get current device settings: Operation not supported 11:23 < Snake007uk> not setting speed 11:23 < Snake007uk> not setting duplex 11:23 < Snake007uk> not setting autoneg 11:24 < MarkW> knewt: you could run xnest in dom0 and have it display apps from the domU, or you could run xnest in the domU and X forward it. the first would be better. 11:24 < MarkW> mael: nope, that's not normal 11:24 < MarkW> what's happening? 11:24 < mael> well nothing :) 11:24 < mael> I'm stuck in the domU 11:24 < MarkW> Snake007uk: don't do that to a Xen virtual ethernet dev, it doesn't need it... 11:24 < knewt> MarkW: i meant if you wanted all the apps in the domU to appear within a single desktop nicely. otherwise you could just point the apps directly at a normal X in dom0 :) 11:24 < mael> can it be a problem with french keyboard? 11:25 < MarkW> knewt: I know, but running the Xnest in the dom0 would still be better 11:25 < Snake007uk> shoudl all things like hardware setting be set out side the vm 11:25 < MarkW> mael: shouldn't be although I've never tried it 11:25 < MarkW> Snake007uk: hardware settings in dom0 11:26 < mael> using a french keyboard you have to use Ctrl+AltGr+) to get Ctrl+] 11:26 < Snake007uk> ok got it 11:26 < MarkW> mael: ah, ok 11:26 < Snake007uk> make sense 11:26 < MarkW> mael: does it work then? 11:26 < mael> no :( 11:26 < MarkW> mael: can you log into dom0 from somewhere else? 11:27 < knewt> mael: i have the same problem if i use my dads german keyboard. try ctrl together with the key 2 to the right of the P 11:27 < mael> mmh no, it is running in qemu and I don't know how to switch the virtual console 11:27 -!- cc [~cc2@c210-49-121-44.eburwd3.vic.optusnet.com.au] has joined #xen 11:28 < MarkW> mael: i don't suppose you could ssh to dom0 from the domU? 11:28 < MarkW> it's like being stuck in the matrix! 11:28 < Snake007uk> riel what image is the vm fc3 ? 11:28 < mael> knewt: that's it 11:28 < knewt> that's where the ] is on a us/gb keyboard ;) 11:29 < mael> Ctrl and the key next to the enter key works 11:29 < mael> hehe 11:29 < mael> well this is obviously a bug :) 11:29 < knewt> it's doing the check based on scancode rather than keycode obviously 11:29 < mael> MarkW: shall I file a bug against xm? 11:30 < MarkW> mael: sounds like a good plan 11:31 < knewt> i'd forgotten about it because i haven't accessed a domU console from my dad's computer for aaaaaaages 11:32 < mael> MarkW: xm is a tool right? 11:32 < MarkW> mael: yup 11:32 * mael hates filing a bug using bugzilla 11:38 < mael> done 11:39 < MarkW> mael: thanks 11:41 < mael> you're welcome, I'm always happy to add work to other ppl 11:42 < MarkW> mael: :-) 11:44 -!- cfreak [cfreak@dsl-084-056-101-071.arcor-ip.net] has joined #xen 11:48 < MarkW> yey! now I can crash domains on demand! there's nothing to this development lark! 11:49 < knewt> so what exactly is the point of xenbus? is it just to add extra sysfs entries, so that the existing class entries can get "device" symlinks? 11:49 < MarkW> knewt: one thing it'll enable is enumerating the devices before their modules are loaded 11:50 < MarkW> so you can know what virtual hardware you have... like you can enumerate a PCI bus without having all the drivers 11:50 < mael> "development lark"... well understanding is far beyond my english capabilities 11:50 < mael> or shall I say capability 11:50 < knewt> ah, so it's a prequel into making the drivers able to be made as modules? 11:50 < MarkW> mael: a "lark" is slang for something that's a trivial bit of fun. in this case, it's a joke ;-) 11:51 < mael> hu! thx for decoding :) 11:51 < MarkW> knewt: yeah, think so. it's also related to the registry stuff. 11:51 < MarkW> mael: in other contexts it's also a kind of bird 11:52 < MarkW> mael: english is really weird ;-) 11:52 < mael> yeah I only had the bird stuff in my dictionnary :) 11:52 < knewt> MarkW: how will the registry be done? will it just be another domain that's always there and the other domains can talk to, but with special code running in it? 11:52 -!- aliguori [~anthony@cpe-70-112-81-91.austin.res.rr.com] has joined #xen 11:52 < mael> yeah! 11:53 < mael> more work to give! 11:53 < mael> humm 11:53 < mael> hello aliguori, howdy? 11:57 < MarkW> knewt: it'll initially be a daemon in dom0 11:57 < MarkW> knewt: but it'll need to scale up beyond that later on... 12:00 < Snake007uk> ok so far so good yum is updating 12:00 < Snake007uk> how slow is the vm compared to native ?? 12:00 < Snake007uk> for example 12:01 < Snake007uk> lets say i had a p4 3ghz 12:01 < Snake007uk> with 1 gig ram 12:01 < Snake007uk> so i gave dom0 256 ram, and 3 vm 256 ram 12:01 < Snake007uk> would they all run okish ? 12:01 < Snake007uk> as if they were on a system p4 3ghz with 256 natively 12:04 < MarkW> Snake007uk: within a few percent for most workloads 12:04 < Snake007uk> wow 12:04 < Snake007uk> that amazing 12:06 < Snake007uk> ok 12:06 < Snake007uk> got a prob 12:06 < Snake007uk> file system has ran out of space 12:06 < Snake007uk> can i expand the size ? 12:08 < MarkW> Snake007uk: possibly, not sure 12:08 < mael> aliguori: ping? 12:08 < aliguori> mael: pong.. i'm a bit busy right now :-) i'll get back you in a bit 12:09 < mael> ok will you still be there in 2hours? 12:09 < mael> I'm almost off 12:09 < aliguori> probably not, feel free to send me an email if you'd like (aliguori@us.ibm.com) 12:09 < mael> hum ok 12:10 < mael> what time is it for you? 12:10 < aliguori> 11am, but i'm pretty much booked solid until 5pm 12:11 < mael> ok 12:11 < mael> hum 5pm will be 00-1am for me 12:12 -!- heb [~harry@blueice3n1.uk.ibm.com] has quit [Quit: Leaving] 12:12 < Snake007uk> aliguori you work for ibm cool :) 12:13 < aliguori> Snake007uk: as do a lot of folks in this channel 12:13 < aliguori> :-) 12:13 < Snake007uk> i need job 12:13 < Snake007uk> :) 12:13 < Snake007uk> uk though 12:13 < Snake007uk> any managers about 12:13 < Snake007uk> :D 12:13 < Snake007uk> hehehe 12:13 < aliguori> I'm just a lowly programmer :-) 12:13 < Snake007uk> ah 12:13 < Snake007uk> oh well 12:13 < Snake007uk> well my cv is up 12:13 < aliguori> and I don't think IBM's doing all that well in that area of the world right now... 12:13 < Snake007uk> yeah 12:14 < Snake007uk> i just remembered :( 12:15 -!- xai [~pasta@cpe-70-112-17-10.austin.res.rr.com] has quit [Remote host closed the connection] 12:18 < demon> anyone know why it is the linux 2.4.29 sparse tree in xen 2.0-testing doesn't build when applied to 2.4.29? 12:18 < MarkW> demon: what breaks? 12:18 < MarkW> demon: 2.4 builds sometimes don't work because nobody tested them for a while! 12:20 < demon> it looks like someone's trying to force some 2.6 stuff into it... lemme ssh into the box I was trying to build on, sec 12:20 < MarkW> demon: i'll give it a test build but it'd be useful if you could have a look 12:21 -!- knewt [~jmb@zeus.pimb.org] has quit [Read error: Connection reset by peer] 12:21 < demon> make[1]: Entering directory `/usr/local/src/linux-2.4.29/arch/xen/kernel' 12:21 < demon> make[1]: *** No rule to make target `gnttab.c', needed by `i386_ksyms.o'. Stop. 12:22 -!- knewt [~jmb@p213.54.97.31.tisdip.tiscali.de] has joined #xen 12:22 < demon> there's no gnttab.c in linux 2.4.29 12:25 < MarkW> demon: hrrrmmmmm 12:25 < Snake007uk> ok guyz i am off MarkWm, reil thanks your for your help 12:25 < Snake007uk> cya 12:25 -!- Snake007uk [~Snake007u@194.159.31.243] has quit [Quit: ] 12:25 < demon> I tried removing the reference to 'gnttab.c' in the makefile in arch/xen/kernel/, and now I get: 12:25 < demon> arch/xen/kernel/kernel.o(.text+0x8462): In function `xen_contig_memory': 12:25 < demon> : undefined reference to `pte_val_ma' 12:25 < MarkW> demon: nasty! 12:25 < demon> when it goes to link the kernel 12:25 < MarkW> lets see what it does on my system... 12:34 < MarkW> demon: it doesn't even get that far on mine at the moment... 12:34 * MarkW investigates 12:34 < demon> well, I'm just building a domU kernel 12:35 < demon> a dom0 kernel might have more issues, but I hadn't had an interest in building a 2.4 dom0 kernel 12:35 -!- aliguori [~anthony@cpe-70-112-81-91.austin.res.rr.com] has quit [Quit: Leaving] 12:35 < MarkW> it won't even start building a 2.4 kernel for me at the moment. 12:36 < demon> ... that's interesting 12:36 < demon> well, at least I know it's not just me 12:37 -!- athomas [~athomas@ppp-0-52.lond-a-1.access.uk.tiscali.com] has quit [Quit: Leaving] 12:37 < MarkW> think i'll blow away my testing tree and start again 12:42 < MarkW> still broken 12:42 < MarkW> this is a different break to the one you're seeing though... the build system is acting up 12:47 < demon> well, I'm using the tarball that's linked off the Xen site 12:47 < demon> I'm guessing you're pulling straight out of the bk tree? 12:48 < MarkW> yes 12:49 < MarkW> i need to fix this first 13:02 < MarkW> demon: Right, I've fixed the latest build problem (which is probably too new for you to see), now I'll let it build and see if your problem occurs too. 13:03 -!- visik7 [~ciao@host124-39.pool80182.interbusiness.it] has joined #xen 13:03 < visik7> hi 13:04 < MarkW> visik7: hi 13:05 < visik7> is there a way to use xen for emulating a computer network like uml ? (so practically 2 guest are connected to eachother but not connected to the host ?) 13:05 < visik7> s/guest/domU 13:05 < visik7> s/host/dom0 13:06 < demon> yeah, you would just set up a separate bridge and have the two domains set to bridge their virtual network devices to that 13:06 < MarkW> well, network is generally routed / bridged by dom0 but dom0 doesn't really have to care about what goes on 13:07 -!- HoraPe [~horape@200.69.230.10] has quit [Read error: Connection reset by peer] 13:07 < visik7> demon and how can I use for example dom1 to route packet from dom2 to dom3 without using dom0 routing ? 13:08 < visik7> 2 ethernet x domain ? 13:08 < visik7> 2 ethernet per domain ? 13:10 -!- franR_ [~franr@inextensia.net] has joined #xen 13:10 -!- franR [~franr@inextensia.net] has quit [Read error: Connection reset by peer] 13:13 < demon> .... I guess I don't entirely understand the layout you're looking for 13:13 < demon> but you can have multiple virtual network devices assigned to a single domain, yes 13:14 < visik7> 3 domU 13:14 < visik7> A B and C 13:14 < visik7> A is connected to B and B is connected to C 13:14 < demon> okay 13:14 < visik7> and B routes packet between A and C 13:14 < demon> so set up two empty bridges 13:14 < visik7> without using dom0 routing capabilities 13:15 < demon> configure A to use the first bridge, B to use both, and C to use the second 13:15 < visik7> ah ok 13:15 < knewt> you can set B up as a network backend and hook A and C up directly to it. only this is, right now you'll have to fake out xen into giving the correct access permissions 13:15 < knewt> s/only this is/only thing is/ 13:16 < visik7> I don't understand why uml use collision domain instead of bridging 13:16 < knewt> but it'll work 13:18 -!- hollis [~hollis@pixpat.austin.ibm.com] has joined #xen 13:31 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has quit [Read error: Connection reset by peer] 13:53 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen 13:53 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has left #xen [] 13:53 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen 13:53 -!- mode/#xen [+o MarkW] by ChanServ 13:54 -!- Tv [~Tv@hq.inoi.fi] has quit [Quit: Client exiting] 14:31 -!- Arnald [~Arnald@81-86-116-102.dsl.pipex.com] has joined #xen 15:01 -!- tierra [~tierra@dsl093-225-126.slc1.dsl.speakeasy.net] has joined #xen 15:18 -!- DEac- [~deac@xdsl-84-44-145-179.netcologne.de] has quit [Quit: Verlassend] 15:19 -!- DEac- [~deac@xdsl-84-44-145-179.netcologne.de] has joined #xen 15:21 -!- riel is now known as | 15:21 -!- | is now known as riel 15:36 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has quit [Remote host closed the connection] 15:50 -!- JViz [Anomaly@cpe-065-190-045-031.triad.res.rr.com] has quit [Quit: YES THEY DESERVE TO DIE, AND I HOPE THEY BURN IN HELL!] 16:02 -!- stekloff [~stekloff@bi01p1.co.us.ibm.com] has joined #xen 16:12 -!- cfreak [cfreak@dsl-084-056-101-071.arcor-ip.net] has quit [Quit: .] 16:16 -!- DEac- [~deac@xdsl-84-44-145-179.netcologne.de] has quit [Ping timeout: 480 seconds] 16:25 < hollis> omg, another 30 commits 16:25 -!- franR_ [~franr@inextensia.net] has quit [Quit: Error inopinée -10s, -9, -8... Unexpected error -10s, -9, -8,... Really ?] 16:27 -!- DEac- [~deac@xdsl-213-196-203-92.netcologne.de] has joined #xen 16:31 -!- soffi_ [~soffi@Tolvudeild-SC-gw.vv.simnet.is] has joined #xen 17:07 -!- soffi_ [~soffi@Tolvudeild-SC-gw.vv.simnet.is] has quit [Quit: Leaving] 17:10 -!- aliguori [~anthony@cpe-70-112-81-91.austin.res.rr.com] has joined #xen 17:10 -!- yarihm [~yarihm@80-218-3-145.dclient.hispeed.ch] has joined #xen 17:16 < aliguori> oy, the hypercalls are changing interfaces like crazy 17:26 < riel> there's a reason x86-64 doesn't even compile part of the time ;) 17:26 < hollis> seems like we're making it as unstable as possible before calling it stable 17:29 < rharper> hey, I sent those patches a month ago =) 17:29 < hollis> :) 17:52 -!- JViz [Anomaly@cpe-065-190-045-031.triad.res.rr.com] has joined #xen 17:57 < aliguori> it's the mad dash before beta/alpha/testing.. it's always a joyious time in any project 18:43 -!- katzj [~katzj@wlanconf-nat-pool-bos.redhat.com] has quit [Read error: Connection reset by peer] 18:56 -!- yarihm [~yarihm@80-218-3-145.dclient.hispeed.ch] has quit [Quit: Leaving] 18:56 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen 18:56 < MarkW> /nickserv identify blibble 18:56 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has left #xen [Kopete 0.10 : http://kopete.kde.org] 18:56 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen 18:56 < visik7> MarkW interessante 18:56 < visik7> :) 18:57 < MarkW> visik7: oops, i think i did that wrong! 18:57 < visik7> ops 18:57 < visik7> :) 18:57 < visik7> change your password asap :) 18:57 < MarkW> yup will do 18:57 < MarkW> ;-) 18:59 < MarkW> that's better! 18:59 < MarkW> I can't believe I just did that (sticks fist in mouth)! 18:59 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has left #xen [Kopete 0.10 : http://kopete.kde.org] 18:59 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen 18:59 -!- mode/#xen [+o MarkW] by ChanServ 18:59 < MarkW> That's more like it! 19:00 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has quit [Remote host closed the connection] 19:00 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen 19:00 < MarkW> oops 19:01 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has left #xen [] 19:01 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen 19:01 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has left #xen [] 19:01 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen 19:01 -!- mode/#xen [+o MarkW] by ChanServ 19:05 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has left #xen [] 19:05 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has joined #xen 19:05 -!- mode/#xen [+o MarkW] by ChanServ 19:07 < hollis> MarkW: you can tell chanserv to op you without leaving and rejoining 19:07 < MarkW> hollis: I'd heard that but I never figured out how! 19:07 < hollis> /msg chanserv help op 19:08 < MarkW> Ahhhhh, ok. 19:08 < MarkW> Thanks :-) 19:08 < visik7> /cs op #xen MarkW 19:08 < MarkW> Ah cool 19:13 < MarkW> hollis: I finally figured out that iomem issue - latest unstable actually claims to have System RAM ;-) 19:14 < hollis> MarkW: cool 19:14 < hollis> what was the issue? 19:15 < MarkW> We build a fake e820 map - just one entry for all of RAM 19:15 < MarkW> But it was starting at address zero, which conflicted with the various random resources at the bottom of the x86 machine address space 19:16 < MarkW> So Linux wouldn't register a "struct resource". I cleaned up some random resources *not* to register in domUs and fixed the e820 map. 19:16 < hollis> what's down there? I know it's full in IO space, but RAM shouldn't conflict with that... 19:16 -!- katzj [~katzj@c-66-30-206-55.hsd1.ma.comcast.net] has joined #xen 19:16 < MarkW> System ROM, video rom, video ram 19:17 < MarkW> It's probably IBM's fault it's like that ;-) 19:17 < aliguori> MarkW: it is :-) 19:17 < aliguori> computers weren't supposed to have more than 4MB of memory 19:18 < MarkW> yes, that went a bit wrong, didn't it? :-) 19:18 < MarkW> anyhow, I only cared about /proc/iomem because the kexec tools get unhappy if they don't find any system RAM!!! 19:20 < MarkW> so now I can go back to my regularly scheduled kernel butchering :-) 19:23 -!- rharper [~rharper@pixpat.austin.ibm.com] has quit [Quit: Leaving] 19:23 -!- soffi [~soffi@proxy.du.vdsl.is] has joined #xen 19:29 -!- rusty [~rusty@bh02i525f01.au.ibm.com] has joined #xen 19:47 < rusty> Hi all. 19:47 < aliguori> howdy rusty 19:47 -!- hollis [~hollis@pixpat.austin.ibm.com] has quit [Quit: leaving] 19:53 < rusty> aliguori: hi! 20:30 -!- rusty [~rusty@bh02i525f01.au.ibm.com] has quit [Ping timeout: 480 seconds] 20:36 -!- stekloff [~stekloff@bi01p1.co.us.ibm.com] has quit [Quit: Leaving] 20:39 -!- JViz [Anomaly@cpe-065-190-045-031.triad.res.rr.com] has quit [Ping timeout: 480 seconds] 20:44 -!- rusty [~rusty@bh02i525f01.au.ibm.com] has joined #xen 20:49 -!- liberie [~root@dsl027-160-029.atl1.dsl.speakeasy.net] has quit [Quit: Lost terminal] 21:17 < rusty> aliguori: DO you know the #include magic to successfully include xc.h? Trying to tie it into Xen Store daemon, and it needs cpumap_t etc. 21:25 < rusty> aliguori: ah, figured it out. Need a symlink "xen" to ..../public. 21:26 < MarkW> rusty: i've dived into the tools header files on a couple of occasions and ended up very confused 21:26 < MarkW> then eventually found the Magic Symlinks 21:27 < rusty> MarkW: Yes, I'm bulding outside the tree, trying to use libxc. If the name "public" were "xen", I could just use the right -I, so xc.h's #include etc. would work. 8( 21:34 < MarkW> rusty: hmm, not so helpful 21:34 < rusty> Well, it "works" for the moment 8) 21:34 < MarkW> "working" is always good :-) 21:35 -!- cc [~cc2@c210-49-121-44.eburwd3.vic.optusnet.com.au] has quit [Ping timeout: 480 seconds] 21:37 < rusty> MarkW: I said I'd have domain <-> Xen store communication working this week... didn't realize it was already Wed night when I said that 8) 21:38 < MarkW> Ah :-) 21:38 < rusty> MarkW: A little pressure is healthy (so I keep telling myself!) 21:39 < MarkW> Are you using the control messages underneath it all? How painful is it mapping the store operations onto them? 21:39 < rusty> MarkW: Since people hated the control messages, I chose a simple share page ringbuffer + eventchn. 21:40 < rusty> MarkW: I already had that code from th Research Hypervisor at IBM, so I scraped off the serial numbers. 21:40 < MarkW> rusty: so control messages are going away completely? 21:40 < rusty> MarkW: Not for 3.0, AFAICT. 21:41 < MarkW> no, that might be a bit short notice 21:41 < MarkW> The the current control message / device channel protocol has caused me substantial pain however, so it'll be very welcome to have something nicer :-) 21:43 < rusty> Right. 22:21 < MarkW> rusty: I always seem to end up needing to add *loads* of control messages whenever I work on something :-) 22:29 < soffi> MarkW: I'm dumb, so tell me... XenFS.. it runs on top of another filesystem just like nfs does? 22:29 < soffi> etc. it exports dirs ? 22:29 < MarkW> soffi: yup 22:29 < soffi> cool 22:30 < soffi> so I could use Coda in-between 22:30 < MarkW> in principle, yes 22:31 < soffi> but in practice... big no no ? :) 22:32 < MarkW> soffi: it should work eventually, provided that coda provides the right ops to be re-exportable 22:32 -!- JViz [Anomaly@cpe-065-190-045-031.triad.res.rr.com] has joined #xen 22:32 < soffi> cool 22:33 < soffi> I've been getting headaches looking at active-active storage in general 22:34 < MarkW> soffi: ? 22:35 < soffi> I'm reading up on what to get for my company and I'm hyped on getting two storage machines that provide active-active failover for the storage 22:35 < MarkW> ah 22:36 < soffi> I'm going on vacation in 2 weeks and I want it up by then, and I want it active-active so I can actually stay on vacation ;) 22:38 < soffi> and I've been so unlucky with storage solutions these past months that almost 90% of my work has been recovery 22:39 < MarkW> soffi: right, sounds like you need something there :-) 22:40 < soffi> then I could do more fun stuff, and hacking on company time :) 22:40 < soffi> hey what's your idea with xen-get ? 22:40 < MarkW> soffi: yeah, i would have sent you a copy already, except my filesystem died immediately 22:41 < MarkW> so I lost it :-( 22:41 < soffi> so we have something in common 22:41 < soffi> filesystems suck 22:41 < soffi> hehehe 22:42 < MarkW> soffi: anyhow, basically it was modelled on apt in some ways... 22:42 < MarkW> it downloaded tarballs from your site and stored them in a cache directory 22:42 < MarkW> then when you install, it'd get them from the local cache if possible. it would install automatically into a file disk, a directory or a device 22:43 < soffi> coool 22:43 < MarkW> the only thing that was missing was some sort of metadata 22:43 < soffi> does it run on windows xp 22:44 < soffi> I might put up an RSS file on the site 22:44 < soffi> describing the tarballs and so on 22:44 < MarkW> Ah I was just going to suggest we had some sort of index file :-) 22:44 < soffi> would that do any good ? 22:44 < soffi> :) 22:44 < soffi> great minds think alike 22:44 < MarkW> That'd be great, then we could have searchable metadata. 22:45 < MarkW> What we could do with at the minimum is package name, tarball path, description. 22:46 < soffi> like; fedoracore4|/soffi/fedoracore4.tar.gz|A wonderful core of fedora 22:46 < soffi> only in xml 22:47 < MarkW> Yeah, that sounds fine. 22:48 < MarkW> The original code was just a bash script. The reincarnation should probably be in Python, which make XML parsing easier and make it easier to integrate with a GUI ;-) 22:49 < soffi> hehe gui 22:49 < soffi> maybe Steve Jobs can help you out 22:49 < soffi> his latest discovery seems to be xml 22:50 < soffi> I just installed his Tiger on my laptop and it has XML in every corner 22:51 < MarkW> yeah, apple do seem to be keen on xml-ifying everything they see 22:51 < MarkW> maybe they'll put xml in the kernel :-D 22:52 < soffi> and not commit it back to FreeBSD 22:52 < soffi> ::) 22:53 < aliguori> i can't even tell you how many times i've heard someone say domain configuration files should be in xml... 22:53 < MarkW> aliguori: they have been 22:53 < aliguori> oh gees :-) 22:53 < MarkW> if you go back into the mists of time and look at Xen 1.0 / 1.1 :-) 22:54 < aliguori> thank god... :-) 22:54 < MarkW> the tools were written in Java back then 22:54 < aliguori> oh yeah 22:54 < aliguori> i remember that 22:54 < aliguori> i remember downloading a copy of xen and not being able to use it because i didn't want to install java on my system :-) 22:55 < MarkW> the guy who wrote them was a smart chap so i imagine they were quite good tools... and there was a web interface... but Java was a bit slow for a one shot tool (esp on the demo CD) 22:58 < aliguori> I imagine.. I haven't seen one of you guys that hasn't been smart :-) 22:58 < soffi> xml is good for parsing though 22:59 < aliguori> it's heavy weight though.. a full xml parser is quite complex 23:00 < soffi> feed://www.kvadratrot.net/~xen/tarballs.xml 23:00 < soffi> feed? 23:00 < soffi> why does Steve Jobs do this 23:00 < soffi> http://www.kvadratrot.net/~xen/tarballs.xml 23:01 < MarkW> soffi: how strange. your feed:// link made my RSS reader open and my taskbar disappear :-S 23:02 < soffi> :S 23:02 < soffi> all blame steve 23:02 < soffi> If you open up the images page, firefox should recognize the feed and be able to display it 23:02 < soffi> but I'm an Apple whore so I only use Safari 23:02 < MarkW> soffi: it'd be good if we could have a "package name", vs the title / description 23:03 < MarkW> I use Konqueror :-) 23:03 < soffi> daerly noted 23:03 < soffi> dearly 23:03 < soffi> isn't Safari based on that? 23:04 < MarkW> yup, although there's now some friction between apple and the khtml people 23:04 < MarkW> konqi gives me a little RSS icon on your site and I can read the feed :-) 23:04 < soffi> cool 23:04 < soffi> try it now.. I've added 23:05 < soffi> maybe it should be title for package name and description for the description 23:05 < MarkW> soffi: it'd be nice to have for the canonical package name 23:06 < soffi> ok 23:06 < MarkW> title for the long format title 23:06 < MarkW> and description for a more verbose description if anyone wants to write one 23:06 < soffi> so I give you the rss feed alpha beta gamma v.0.1 23:06 < MarkW> eh? 23:08 < soffi> hey although I'm not a developer I can still make up silly version names 23:08 < MarkW> heh :-) 23:08 < MarkW> Hey, you're an Apple user, that probably makes you more qualified to do that ;-) 23:09 < soffi> ok so this has <package><description><link><date> 23:09 < soffi> and only 2 more weeks until Safari is going to refresh it 23:09 < MarkW> That sounds good to me 23:09 < soffi> this thing is caching like a *other****er 23:09 < soffi> and I can't get it to not do that 23:10 < MarkW> Oh, and could you stick a version field for the feed as a whole? So that if we change some fields we can have a "you need to upgrade your xen-get" message? 23:10 < soffi> hmm ok 23:10 < soffi> I've added <revision> within <channel> 23:11 < MarkW> better to have it in there, just in case we want to change the protocal :-) 23:11 < soffi> yup 23:11 < soffi> Steve Jobs would be so proud of us now 23:11 < MarkW> perhaps we should start wearing black turtleneck sweaters? 23:12 < soffi> not my thing 23:12 < MarkW> but Steve wears them and he can't possibly be wrong! 23:12 < MarkW> dare you defy steve jobs? your computer will break! 23:12 < soffi> Steve Jobs is two whole mousebuttons close to insane 23:13 < MarkW> hmmm. the two ends of this communications code i've written have very different ideas about what the actual data is. 23:13 < soffi> let them be alone for a few hours 23:13 < MarkW> oh actually, no they don't, it's just the wrong data :-) 23:14 < soffi> get to know each other 23:14 < soffi> hehee 23:14 < MarkW> it's not what i *wanted* the data to say, anyhow 23:14 < MarkW> so i'll blame them anyway 23:20 < soffi> hehe 23:20 < soffi> usa requiring every passenger with an rfid passport after june 27th. 23:21 < soffi> I'll blame Steve 23:24 < soffi> well.. I'm going to fall asleep over a BBC documentary 23:28 -!- katzj [~katzj@c-66-30-206-55.hsd1.ma.comcast.net] has quit [Quit: .] 23:28 < MarkW> soffi: ok. think i should go to bed soon. i don't think i'm going to get this nailed down tonight... 23:28 < MarkW> and the sun is starting to rise again 23:28 < MarkW> i hate it when it does that 23:40 -!- MarkW [~MarkWilli@maw48.kings.cam.ac.uk] has quit [Read error: Connection reset by peer] 23:51 -!- aliguori [~anthony@cpe-70-112-81-91.austin.res.rr.com] has quit [Quit: Leaving] 23:55 -!- homebaum [~michael@wbar1.sea1-4-5-031-104.sea1.dsl-verizon.net] has joined #xen --- Log closed Thu May 12 23:59:00 2005