TheGrebs.com:IRC Logs:#xen:2006:02:22

---	Log	opened Wed Feb 22 00:00:22 2006
00:20	\|-\|	JViz` [Anomaly@adsl-065-013-131-023.sip.int.bellsouth.net] has joined #xen
00:20	\|-\|	JViz [Anomaly@adsl-065-013-131-023.sip.int.bellsouth.net] has quit [Read error: Connection reset by peer]
00:30	<teferi>	Does anyone have a newer version of the Forcedeth driver than the one included in 2.6.12.6-xen that works with it?
00:35	<teferi>	Never mind!
00:44	\|-\|	muli [~muli@87.69.40.180.cable.012.net.il] has quit [Quit: My damn controlling terminal disappeared!]
01:08	\|-\|	rusty [~rusty@bh02i525f01.au.ibm.com] has quit [Quit: Client exiting]
01:17	\|-\|	muli [~muli@192.114.107.4] has joined #xen
01:26	\|-\|	yorkedork [~thom@nom11167it.nomadic.ncsu.edu] has joined #xen
01:37	\|-\|	cods [~cods@tuxee.net] has quit [Ping timeout: 480 seconds]
01:41	\|-\|	ns [~nivedita@c-67-171-181-157.hsd1.or.comcast.net] has quit [Quit: Leaving]
01:55	\|-\|	yorkedork [~thom@nom11167it.nomadic.ncsu.edu] has quit [Quit: Leaving]
02:27	\|-\|	pvanhoof [~pvanhoof@mailhost.newtec.be] has joined #xen
02:28	\|-\|	psuresh [~psuresh@125.16.129.15] has joined #xen
02:32	\|-\|	Teltariat [~tekron@h-68-164-200-156.nycmny83.dynamic.covad.net] has joined #xen
02:32	<Teltariat>	whoa
02:32	<Teltariat>	didn't know there was a #xen here
02:32	<Teltariat>	eh heh heh... err... greets folks :P
02:32	\|-\|	psuresh [~psuresh@125.16.129.15] has quit [Quit: ]
02:33	<Teltariat>	I'm wondering if anyone has gotten FreeBSD going in Xen 3
02:33	<Teltariat>	I've found the tools over at fsmware.com, but I can't get them to work for the life of me
02:58	\|-\|	Basic_py [~Basic@warden.real-time.com] has quit [Ping timeout: 480 seconds]
03:05	<Teltariat>	.....anyone...? :=[
03:42	\|-\|	rusty [~rusty@ppp61-206.lns1.cbr1.internode.on.net] has joined #xen
03:46	\|-\|	sdog [~sdog@62.58.98.210] has joined #xen
04:03	\|-\|	ivan [~ikelly@86.41.193.236] has joined #xen
04:08	\|-\|	sdog [~sdog@62.58.98.210] has left #xen []
04:18	\|-\|	tab [~tab@darwin.snarc.org] has joined #xen
04:20	\|-\|	tab_ [~tab@darwin.snarc.org] has quit [Ping timeout: 480 seconds]
04:27	<mancdaz>	I have xen 3 runnig on fc5t3, with an fc5t3 guest. I notice the guest is running with kernel 2.6.15-1.1955_FC5guest, but cannot find this kernel anywhere on the host system. Where does this guest kernel come from?
04:29	<Teltariat>	uname output != name of kernel file
04:29	<Teltariat>	You should check the guest's configuration file, prolly somewhere in /etc/xen
04:29	<Teltariat>	the kernel its using should be specified there
04:30	<mancdaz>	there is no kernel specified there
04:30	<mancdaz>	this config file was generated by xenguest-install.py
04:30	<mancdaz>	in the sample config files a guest kernel is specified
04:30	<Teltariat>	Dunno, you may have to read that py file
04:30	<mancdaz>	this is what's confusing me
04:31	<Teltariat>	then use locate or something to try to find the file
04:31	<Teltariat>	thats specified.
04:31	<mancdaz>	is it possible to run a suse guest on a fc xen host?
04:32	<Teltariat>	I dunno. Never tried anything other than Gentoo and Debian.
04:32	<mancdaz>	but you have, say a gentoo guest on a debian system for example
04:34	\|-\|	boncos [~boncos@202.162.208.126] has joined #xen
04:36	<boncos>	anybody can help me ? i can't start xen on centos 4.2
04:36	<boncos>	i got this error msg: gaierror: (-2, 'Name or service not known')
04:43	<Teltariat>	mancdaz: I think it depends more on how well a distro can work with a (relatively) vanilla kernel
04:43	<Teltariat>	(and certain missing amenities)
04:44	<mancdaz>	I found the guest kernel - the xenguest-install.py script pulls it from the install source
04:44	<mancdaz>	rather than using a local copy from the host
04:44	<mancdaz>	suse 10 has xen-3 support so in theory it should be able to work if I can get the guest kernel for it
04:45	<mancdaz>	and find some way to actually install the OS into the guest
04:49	<Teltariat>	I need to do the same for Fedora Core. So I need to figure out how to get yum running in Gentoo. :P
04:50	<Teltariat>	(don't want to have to reboot to install using the Fedora CDs, need to install FC into a guest while the machine stays running)
04:50	<Teltariat>	might have to do it on another machine and tarball the filesystem
04:50	\|-\|	ivan_ [~ikelly@86.41.204.130] has joined #xen
04:50	<mancdaz>	you can use the process from the fedora 4 quickstart guide - it uses a kind of chroot to install the fc base system into your guest filesystem
04:51	<mancdaz>	in theory
04:51	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has joined #xen
04:52	<Teltariat>	yea, but you're using yum to install Fedora into the chroot
04:52	<mancdaz>	ah right
04:52	<mancdaz>	you have no yum
04:52	\|-\|	ivan [~ikelly@86.41.193.236] has quit [Ping timeout: 480 seconds]
04:55	<mancdaz>	I have no yast
05:08	<mancdaz>	it seems that maybe suse doesn't use separate guest and hypervisor kernels
05:08	<mancdaz>	just vmlinuz-xen for both
05:09	<hensema>	they don't
05:09	<hensema>	both dom0 and domU use the same kernel in suse
05:09	<hensema>	all bare metal hardware drivers are kernel modules
05:09	<hensema>	they simply aren't loaded in domU
05:10	<hensema>	so you waste a couple of MB of diskspace
05:10	<hensema>	big deal ;)
05:12	<mancdaz>	fair point :-)
05:12	<boncos>	anybody know what error cause this msg? gaierror: (-2, 'Name or service not known')
05:14	<mancdaz>	when does it appear?
05:14	<boncos>	when i start xend ..
05:15	<mancdaz>	quick google suggests it might be something to do with python ftplib
05:15	<boncos>	mmhh...i'll check it first
05:17	<boncos>	strange....my other machine with same distro, same package, didn't failed
05:22	\|-\|	rusty [~rusty@ppp61-206.lns1.cbr1.internode.on.net] has quit [Quit: Client exiting]
05:34	\|-\|	cehteh [foobar@cehteh.homeunix.org] has joined #xen
05:34	<cehteh>	hi
05:37	<cehteh>	short question: will Xen running on a processor with Vanderpool/Pacifica support uncommon hardware like Firewire and such for uncooperative OS'es like windows?
05:41	\|-\|	athomas [~athomas@host86-136-81-69.range86-136.btcentralplus.com] has joined #xen
05:54	\|-\|	athomas [~athomas@host86-136-81-69.range86-136.btcentralplus.com] has quit [Remote host closed the connection]
06:02	\|-\|	athomas [~athomas@host86-136-81-69.range86-136.btcentralplus.com] has joined #xen
06:15	<mancdaz>	how easy would it be to clone a guest that has a disk file as it's disk - is it simply a case of dd'ing the image file and copying/editing the config file to bring it up?
06:34	<sh0ck>	yes
06:50	<DV>	just make sure it's not mounted :-)
06:58	<sh0ck>	and is very god have hardcoded network settings in domU and then run it few times :>
06:59	<sh0ck>	god/good :> evil
06:59	<mancdaz>	yes so bring one up at a time
06:59	<mancdaz>	or bring up in single user
07:07	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has quit [Quit: mdday]
07:10	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has joined #xen
07:28	<mancdaz>	what tools are available for managing the virtual disk from within the guest environment?
07:28	<mancdaz>	fdisk sees nothing
07:38	<probonic>	Hey, does anyone have a recommendation for a Xen host distro for 3.x that's proven nice and stable?
07:53	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has quit [Quit: mdday]
07:57	\|-\|	boncos [~boncos@202.162.208.126] has quit [Quit: Leaving]
08:05	\|-\|	johnlev [~johnlev@nwkea-socks-1.sun.com] has joined #xen
08:41	\|-\|	Basic_py [~Basic@fortress.tanners.org] has joined #xen
08:52	\|-\|	Basic_py [~Basic@fortress.tanners.org] has quit [Quit: Leaving]
09:00	\|-\|	Shoragan [~shoragan@d072.apm.etc.tu-bs.de] has joined #xen
09:01	<mancdaz>	probonic: fedora core 5 test 3 is pretty good and has a recent snapshot of xen-3
09:01	<mancdaz>	and installs of guests can be done via anaconda
09:03	\|-\|	Shoragan [~shoragan@d072.apm.etc.tu-bs.de] has quit [Read error: Connection reset by peer]
09:03	\|-\|	Shoragan_ [~shoragan@d072.apm.etc.tu-bs.de] has joined #xen
09:11	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has joined #xen
09:15	\|-\|	dansmith [~dan@c-24-21-32-166.hsd1.or.comcast.net] has joined #xen
09:22	\|-\|	rharper [~rharper@pixpat.austin.ibm.com] has joined #xen
09:26	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has quit [Quit: mdday]
09:35	\|-\|	ns [~nivedita@c-67-171-181-157.hsd1.or.comcast.net] has joined #xen
09:46	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has joined #xen
09:50	\|-\|	Basic_py [~Basic@gatekeeper.real-time.com] has joined #xen
09:54	\|-\|	ns [~nivedita@c-67-171-181-157.hsd1.or.comcast.net] has quit [Ping timeout: 480 seconds]
10:02	\|-\|	MarkW [~MarkW@maw48.kings.cam.ac.uk] has joined #xen
10:02	\|-\|	mode/#xen [+o MarkW] by ChanServ
10:03	<@MarkW>	aliguori: yo?
10:03	\|-\|	ns [~nivedita@c-67-171-181-157.hsd1.or.comcast.net] has joined #xen
10:03	<sdague>	MarkW: aliguori thends to be at school on wed
10:04	<@MarkW>	sdague: ah
10:04	<@MarkW>	sdague: maybe you can help me?
10:04	<@MarkW>	sdague: any idea why we ship arch/um/ in the sparse tree now????
10:06	<sdague>	no idea
10:06	<sdague>	there is only one file in there
10:13	<@MarkW>	sdague: indeed. It's a bit of a mystery to me.
10:31	\|-\|	mejlholm [~mejlholm@port79.ds1-abc.adsl.cybercity.dk] has joined #xen
10:43	\|-\|	hollisb [~hollisb@198.4.83.123] has joined #xen
10:43	\|-\|	Bicster [~Bicster@ip195.planetfall.COM] has joined #xen
10:43	\|-\|	pvanhoof [~pvanhoof@mailhost.newtec.be] has quit [Remote host closed the connection]
10:45	\|-\|	dopez [~unknown@muedsl-82-207-247-137.citykom.de] has joined #xen
10:54	\|-\|	gerrit [~gerrit@129.33.1.37] has joined #xen
10:55	\|-\|	hollisb [~hollisb@198.4.83.123] has quit [Quit: leaving]
10:57	\|-\|	stekloff [~stekloff@129.33.1.37] has joined #xen
11:15	\|-\|	hbaum [~hbaum@pixpat.austin.ibm.com] has joined #xen
11:17	\|-\|	johnlev [~johnlev@nwkea-socks-1.sun.com] has quit [Ping timeout: 480 seconds]
11:18	\|-\|	jerone [~jerone@pixpat.austin.ibm.com] has joined #xen
11:23	\|-\|	johnlev [~johnlev@nwkea-socks-2.sun.com] has joined #xen
11:24	\|-\|	xenic [~xenic@85-250-42-132.bb.netvision.net.il] has joined #xen
11:26	\|-\|	Bicster [~Bicster@ip195.planetfall.COM] has quit [Remote host closed the connection]
11:27	\|-\|	jerone [~jerone@pixpat.austin.ibm.com] has quit [Quit: Leaving]
11:28	\|-\|	Bicster [~Bicster@ip195.planetfall.COM] has joined #xen
11:29	<Bicster>	yay, tip works
11:30	<tessier>	tip?
11:30	<tessier>	Is that a uucp dialer thing?
11:30	<Bicster>	tip == cvs head
11:30	<Bicster>	but for mercurial
11:30	<tessier>	oh
11:31	<tessier>	The religious wars over revision control systems amuse me. :0
11:31	<tessier>	Building your own seems to be one of those geek rites of passage or something.
11:31	<Bicster>	I think hg was born of bk licensing issues
11:32	<Bicster>	subversion and cvs are really not similar to bk/hg
11:32	\|-\|	sdague [~sdague@serenity.dague.org] has quit [Ping timeout: 480 seconds]
11:32	<Bicster>	git just does what linus wants
11:32	<Bicster>	did I miss one?
11:32	<movement>	one hundred more like
11:32	<Bicster>	a relevant one? ;)
11:33	<xenic>	does anybody have an experience with vt-x here?
11:33	<Bicster>	yes xenic
11:33	<tessier>	darcs
11:33	<xenic>	Question: i assume you tried to ping from domu with vt-x
11:34	\|-\|	athomas [~athomas@host86-136-81-69.range86-136.btcentralplus.com] has quit [Quit: Leaving]
11:34	\|-\|	ns [~nivedita@c-67-171-181-157.hsd1.or.comcast.net] has quit [Quit: Leaving]
11:34	<xenic>	(or any other network activity such as ssh,telnet ...)
11:35	<Bicster>	networking works fine with vt-x
11:35	<xenic>	Bicster : thnks
11:35	<xenic>	Now what I am getting at is: does anybody know
11:35	<Bicster>	you need to have 'tun' loaded in your kernel, since vt-x networking is somewhat different from xenlinux
11:36	<xenic>	how networking works ? I mean : is it with qemu ?
11:36	<Bicster>	qemu provides a virtual AMD pcinet adapter
11:36	<xenic>	or is is with netfront/netback (which I assume is NOT the case
11:36	<xenic>	Bicster: did you uses some special nic?
11:37	<Bicster>	xenic, huh?
11:37	<xenic>	BIcster: in non-vt-x you work with netfront/netback
11:37	<xenic>	Is it also the case with vt-x?
11:37	<Bicster>	no
11:38	<Bicster>	vt-x uses tun/tap
11:38	<Bicster>	and qemu emulates a pci ethernet adapter
11:38	<xenic>	Bicster: so you MUST have both qemu and tun/tap : is it so?
11:39	<Bicster>	you don't need to install qemu, no
11:39	<Bicster>	everything you need comes with xen
11:39	<Bicster>	xen merely re-uses qemu code to provide emulated hardware
11:39	\|-\|	sdague [~sdague@216.179.125.46] has joined #xen
11:40	<xenic>	I see; did you have any idea on performance ? (I mean if it was a server getting thousands of packets per second)
11:41	<Bicster>	performance is probably not very good, but I haven't tested it
11:41	<Bicster>	it's fine for rdesktop and tasks like that
11:41	<Bicster>	windowsupdate downloads are as fast as my broadband connection allows
11:42	\|-\|	johnlev [~johnlev@nwkea-socks-2.sun.com] has quit [Ping timeout: 480 seconds]
11:42	<xenic>	I see; any idea if in the future intentions is that it will work with qemu emulation or that there are intentions to use the backend/frontend virtual devices?
11:42	\|-\|	johnlev [~johnlev@nwkea-socks-1.sun.com] has joined #xen
11:42	<xenic>	because as I understand in fact dom0 CAN access the NIC
11:43	<Bicster>	I would guess that ultimately someone will write device drivers to run under VT-x that bypass the hardware emulation and run much faster.
11:43	<Bicster>	Those drivers would have to be written for each OS you want to run fast ;)
11:44	<xenic>	I know a little about the implementation of the netback/nerfront ; what does prevent xen from using them also in vt-x?
11:45	<Bicster>	in vt-x the OS is unaware of Xen, so how would it use that?
11:46	<xenic>	Bicster : you are right
11:46	\|-\|	ivan_ [~ikelly@86.41.204.130] has quit [Quit: leaving]
11:49	<Bicster>	xenic, so the point is that you could write device drivers that run in the guest O/S domain, to speed things up by talking to netfront (I suppose).
11:52	<Bicster>	I am getting 1ms ping times between dom0 and a VT-x guest running Windows Server 2003
12:00	\|-\|	ns [~niv@129.33.1.37] has joined #xen
12:04	\|-\|	AlexMBas [~Alexandre@200.138.34.4] has joined #xen
12:33	<hensema>	does anyone have experience with a linux iscsi server and a xen host using a root fs on an iscsi volume?
12:33	<Bicster>	that was a mouthful
12:33	<Bicster>	I hope to try it tomorrow
12:34	<Bicster>	it could require some annoying initrd tricks
12:34	<hensema>	probably next weekend ;-)
12:34	<hensema>	yeah, at http://wiki.xensource.com/xenwiki/UniversityOfCambridgeXenExample I read they use cisco initiator on an initrd
12:34	<hensema>	would be scary though :-)
12:35	<Bicster>	that is oldy moldy
12:35	<hensema>	what kind of storage layer would be preferred with xen 3.0?
12:35	<hensema>	nfs root?
12:36	<Bicster>	nfs is slow
12:36	<Bicster>	I think several people are using iscsi successfully
12:37	<cfreak>	has anyone experience with open-e iscsi (http://www.open-e.com/iscsi_san/storage_area_network/?lang=en&product_code=iSCSI%20ENTERPRISE&show=description) ?
12:38	<hensema>	Bicster: what other options would there be for live migration of domains?
12:38	<teferi>	sounds like initrd tricks are the way to go
12:38	<teferi>	personally
12:39	<Bicster>	cfreak, open-e seems to be selling for $500-$1000 the same iscsi stuff you can download for free
12:39	<cfreak>	but it has a simple interface and support for snapshots and so on
12:40	<Bicster>	hensema, um ... firewire (lame), FC, scsi, AoE [maybe], nbd, gnbd... I think I missed one or two
12:40	<Bicster>	cfreak, yeah, I'd be curious to try it
12:41	<hensema>	AoE?
12:41	<Bicster>	ATA over Etherenet. I don't know if the current versions support multiple initiators per target or not.
12:41	<hensema>	ah
12:41	<Bicster>	I think gnbd may be faster than iscsi, possibly a good choice for linux clients
12:43	<hensema>	would there be an advantage to run gnbd instead of nbd when running a single file server?
12:43	<Bicster>	I ordered the first few bits of my new san box yesterday :)
12:43	<hensema>	hmmm
12:43	<hensema>	gnbd would be scalable to multiple servers of course
12:43	[~]	Bicster is pretty ignorant about gnbd and nbd
12:44	[~]	hensema 2
12:44	<Bicster>	I have some domU's running centos with the clustering stuff so I can play with it
12:44	<Bicster>	it seemed like the path of least resistance to try it out
12:44	<hensema>	lol
12:45	<Bicster>	ultimately I need to make it work with gentoo and xenlinux kernels, but for now centos ("linux for dummies") is fine
12:50	\|-\|	pdx6 [~ballew@screen.sublinear.net] has quit [Quit: switching subnets]
12:55	[~]	Bicster is away: meetings
12:58	[~]	hensema decides he's probably going to setup an NBD server in domain0 and an NBD client in a guest
13:02	\|-\|	aliguori-uni [aliguori@jib.cs.utexas.edu] has joined #xen
13:13	<cfreak>	Bicster you can download a 30day trial at http://www.open-e.com/data_storage_solution/server/service_and_support.php?lang=en&subserv=trial_cd_downloads
13:21	<@MarkW>	Bicster, xenic: Regarding netback / netfront in VT-x
13:22	<@MarkW>	Bicster, xenic: hypercalls can be implemented in VT-x, and intel have written a driver to implement the various xen services. So eventually netfront will probably run in VT-x domains
13:22	<@MarkW>	for maximal performance
13:23	<@MarkW>	There's also a plan to replace the ioemu stuff with something that talks to netback
13:23	<@MarkW>	rather than being a userspace process
13:28	\|-\|	jdmason [~jonmason@pixpat.austin.ibm.com] has quit [Ping timeout: 480 seconds]
13:30	\|-\|	aliguori-uni [aliguori@jib.cs.utexas.edu] has quit [Quit: Leaving]
13:32	\|-\|	jdmason [~jonmason@pixpat.austin.ibm.com] has joined #xen
13:48	<xenic>	MarkW: I am quite ignorant about vt-x and ioemu ; can I verify 2 things which I understand (indirectly) from your answer: 1)in context of vt-x and xen , are ioemu and qemu identical terms ? the same thing (emulating a NIC)? 2) does this ioemu emulation in vt-x currently run in user space?
13:49	<xenic>	Markw: and BTW, is there is also an ioemu/qemu emulation for block devices?
14:03	\|-\|	Bicster [~Bicster@ip195.planetfall.COM] has quit [Ping timeout: 480 seconds]
14:04	\|-\|	Homere [~homere@did75-9-82-229-149-96.fbx.proxad.net] has joined #xen
14:04	<Homere>	hello
14:09	\|-\|	mooli [~muli@87.69.40.180.cable.012.net.il] has joined #xen
14:11	\|-\|	jerone [~jerone@pixpat.austin.ibm.com] has joined #xen
14:13	<Homere>	do you know if it's possible to have a sticky domid for a given domU ? (domid is increased at each start)
14:15	\|-\|	xenic [~xenic@85-250-42-132.bb.netvision.net.il] has quit [Quit: Leaving]
14:20	<teferi>	anyone around to answer a question about Xen networking and traffic shaping?
14:23	\|-\|	Bicster [~Bicster@ip195.planetfall.COM] has joined #xen
14:30	\|-\|	AlexMBas [~Alexandre@200.138.34.4] has quit [Ping timeout: 480 seconds]
14:35	\|-\|	lilo [~lilo@lilo.usercloak.oftc.net] has quit [Quit: leaving]
14:35	\|-\|	lilo [~lilo@lilo.usercloak.oftc.net] has joined #xen
14:40	\|-\|	AlexMBas [~Alexandre@200.103.152.57] has joined #xen
14:41	\|-\|	alexandre [~Alexandre@200.138.131.88] has joined #xen
14:48	\|-\|	AlexMBas [~Alexandre@200.103.152.57] has quit [Ping timeout: 480 seconds]
14:59	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has quit [Quit: mdday]
14:59	\|-\|	yorkedork [~thom@adsl-214-246-113.rmo.bellsouth.net] has joined #xen
15:02	\|-\|	Basic_py [~Basic@gatekeeper.real-time.com] has quit [Quit: Leaving]
15:04	[~]	Bicster is back (gone 00:40:56)
15:05	<Bicster>	thanks for the info MarkW
15:08	<yorkedork>	i'm trying to compile xenified msr support into xen-unstable, and am having problems that seem to indicate my relative lack of familiarity with how xen compiles against/builds with the kernel. when it tries to compile my "msr-xen.c" is bombs on a number of include syntax errors (i.e. uint64_t undefined, dom0_op_t undefined)
15:09	<yorkedork>	i thought i had found all the relevant definitions to include - any pointers as to what i may be misunderstanding about the build process?
15:09	\|-\|	Basic_py [~Basic@gatekeeper.real-time.com] has joined #xen
15:12	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has joined #xen
15:27	<yorkedork>	ah, discovered what i was doing incorrectly re: above embarrassed
15:29	\|-\|	m0dY [m0dY@62.139.160.212] has joined #xen
15:40	\|-\|	Homere [~homere@did75-9-82-229-149-96.fbx.proxad.net] has quit [Quit: Client exiting]
15:43	<jerone>	rharper: are you using notes 7 under ubuntu ?
15:43	\|-\|	pdx6 [~ballew@209.237.242.37] has joined #xen
15:44	\|-\|	mejlholm [~mejlholm@port79.ds1-abc.adsl.cybercity.dk] has quit [Quit: Client exiting]
15:44	<rharper>	jerone: no
15:53	\|-\|	yorkedork [~thom@adsl-214-246-113.rmo.bellsouth.net] has quit [Quit: Leaving]
15:53	<nod>	opinion question - I've got some domU's i need to test but dont happen to have a xen install right now... what's the fastest/easiest install? Xen Live CD and read from hdd? Ubuntu Xen? Fedora?
15:54	<nod>	for performance reasons (limited ram) i dont think livecd will work for what i'm trying to do
15:59	\|-\|	mikegrb Your nick is now startkeylogger
16:00	\|-\|	startkeylogger Your nick is now mikegrb
16:03	\|-\|	m0dY [m0dY@62.139.160.212] has quit [Ping timeout: 480 seconds]
16:05	\|-\|	Basic_py [~Basic@gatekeeper.real-time.com] has quit [Remote host closed the connection]
16:05	\|-\|	Shaun222 [Shaun@tina.ndcservers.net] has joined #xen
16:06	<Shaun222>	aliguori: about having dom0 only use 1 cpu, i though it was a good idea to dedicated a cpu to dom0 to improve IO?
16:09	\|-\|	dopez [~unknown@muedsl-82-207-247-137.citykom.de] has quit [Quit: KVIrc 3.2.0 'Realia']
16:18	\|-\|	Bicster [~Bicster@ip195.planetfall.COM] has quit [Quit: Leaving]
16:31	\|-\|	jerone [~jerone@pixpat.austin.ibm.com] has quit [Quit: Leaving]
16:37	\|-\|	jerone [~jerone@pixpat.austin.ibm.com] has joined #xen
16:47	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has quit [Quit: mdday]
16:51	\|-\|	hbaum [~hbaum@pixpat.austin.ibm.com] has quit [Quit: Client exiting]
17:09	\|-\|	athomas [~athomas@hardpress.demon.co.uk] has joined #xen
17:23	\|-\|	athomas [~athomas@hardpress.demon.co.uk] has quit [Quit: Leaving]
17:28	\|-\|	rharper [~rharper@pixpat.austin.ibm.com] has quit [Quit: Leaving]
17:29	\|-\|	rusty [~rusty@bh02i525f01.au.ibm.com] has joined #xen
17:30	\|-\|	jerone [~jerone@pixpat.austin.ibm.com] has quit [Quit: Leaving]
17:32	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has joined #xen
17:34	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has quit [Quit: ]
17:42	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has joined #xen
17:46	\|-\|	jevere [~jeverever@jffwpr01.jf.intel.com] has joined #xen
17:50	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has quit [Quit: mdday]
17:54	\|-\|	jevere [~jeverever@jffwpr01.jf.intel.com] has quit [Ping timeout: 480 seconds]
18:00	\|-\|	jimix [~jimix@p1.almaden.ibm.com] has joined #xen
18:00	\|-\|	jimix [~jimix@p1.almaden.ibm.com] has quit [Quit: ]
18:03	\|-\|	ivan [~ikelly@86.41.204.130] has joined #xen
18:05	\|-\|	stekloff [~stekloff@129.33.1.37] has quit [Quit: Leaving]
18:14	\|-\|	dansmith [~dan@c-24-21-32-166.hsd1.or.comcast.net] has quit [Quit: Leaving]
18:19	\|-\|	ns [~niv@129.33.1.37] has quit [Quit: Leaving]
18:25	\|-\|	anball [~anball@bi01p1.nc.us.ibm.com] has quit [Quit: Leaving]
18:29	\|-\|	anthony_ [~anthony@cpe-70-116-13-229.austin.res.rr.com] has joined #xen
18:34	\|-\|	Basic_py [~Basic@warden.real-time.com] has joined #xen
18:35	\|-\|	anthony_ [~anthony@cpe-70-116-13-229.austin.res.rr.com] has quit [Quit: Ex-Chat]
18:43	\|-\|	ivan [~ikelly@86.41.204.130] has quit [Quit: leaving]
19:09	<Shaun222>	boy sending a shutdown to 40 guests really puts xend into a lag :)
19:14	\|-\|	gpd [~gpd@70.85.16.173] has joined #xen
19:16	<aliguori>	Shaun222, really?
19:16	<aliguori>	yeah, i guess that kind of makes sense
19:17	<Shaun222>	ya
19:17	<Shaun222>	about 800 xend processes fork too :)
19:18	<Shaun222>	aliguori: has me alittle concerned, i'm wondering if say a bunch of guests start to put a good load on the server, nothing crazy but if it's goign to be a problem with the responsivness of xend
19:19	<Shaun222>	almost wondering if i should put dom0 on the mb controller and seperate disk and run the guests off the 3ware 9550 controller/array...
19:19	<Shaun222>	right now everything is on the 3ware controller/array.
19:20	<Shaun222>	right now with the host idle and all guests idle (running but not really doing anything), xm list takes alittle over 2 seconds to run and spit out output...
19:21	<Shaun222>	40 guests running
19:21	<Shaun222>	actually more like 3 seconds..
19:23	<tonfa>	how many ram do you have ?
19:23	<tonfa>	(just curious)
19:23	<Shaun222>	8GB
19:23	<Shaun222>	dom0 has 512 assigned to it
19:23	<Shaun222>	dom0 isnt using any swap..
19:23	<Shaun222>	only using about 400 of the 51
19:23	<Shaun222>	512*
19:24	<Shaun222>	bahh,
19:24	<Shaun222>	setting maxcpus on dom0 kernel line makes it so that the guests cant use any of the other cpu's.
19:24	<Shaun222>	thats no good...
19:25	<iamlost>	ok
19:25	<iamlost>	how do i tell xen to give more cpu to my domU
19:25	<iamlost>	its taking half an hour to boot (so far)....
19:26	<iamlost>	xm top says its using 0% cpu oddly enough ... or 1 second of cpu time
19:29	<Shaun222>	i must be the only guy attempting to do what the docs recommend for high IO :)
19:35	<aliguori>	iamlost, half an hour to boot? that's not right
19:35	<aliguori>	Shaun222, xend doesn't scale very well :-(
19:36	<aliguori>	as you're experiencing
19:37	<iamlost>	aliguori: no, its painful
19:37	<iamlost>	NAME STATE CPU(sec) CPU(%) MEM(k) MEM(%) MAXMEM(k) MAXMEM(%) VCPUS NETS NETTX(k) NETRX(k)SSID
19:37	<iamlost>	qplex --b--- 2 0.0 65504 14.3 65536 14.3 1 1 5 6 0
19:37	<iamlost>	its only used 2 seconds of cpu time so far
19:38	<iamlost>	so sometimes not right with the scheduler.
19:40	<aliguori>	is this a domU booting?
19:40	<aliguori>	what's the domU?
19:40	<aliguori>	what's the hardware like?
19:41	<iamlost>	yes, its linux 2.6 kernel, xenU
19:41	<iamlost>	hardware is amd athlon XP cpu, like 1.8ghz
19:41	<iamlost>	dom0 runs fine, but it uses 97% of cpu time
19:47	<iamlost>	im guessing somethings wrong with either the cpu paramaters or the scheduler
19:50	<iamlost>	with 'xm sched-sedf 11 0 0 0 1 5' there is some improvement
19:52	<aliguori>	um, what's eating the cpu in dom0/
19:52	<aliguori>	?
19:52	<iamlost>	a badly broken firefox :(
19:52	<aliguori>	hehe
19:52	<aliguori>	oops
19:53	<aliguori>	killall firefox-bin :-)
19:53	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has joined #xen
19:54	\|-\|	gerrit [~gerrit@129.33.1.37] has quit [Ping timeout: 480 seconds]
19:54	<iamlost>	ah yes
19:54	<iamlost>	thats much better
19:57	<aliguori>	unfortunately, eating CPU in dom0 means IO requests won't complete for domUs...
19:57	<aliguori>	eating 100% cpu in a domU wouldn't have as big of an impact
19:59	<tonfa>	is there an easy way to pass information from xen to dom0 ?
19:59	<tonfa>	(like xenstore for domU<->dom0
19:59	<tonfa>	)
19:59	<riel>	eating CPU in dom0 means domU won't get scheduled
19:59	<riel>	thanks to unsafe scheduler settings
20:00	<aliguori>	riel, yeah, I saw your note the other day on xen-devel :-)
20:00	<aliguori>	is there a good solution though?
20:01	<aliguori>	that doesn't hurt IO bound workloads
20:01	<riel>	give dom0 a higher CPU weighting, but no real time priority
20:02	<riel>	if your workload is in domU, dom0 will get the CPU it needs
20:02	<riel>	if your workload is in dom0, domU is screwed regardless ;)
20:02	<aliguori>	:)
20:03	<riel>	regardless meaning:
20:03	<riel>	if dom0 gets all the CPU, domU gets none
20:03	<riel>	and if dom0 doesn't get all the CPU, domU gets worse IO latency
20:03	[~]	aliguori wonders when the new scheduler will go in
20:12	\|-\|	JViz` [Anomaly@adsl-065-013-131-023.sip.int.bellsouth.net] has quit [Read error: Connection reset by peer]
20:12	\|-\|	JViz [Anomaly@adsl-065-013-131-023.sip.int.bellsouth.net] has joined #xen
20:13	<iamlost>	u no, it may just prove easier to put qvm86 directly in the kernel
20:13	<iamlost>	:(
20:13	<aliguori>	?
20:13	<aliguori>	what do you mean?
20:13	<iamlost>	aliguori: i've been talking to kip macy over email....
20:14	<iamlost>	aliguori: i've discovered that i can use breakpoints on functions internal to the kernel
20:14	<iamlost>	but not those in modules.
20:14	<aliguori>	i see
20:14	<aliguori>	so just compile it statically i presume?
20:14	<iamlost>	yup
20:14	<iamlost>	just need to stick it somewhere and do some makefile munging i suppose
20:16	\|-\|	Shoragan_ [~shoragan@d072.apm.etc.tu-bs.de] has quit [Quit: Leaving]
20:17	\|-\|	Teltariat [~tekron@h-68-164-200-156.nycmny83.dynamic.covad.net] has quit [Ping timeout: 480 seconds]
20:38	\|-\|	apathy [~apathy@c220-237-71-241.kelvn1.qld.optusnet.com.au] has joined #xen
20:39	\|-\|	apathy [~apathy@c220-237-71-241.kelvn1.qld.optusnet.com.au] has left #xen []
20:39	\|-\|	apathy [~apathy@c220-237-71-241.kelvn1.qld.optusnet.com.au] has joined #xen
20:58	\|-\|	mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has quit [Quit: mdday]
21:35	\|-\|	psuresh [~psuresh@125.16.129.15] has joined #xen
21:39	\|-\|	psuresh [~psuresh@125.16.129.15] has quit [Quit: ]
22:06	<teferi>	hey, anyone done stuff with QoS/traffic shaping for Xen guests?
22:11	<teferi>	failing that, anyone want to advise me about the best way to allocate CPUs in my Xen setup?
22:14	<xterminus>	teferi: yah, there is nothing special xen specific about traffic shaping
22:14	<teferi>	xterminus: yeah, I know
22:14	<teferi>	xterminus: if I want to do traffic shaping, I'll have to do routing instead of just bridging, right?
22:15	<xterminus>	nod, you limit traffic on your egress interface - not on vifs
22:15	<xterminus>	s/nod/no
22:15	<teferi>	oh? care to elaborate?
22:16	<xterminus>	setup your queues and classes on your egress interface, then use iptables to mark or classify traffic into those classes
22:16	<teferi>	er, could you give me an example?
22:17	<xterminus>	sure - load the mark iptables module, then do something like:
22:18	<xterminus>	iptables -t mangle -A POSTROUTING -s 192.168.1.4 -j MARK --set-mark 4
22:18	<xterminus>	tc filter add dev eth0 protocol ip handle 4 fw flowid 1:40
22:19	<xterminus>	that grabs packets marked with mark "4", and places them in the 1:40 classid
22:19	<xterminus>	you could do that in one line too
22:19	<teferi>	where 192.168.1.4 is one of the domUs?
22:20	<xterminus>	iptables -t mangle -A POSTROUTING -p tcp -s 192.168.1.4 -j CLASSIFY --set-class 1:40
22:20	<xterminus>	yah
22:20	<teferi>	awesome
22:20	<teferi>	which qdisc is that using?
22:20	<xterminus>	doesnt matter
22:20	<xterminus>	i use htb and sfq
22:21	<teferi>	but it's gotta be classful, with one class per VM, right?
22:21	<xterminus>	hdb to define classes, and sfq inside the classes to round robin packets with the same priority
22:21	<xterminus>	you could do that sure
22:22	<xterminus>	you could also do bit counting with iptables and allow all domu's to go crazy for say 1 gig a day upstream, then throttle them after that
22:23	<xterminus>	encourages them to behave :)
22:23	<teferi>	heh
22:23	<teferi>	well, our colo plan gives us 250G/mo
22:23	<xterminus>	rate limit em to 5kbs up after 1 gig or something
22:23	<teferi>	and it's only the 5 of us so far
22:24	<xterminus>	ahh, i'd setup 5 classes then, give each 20% rate, and 100% ceiling
22:24	<teferi>	I haven't really done much with tc yet, unfortunately
22:24	<xterminus>	teferi, its fun :)
22:24	<teferi>	so I'm really not sure how to go about doing most of this
22:25	<xterminus>	teferi, read the advanced routing howto for linux
22:25	<xterminus>	the htb homepage is essential too
22:25	<teferi>	I'm just kinda amazed that I can do this with just bridging
22:25	<teferi>	the advanced routing howto is very specific about how this stuff only happens at the ip level
22:26	<xterminus>	bridging has nothing to do with it really
22:26	<xterminus>	i use routing only here with xen
22:26	<teferi>	routing just seems like it'll be more efort
22:26	<xterminus>	bridging + ipv6 + xen is a total no-go
22:27	<xterminus>	its less
22:27	<xterminus>	bridging makes it all that much more complex i think
22:27	<teferi>	well, with bridging, I have networking working out of the box with practically no configuration
22:28	<xterminus>	yah, i think the xen folks prefer bridging because it makes moving domu's around (live) easier
22:30	<xterminus>	i do all of my network setup with one script for example with routing
22:30	<xterminus>	http://pastebin.com/568055
22:30	<teferi>	could you talk about your setup?
22:31	<teferi>	ah
22:34	<xterminus>	if you rip the ipv6 hackery out - its really simple
22:35	\|-\|	apathy [~apathy@c220-237-71-241.kelvn1.qld.optusnet.com.au] has quit [Quit: leaving]
22:36	[~]	teferi examines
22:40	<xterminus>	im gonna eventually put all of my configs and changes online - but in need to do some subversion foolery first, still have not figured out a good way of sharing custom configs in /etc via svn easily
22:48	<teferi>	I've been using SVN for project management, but not configuration management
22:49	<xterminus>	ive had all of my homedirs in svn (and cvs before svn) for years
22:50	<xterminus>	i try and keep as much as possible in subversion repos - makes an awesome backup system among other things
22:54	<xterminus>	also makes it easy to backup all of your important info in one shot (by burning the repo to cd or tarring it up onto a tape)
22:57	<teferi>	yeah
22:57	<teferi>	I've been using rsync to back up my work repository
22:59	\|-\|	VS_ChanLog [~stats@ns.theshore.net] has left #xen [Rotating Logs]
22:59	\|-\|	VS_ChanLog [~stats@ns.theshore.net] has joined #xen
23:01	<rusty>	aliguori: EM64T Linux boxes return x86_64 from uname -m, right?
23:02	<teferi>	rusty: I'd imagine so
23:03	<teferi>	speaking of CPUs, if I have a dual-core machine, what's the optimal cpu-pinning configuration
23:03	<teferi>	?
23:32	\|-\|	hollisb [~hollisb@adsl-67-115-102-122.dsl.sntc01.pacbell.net] has joined #xen
23:38	\|-\|	gpd [~gpd@70.85.16.173] has left #xen []
---	Log	closed Thu Feb 23 00:00:14 2006