| --- | Log | opened Tue Jun 20 00:00:46 2006 |
| 01:13 | |-| | visik7 [~visi@host104-47.pool877.interbusiness.it] has quit [Remote host closed the connection] |
| 01:16 | |-| | gerrit [~gerrit@c-67-160-146-170.hsd1.or.comcast.net] has quit [Ping timeout: 480 seconds] |
| 01:22 | |-| | SNy [63fe5ce50b@bmx-chemnitz.de] has quit [Remote host closed the connection] |
| 01:22 | |-| | SNy [c37e4bcf2a@bmx-chemnitz.de] has joined #xen |
| 01:27 | |-| | mastermind [~mastermin@mastermind.kaltenbrunner.cc] has quit [Quit: Client exiting] |
| 01:35 | |-| | gerrit [~gerrit@c-67-160-146-170.hsd1.or.comcast.net] has joined #xen |
| 01:37 | |-| | koki [~klaus@baikonur.stro.at] has quit [Ping timeout: 480 seconds] |
| 01:41 | |-| | horms [~horms@vagw.valinux.co.jp] has quit [Quit: Leaving] |
| 01:42 | |-| | koki [~klaus@baikonur.stro.at] has joined #xen |
| 01:42 | |-| | lazyb0y_ changed nick to lazyb0y |
| 01:49 | |-| | womble [~mpalmer@106.135.233.220.exetel.com.au] has quit [Quit: What's behind the round window...] |
| 01:57 | |-| | JViz [Anomaly@adsl-065-013-131-023.sip.int.bellsouth.net] has quit [Read error: Connection reset by peer] |
| 02:25 | |-| | Tv [~tv@GMMDXXVII.dsl.saunalahti.fi] has quit [Quit: foo] |
| 02:28 | |-| | JViz [Anomaly@adsl-065-013-131-023.sip.int.bellsouth.net] has joined #xen |
| 02:41 | |-| | parabelboi [~parabelbo@blueice1n1.de.ibm.com] has joined #xen |
| 02:43 | |-| | Shoragan_ [~shoragan@d072.apm.etc.tu-bs.de] has joined #xen |
| 03:19 | |-| | visik7 [~visi@155.185.28.21] has joined #xen |
| 03:28 | |-| | Tv [~tv@i1.inoi.fi] has joined #xen |
| 03:31 | |-| | athomas [~athomas@host86-136-84-147.range86-136.btcentralplus.com] has joined #xen |
| 03:33 | |-| | tomaso [tomas@twin.jikos.cz] has quit [Quit: Changing server] |
| 03:40 | |-| | horms [~horms@vagw.valinux.co.jp] has joined #xen |
| 03:58 | |-| | JViz [Anomaly@adsl-065-013-131-023.sip.int.bellsouth.net] has quit [Read error: Connection reset by peer] |
| 04:27 | |-| | visik7 [~visi@155.185.28.21] has quit [Remote host closed the connection] |
| 04:32 | |-| | jasonmc [jason@spoon.netsoc.tcd.ie] has joined #xen |
| 05:04 | |-| | cableman [~cableman@3e6b5b45.rev.stofanet.dk] has joined #xen |
| 05:07 | |-| | mejlholm [~mejlholm@port79.ds1-abc.adsl.cybercity.dk] has joined #xen |
| 05:19 | |-| | horms [~horms@vagw.valinux.co.jp] has quit [Quit: Leaving] |
| 05:20 | |-| | Shaun [~ndci@68.5.63.223] has quit [Ping timeout: 480 seconds] |
| 05:35 | |-| | visik7 [~visi@155.185.28.21] has joined #xen |
| 06:05 | <jasonmc> | does anybody know how to go about patching a kernel SRPM they have with xen? |
| 06:06 | <Guy-> | jasonmc: not actually "know", as such, but I guess you unpack the srpm, extract the source, and then patch it - at least that what I'd do |
| 06:07 | |-| | harry_ [~harry@blueice2n1.uk.ibm.com] has joined #xen |
| 06:09 | <harry_> | Hi all. I think bugzilla 625 is caused by compiling the scsi disk driver into the domU kernel. xen_blk can't then get the SCSI major number 8. This is a usability issue because people create broken domU kernels and then raise bugs. What's the best fix? Changing the Kconfig files to prohibit built in scsi disk when blkfront is enabled? Is there a better option? |
| 06:10 | <jasonmc> | yea, that's what i'm doing. just asking if anyone had any experience with it. cheers |
| 06:10 | <harry_> | jasonmc: what exactly are you doing? |
| 06:11 | <jasonmc> | building an rpm for RHEL4. |
| 06:11 | <harry_> | maybe I should go back and read the logs---i just joined |
| 06:11 | <jasonmc> | (x86_64) |
| 06:14 | <jasonmc> | oh, sorry. i asked does anybody know how to go about patching a kernel SRPM they |
| 06:14 | <jasonmc> | have with xen? |
| 06:16 | <harry_> | OK, i just skimmed the logs. |
| 06:18 | <harry_> | I've never tried patching a kernel tree with the xen patches so I don't know how hard it will be. |
| 06:21 | <jasonmc> | hmmm ok. |
| 06:21 | <Guy-> | I did it yesterday, no issues |
| 06:22 | <Guy-> | (although I patched a vanilla kernel, not one from redhat) |
| 06:23 | <jasonmc> | is it a matter of patching everything from the linux-2.6-xen-sparse directory. or are there more required actions? |
| 06:24 | <harry_> | You need the sparse stuff and some of the stuff in the patches dir depending on what of that is already in your kernel. |
| 06:24 | <harry_> | But all the code has to be consistent to work correctly. |
| 06:25 | <harry_> | I would have thought it would be less risky to use a standard xen kernel that has had a lot of testing and then fix any userspace issues that arise from using a newer kernel. |
| 06:26 | <jasonmc> | point taken |
| 06:26 | <harry_> | I guess it depends how much userspace pain you get. |
| 06:26 | <jasonmc> | should xen only be patched on the 2.6.16 kernel? |
| 06:26 | <Guy-> | what's the preferred 3.x version to use now? xen-3.0.2+hg9697 is published on alioth.debian.org, any known trouble with that? or should I just pull the latest from mercurial? |
| 06:26 | <jasonmc> | yea, lvm and udev break i think |
| 06:29 | |-| | horms [~horms@p7209-ipbf22marunouchi.tokyo.ocn.ne.jp] has joined #xen |
| 06:29 | <am88b> | does anyone know how far is xen from being merged into vanilla kernel? |
| 06:31 | <Guy-> | am88b: quite far, afaik - many months to go, surely |
| 06:31 | <Guy-> | (but I'm not authoritative) |
| 06:32 | <horms> | am88b: a very long way |
| 06:33 | <horms> | i believe that cdub_ is cordinating the current effort. which is to try and get (UP, i386, domU, shadow page table) in as a start |
| 06:33 | <murb> | Guy-: according to the FOSDEM talk about a month ago :-) |
| 06:33 | <murb> | s/Guy-/am88b/ |
| 06:35 | <am88b> | ok, but does anyone know when new stable version will come out? I was going to use Xen in production enviroment but if new stable is coming soon I might want to wait... |
| 06:35 | <horms> | 3.0.3 should be out within the next month |
| 06:36 | <am88b> | cool |
| 06:37 | <am88b> | then maybe using xen-testing might be good idea.. |
| 06:37 | <horms> | the plan is to get it done by ols, which is about this time next month |
| 06:37 | <horms> | i guess if that is missed it will slip a few weeks |
| 06:37 | <am88b> | what is ols? |
| 06:38 | <horms> | ottawa linux symposium |
| 06:38 | <am88b> | ok, thanks |
| 06:39 | <Guy-> | how usable can I normally expect -testing to be? is it totally bleeding edge that may not even compile, or somewhat consolidated and probably stable? |
| 06:39 | <horms> | the bleeding edge stuff should be in -unstable |
| 06:39 | <horms> | so i would imagine that -testing is gernerally ok |
| 06:40 | <Guy-> | OK, thanks |
| 06:40 | <horms> | i'd actually be more worried about it being out of date |
| 06:40 | <horms> | but you could verify that with the commit logs |
| 06:40 | <Guy-> | out of date as compared to what? |
| 06:40 | <horms> | also, keep in mind i've only ever used -unstable |
| 06:40 | <horms> | compared to -unstable |
| 06:40 | <am88b> | but compared to -stable ? |
| 06:41 | <horms> | no idea :( |
| 06:41 | <am88b> | I think using -unstable in production wouldn't be good idea |
| 06:41 | <Guy-> | well, debian unstable in production is perfectly fine |
| 06:41 | <horms> | am88b: agreed |
| 06:41 | <Guy-> | the question is how "unstable" -unstable actually is |
| 06:41 | <am88b> | Guy-: if you're lucky ;) |
| 06:42 | <Guy-> | am88b: or determined :) |
| 06:42 | <horms> | from a build point of view, reasonably |
| 06:42 | <am88b> | hehe |
| 06:42 | <horms> | from a trashing your system point of view, not sure, but i wouldn't trust it myself |
| 06:42 | <harry_> | debian unstable can't be compared to xen unstable. |
| 06:43 | <horms> | harry_: yes, i agree with that |
| 06:43 | <Guy-> | certainly not |
| 06:43 | <Guy-> | I was just trying to say that just because something is called "unstable" it needn't be unusable |
| 07:02 | |-| | bernarde [~bernarde@201.44.214.48] has joined #xen |
| 07:11 | |-| | SLot [~SLot___@gustavo.gruposim.com.br] has joined #xen |
| 07:12 | |-| | Shaun [~ndci@ip68-5-63-223.oc.oc.cox.net] has joined #xen |
| 07:21 | |-| | mr_flea [~mr_flea@cpe-70-113-197-207.austin.res.rr.com] has quit [Quit: leaving] |
| 08:01 | |-| | JViz [Anomaly@adsl-065-013-131-023.sip.int.bellsouth.net] has joined #xen |
| 08:14 | |-| | plars [~plars@pixpat.austin.ibm.com] has joined #xen |
| 08:16 | |-| | surriel changed nick to riel |
| 08:39 | |-| | visik7 [~visi@155.185.28.21] has quit [Quit: ] |
| 08:57 | <Guy-> | I'm confused about the way xen manages memory |
| 08:57 | <Guy-> | can I give dom0 all my physical RAM and still run domU? |
| 08:58 | <Guy-> | i.e. dom0_mem=size_of_physical_memory |
| 08:58 | <harry_> | yes i think so |
| 08:58 | <Guy-> | what are the disadvantages of this? |
| 08:58 | <Guy-> | (there must be some, otherwise it could be the default) |
| 09:00 | <harry_> | isn't it the default? |
| 09:00 | <harry_> | I don't have a dom0_mem setting when I do testing |
| 09:00 | <Guy-> | The maximum is 40MB less than you have in the system (the Xen hypervisor requires 40MB for itself) |
| 09:00 | <Guy-> | http://wiki.xensource.com/xenwiki/GettingStarted?highlight=%28dom0_mem%29 |
| 09:00 | <murb> | i think there were bugs before with the balloon driver |
| 09:01 | <murb> | i certainly use sill dom0_mem= |
| 09:01 | <murb> | because of this, but it is probably just voodoo |
| 09:01 | <Guy-> | according to the docs, the default is |
| 09:01 | <Guy-> | 16000kBytes |
| 09:02 | <Guy-> | ah, this is for 2.0.5 |
| 09:03 | <Guy-> | I can't seem to find what the default is supposed to be in 3.x |
| 09:03 | <harry_> | I don't use a dom0_mem setting and dom0 seems to get most of my 8GB |
| 09:03 | <Guy-> | but: why would I want to limit the amount of memory dom0 sees? don't all my domains share physical memory to the extent they need it? |
| 09:04 | <Guy-> | well... maybe not |
| 09:04 | <Guy-> | because xen itself doesn't do virtual memory management (no paging or swap) |
| 09:04 | <harry_> | I guess if you have lots of ram then giving dom0 too much would cause it to allocate metadata for memory it was never going to use and therefore waste space. |
| 09:05 | <Guy-> | what I'm really concerned about is where the memory I want to give my domU's will come from |
| 09:05 | <harry_> | The balloon driver in dom0 |
| 09:05 | <Guy-> | I'll try to read up on that |
| 09:05 | <Guy-> | no reference to it in the user guide :/ |
| 09:06 | <harry_> | The balloon driver allocates pages in the kernel. Since the pages are allocated to the balloon driver the balloon driver knows they are free and can be used by a domU. |
| 09:06 | <harry_> | Hence the name balloon driver---it's like a balloon of domU space expanding in dom0 memory. |
| 09:07 | <Guy-> | I'm not sure I understand |
| 09:07 | <Guy-> | are you saying that the balloon driver is part of the dom0 kernel, and that it keeps track of what part of memory isn't in use by dom0? |
| 09:07 | <harry_> | Yes it's part of the dom0 kernel. |
| 09:08 | <harry_> | When it's idle, it hasn't allocated any RAM. |
| 09:08 | <Guy-> | OK, so if my dom0 doesn't use all the physical RAM, then the physical RAM can be used by domU, right? |
| 09:08 | <harry_> | Yes. |
| 09:08 | <Guy-> | what if my dom0 _does_ use all physical RAM, e.g. for cache? |
| 09:09 | <Guy-> | domU's can't be swapped out onto the swap of dom0, can they? |
| 09:09 | <harry_> | Well the balloon driver will call kmalloc which will flush the cache. |
| 09:09 | <harry_> | No, when the memory is allocated to the balloon driver it is pinned. |
| 09:09 | <Guy-> | so, basically, cache and buffers count as 'free' memory as far as xen is concerned |
| 09:09 | <harry_> | Yes. |
| 09:10 | <Guy-> | OK, will the balloon driver cause dom0 processes to be swapped out if a domU needs memory? |
| 09:10 | <harry_> | Yes. I think so. |
| 09:10 | <Guy-> | cool |
| 09:10 | <harry_> | I think it's just like a normal kernel memory allocation. |
| 09:11 | <harry_> | Which would push out any cache/buffers/user-space allocation if necessary. |
| 09:11 | <Guy-> | what happens if I have, say, 1GB of RAM and try to run two domU's with mem=768M each? |
| 09:11 | |-| | horms [~horms@p7209-ipbf22marunouchi.tokyo.ocn.ne.jp] has quit [Quit: Leaving] |
| 09:12 | <harry_> | I think if you balloon one down so it is only using <150MB then the other one should be able to start. |
| 09:12 | <harry_> | But I think there are some bugs in this area. |
| 09:12 | <Guy-> | so memory is not completely virtualized in the sense that I can't pretend I have more RAM than I actually have |
| 09:13 | <Guy-> | (the rest would be swap) |
| 09:13 | <harry_> | Like a suspended ballooned down domain apparently won't restart if there isn't the full memory available. |
| 09:13 | <harry_> | No, there is no swap for domain memory. |
| 09:13 | <Guy-> | I see |
| 09:13 | <Guy-> | OK, thanks a lot |
| 09:14 | <Guy-> | (this is slightly worse than with UML, unfortunately; UMLs can be swapped out, no problem) |
| 09:14 | <harry_> | Xen is different from UML. UML can't run non-linux guests. |
| 09:14 | <Guy-> | I know |
| 09:15 | <Guy-> | and there's the performance benefit, of course |
| 09:15 | <harry_> | I don't think UML can support different versions of the linux kernel concurrently either, |
| 09:15 | <Guy-> | it can |
| 09:15 | <harry_> | Really? I didn't know that. |
| 09:15 | <Guy-> | and, theoretically anyway, you could write a user-mode-freebsd to run on linux, but it hasn't happened yet |
| 09:16 | <Guy-> | sure, I run 2.4 UMLs on 2.6 hosts |
| 09:16 | <Guy-> | and you can run several different versions of UML on the same host |
| 09:17 | <harry_> | What's the performance like? |
| 09:17 | <Guy-> | it's OK for CPU-intensive workloads |
| 09:17 | <Guy-> | less so for I/O intensive ones |
| 09:18 | <Guy-> | being able to swap out an entire UML instance is actually a large advantage in some cases |
| 09:18 | <harry_> | What cases? |
| 09:18 | <harry_> | It would be good to know when that is useful. |
| 09:18 | <Guy-> | for example, at one of our universities, they have a box with 4GB RAM and they run a multitude of low-volume UMLs on it |
| 09:19 | <Guy-> | most only contain webservers |
| 09:19 | <Guy-> | many of them don't get hits for several hours |
| 09:19 | <Guy-> | these spend much of their time being swapped out |
| 09:19 | <harry_> | That's interesting. |
| 09:19 | <harry_> | Any other examples? |
| 09:19 | <Guy-> | which leaves more RAM for the others (they don't actually have as much physical RAM as the total amount of memory given to the UMLs) |
| 09:20 | <Guy-> | well, obviously software suspend |
| 09:20 | <Guy-> | you can suspend an UML host and the UML guest will suspend with it |
| 09:21 | <harry_> | Does that have anything do do with memory overcommit or is it just another advantage you see over Xen? |
| 09:21 | <Guy-> | I don't see how it would have anything to do with overcommitting memory |
| 09:21 | <harry_> | OK. Just checking. |
| 09:21 | <Guy-> | it's just the convenience of being able to swap out a UML instance, like any other user-space process |
| 09:21 | <Guy-> | it'd be nice if xen could do that |
| 09:22 | <Guy-> | but it'd need its own VMM |
| 09:22 | <Guy-> | (or use dom0's, but I don't know how feasible that is) |
| 09:27 | <harry_> | It's possible for the hypervisor to have a vmm. Not sure how easy to integrate into Xen. I had always assumed that the performance would suck so badly that there wouldn't be any use cases so hearing that it's useful is very interesting. |
| 09:27 | <harry_> | thanks. |
| 09:27 | <Guy-> | my pleasure :) |
| 09:31 | |-| | rharper [~rharper@pixpat.austin.ibm.com] has joined #xen |
| 09:37 | |-| | jerone [~jerone@pixpat.austin.ibm.com] has joined #xen |
| 09:46 | |-| | aliguori [~anthony@cpe-70-116-9-243.austin.res.rr.com] has quit [Remote host closed the connection] |
| 09:57 | |-| | aliguori-uni [aliguori@windlass.cs.utexas.edu] has joined #xen |
| 10:08 | |-| | jimix [~jimix@yktgi01e0-s4.watson.ibm.com] has joined #xen |
| 10:08 | <Guy-> | harry_: oh, another thing: you can run a 32bit UML on a 64bit host, which is sometimes also nice, and something xen doesn't (yet?) do |
| 10:09 | <aliguori-uni> | uml is fundamentally different than xen |
| 10:09 | <Guy-> | I know |
| 10:09 | <Guy-> | I was just comparing them from the end-user's perspective |
| 10:09 | <aliguori-uni> | just get VT hardware... |
| 10:09 | <Guy-> | and I'm well aware of the numerous advantages of xen :) |
| 10:09 | <aliguori-uni> | in 3 years, most people will have it |
| 10:10 | <Guy-> | don't get me wrong, I wasn't knocking xen |
| 10:10 | <aliguori-uni> | i understand |
| 10:10 | <harry_> | aliguori-uni: my home machine is svm enabled---i got the svm flag with the latest xen kernel. |
| 10:10 | <harry_> | :-) |
| 10:10 | <aliguori-uni> | harry_, sweet, have you gotten it working yet? |
| 10:11 | <aliguori-uni> | i'm going to build an svm system in the near future it hink |
| 10:11 | <Guy-> | what's svm? |
| 10:11 | <aliguori-uni> | amd's virtualization extensions |
| 10:11 | <harry_> | I'm not going to be running xen on it. |
| 10:11 | <Guy-> | pacifica? |
| 10:11 | <harry_> | yes |
| 10:11 | |-| | bernarde [~bernarde@201.44.214.48] has quit [Remote host closed the connection] |
| 10:11 | <aliguori-uni> | Guy-, pacifica was the codename, svm is the official name i believe |
| 10:11 | <Guy-> | I see |
| 10:11 | <harry_> | I want to play around with the SVM primitives. |
| 10:13 | <aliguori-uni> | skinit? |
| 10:13 | <riel> | speaking of VT hardware ... it left the local UPS depot at 6:11 AM |
| 10:13 | [~] | riel can't wait |
| 10:13 | <riel> | but it looks like I'll have to |
| 10:13 | <aliguori-uni> | riel, dying to run windows under xen? :-) |
| 10:14 | <riel> | no |
| 10:14 | <riel> | but full-virt RHEL4 |
| 10:30 | |-| | Basic_py [~Basic@warden.real-time.com] has quit [Quit: Leaving] |
| 10:32 | <jerone> | aliguori: svm amd cpus are going for as low as $105 on newegg ...just get a board and ram and your ready |
| 10:34 | <riel> | ummmm, $105? |
| 10:34 | <riel> | which cpus are those? |
| 10:34 | [~] | riel wasn't aware svm capable cpus were selling already |
| 10:35 | <riel> | AMD.com certainly didn't mention it when I looked last week |
| 10:35 | <@MarkW> | Wow there are people here! |
| 10:35 | <@MarkW> | Hello people! |
| 10:36 | <jerone> | riel: yeah if all AM2 cpus are svn enable (not the semptrons) then it's pretty low ...hold on let me check $105 may have been too low |
| 10:37 | |-| | stekloff [~stekloff@bi01p1.co.us.ibm.com] has joined #xen |
| 10:37 | <jerone> | reil: actuall on $94 |
| 10:37 | <jerone> | http://www.newegg.com/Product/Product.asp?Item=N82E16819103639 |
| 10:38 | <jerone> | http://www.newegg.com/Product/ProductList.asp?Order=PRICE&Page=1&N=2010340343+1051720996&Submit=ENE&Nty=1&SubCategory=343 |
| 10:38 | <riel> | oh well, I wanted a dual core cpu anyway |
| 10:38 | <Guy-> | what options must be enabled in the domU kernel for it to be able to boot off "hda1"? I get Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(3,1) |
| 10:38 | <Guy-> | I have IDE disk support compiled in |
| 10:38 | <Guy-> | and xen block frontend too |
| 10:38 | <jerone> | reil: haha yeah ... they still have an outragous premium for the X2 cpus |
| 10:39 | <riel> | Intel 930 is a fairly cheap CPU, with 2x 2MB L2 cache |
| 10:39 | <harry_> | Guy: xen block front |
| 10:39 | <Guy-> | I have that |
| 10:39 | <@MarkW> | Guy-: what does your config file say? |
| 10:39 | <bored2sleep> | the X2 for $297 doesn't seem that outrageous. |
| 10:40 | |-| | Tv [~tv@i1.inoi.fi] has quit [Quit: foo] |
| 10:40 | <jerone> | riel: I'm still holding on to the amd fanboy in me |
| 10:40 | <Guy-> | MarkW: for the kernel? CONFIG_XEN_BLKDEV_FRONTEND=y |
| 10:40 | <@MarkW> | Guy-: No, for the domain |
| 10:41 | <Guy-> | disk = [ 'file:/store/elan-uml/root.img,hda1,w', 'file:/store/elan-uml/swap.img,hda2,w' ] |
| 10:41 | <Guy-> | root = "/dev/hda1 ro" |
| 10:41 | <jerone> | bored2sleep: actually your kind of right .. I bought one of the first athlon64 3200+ that came out for ~$500 .. but that's what you get with first gen...pay more...get more bugs |
| 10:41 | <@MarkW> | Guy-: Hrrmmmm. Is this a fresh install of Xen? |
| 10:41 | <Guy-> | yes |
| 10:42 | <Guy-> | the xenu kernel doesn't have PCI support enabled |
| 10:42 | <Guy-> | can that be a problem? |
| 10:42 | <aliguori-uni> | Guy-, if you have IDE compiled in the domU kernel that's your problem |
| 10:42 | |-| | hollisb [~hollisb@pixpat.austin.ibm.com] has joined #xen |
| 10:42 | <riel> | jerone: my Dell 9150 is supposed to be delivered today |
| 10:42 | <aliguori-uni> | xenu kernel shouldn't have any hardware enabled |
| 10:42 | <riel> | jerone: Intel 930 CPU, otherwise fairly standard |
| 10:43 | <Guy-> | aliguori-uni: it is? yesterday I played with xen on a different computer and the xenu kernel had IDE support and it appeared to work... |
| 10:43 | <Guy-> | aliguori-uni: but I'll try without it |
| 10:43 | <jerone> | riel: you mean you didn't get a dual SLI nvidia card with that...how can you possibly use that :-) |
| 10:44 | <riel> | apparently the nvidia cards have a loud fan |
| 10:44 | <riel> | that needs the binary-only driver to make the fan go slower and quieter |
| 10:45 | <Guy-> | strangely, with ATI it's sort of the other way round :) the Linux driver makes the fan a lot louder than it is under windows |
| 10:45 | <@MarkW> | aliguori-uni: Really? I knew we occasionally had trouble with SCSI but not IDE. |
| 10:45 | <@MarkW> | aliguori-uni: How come the xen0 kernel works in that case??? |
| 10:47 | <Guy-> | aliguori-uni: removing IDE support didn't help |
| 10:50 | <aliguori-uni> | Guy-, that error only occurs for a few reasons: 1) you have a bad config 2) udev/hotplug is busted 3) blkfront can't register the IDE major number (you'll see an error message) 4) blkfront isn't supported in the kernel |
| 10:50 | <aliguori-uni> | take your pick :-) |
| 10:50 | <murb> | what happend to the xbd devices or whatever we are supposed to be using? |
| 10:50 | <aliguori-uni> | MarkW: the IDE driver will register the IDE major numbers on startup IIRC |
| 10:51 | <aliguori-uni> | MarkW, xen0 works b/c the blkfront only attempts to register the IDE major on demand again IIRC |
| 10:51 | <aliguori-uni> | murb, people aren't being encouraged to use them as they ought to be |
| 10:51 | <murb> | aliguori-uni: does blkfront now have its own major/mono |
| 10:51 | <Guy-> | aliguori-uni: 'bad config' as in ...? |
| 10:51 | <aliguori-uni> | murb, yes |
| 10:51 | <@MarkW> | aliguori-uni: but you can run a xen0 kernel in a domU and it works... |
| 10:51 | <murb> | aliguori-uni: i spent about 5 minutes trying to get it to work. |
| 10:51 | <aliguori-uni> | Guy-, as in your configuration isn't good :-) |
| 10:51 | <@MarkW> | aliguori-uni: or am I missing the point of the problem? |
| 10:51 | <aliguori-uni> | MarkW, this all strongly depends on what your config is for xen0 |
| 10:52 | <Guy-> | aliguori-uni: well, I'm trying to find out what part of it isn't good :) I guess we can rule out the others |
| 10:52 | <aliguori-uni> | MarkW, it depends on what is modules and what isn't |
| 10:52 | <aliguori-uni> | Guy-, post your config and the kernel startup messages in pastebin.com |
| 10:52 | <aliguori-uni> | and xend.log/xend-debug.log |
| 10:53 | <aliguori-uni> | and it will be pretty easy to tell what's going on |
| 10:53 | <aliguori-uni> | provided you can do that in the next 7 minutes before i got get lunch :-) |
| 10:53 | <Guy-> | I guess so :) |
| 10:54 | <aliguori-uni> | it's almost certainly a bad config or bad kernel config. it's rare we get hotplug problems anymore |
| 10:54 | <murb> | anyone got an example config using then ative blockdev? |
| 10:54 | <harry_> | Guy- why is your kernel trying tou mount hdd2 when your config is only giving hda1 and hda2? I think you have passed the wrong root= parameter. |
| 10:55 | <harry_> | unknown-block(3,1) is hdd2 right? |
| 10:55 | <dansmith> | 3,1 is hda1 |
| 10:55 | <harry_> | that's not a grub number then? |
| 10:56 | <dansmith> | no |
| 10:56 | <aliguori-uni> | harry_, major/minor device number |
| 10:56 | <dansmith> | that's a maj/min |
| 10:56 | <harry_> | oh well ignore the above then |
| 10:57 | |-| | bernarde [~bernarde@201.44.214.48] has joined #xen |
| 10:57 | <harry_> | if the filesystem wasn't compiled in would you get that error? I had something similar when I forgot cramfs for my initrd. |
| 10:57 | <Guy-> | aliguori-uni: http://pastebin.ca/67773 (xend.log) |
| 10:57 | <aliguori-uni> | harry_, no, you'd get invalid superblock |
| 10:58 | <harry_> | i don't remember getting invalid superblock. |
| 10:58 | <Guy-> | aliguori-uni: http://pastebin.ca/67774 (kernel startup) |
| 10:58 | <harry_> | I think i would have found that a lot easier to debug. |
| 11:01 | <Guy-> | aliguori-uni: http://pastebin.ca/67777 (kernel config) |
| 11:01 | <aliguori-uni> | # |
| 11:01 | <aliguori-uni> | UDF-fs: No VRS found |
| 11:01 | <aliguori-uni> | # |
| 11:01 | <aliguori-uni> | Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(3,1) |
| 11:02 | <aliguori-uni> | whatever your root device is, the kernel you configured doesn't look it |
| 11:02 | <Guy-> | it's supposed to be hda1, which is an image file |
| 11:02 | <aliguori-uni> | what is it an image of? |
| 11:02 | <aliguori-uni> | what filesystem type? |
| 11:02 | <Guy-> | of an xfs filesystem |
| 11:03 | <aliguori-uni> | you have xfs compiled as a module |
| 11:03 | <aliguori-uni> | but no initrd |
| 11:03 | <aliguori-uni> | so that's not gonna work |
| 11:03 | <Guy-> | indeed I have :) |
| 11:03 | <Guy-> | silly me |
| 11:03 | <Guy-> | I was confused by the unknown-block message |
| 11:03 | <aliguori-uni> | like i said, it's always one of those four things :-) |
| 11:03 | <aliguori-uni> | it's an exceedingly common error unfortunately :-) |
| 11:03 | <Guy-> | I thought it meant it didn't even get as far as seeing what fs was there |
| 11:04 | <harry_> | Yeah. I think i got unknown block when i forgot cramfs. It took allmorning for me to find the problem. |
| 11:04 | <aliguori-uni> | time for me to go, cya |
| 11:04 | |-| | aliguori-uni [aliguori@windlass.cs.utexas.edu] has quit [Quit: Leaving] |
| 11:05 | <Guy-> | thanks, bye |
| 11:08 | |-| | parabelboi [~parabelbo@blueice1n1.de.ibm.com] has quit [Read error: Connection reset by peer] |
| 11:14 | |-| | Tv [~tv@GMMDXXVII.dsl.saunalahti.fi] has joined #xen |
| 11:18 | |-| | bartman_ [~bart@CPE0014bfaf982f-CM0011ae8a8ba6.cpe.net.cable.rogers.com] has quit [Ping timeout: 480 seconds] |
| 11:38 | |-| | MDC [~marcus@83.227.146.130] has joined #xen |
| 11:39 | <MDC> | quick question: is there a debian package with [hvm|vmx]loader in unstable? or is there any other repository I can use? |
| 11:42 | |-| | bartman_ [~bart@CPE0014bfaf982f-CM0011ae8a8ba6.cpe.net.cable.rogers.com] has joined #xen |
| 11:47 | |-| | harry_ [~harry@blueice2n1.uk.ibm.com] has quit [Quit: Leaving] |
| 11:47 | |-| | Basic_py [~Basic@65.193.16.100] has joined #xen |
| 11:53 | |-| | mastermind [~mastermin@mastermind.kaltenbrunner.cc] has joined #xen |
| 11:58 | |-| | SLot [~SLot___@gustavo.gruposim.com.br] has quit [Remote host closed the connection] |
| 11:58 | |-| | athomas [~athomas@host86-136-84-147.range86-136.btcentralplus.com] has quit [Quit: Leaving] |
| 12:15 | <MDC> | does the debian packages have support for vnc? |
| 12:15 | <Guy-> | as in, virtual network computing? |
| 12:17 | <MDC> | as in tightvnc for example... don't now if it stands for virtual network computing.. |
| 12:18 | <Guy-> | it does |
| 12:18 | <Guy-> | but what does it have to do with xen? |
| 12:18 | <MDC> | i'm (trying) to get win2003 to get up and running and it doesnt have a console ;-) so I have to use vnc to see whats coming from it.. |
| 12:19 | <MDC> | ok, i see, i mean if the debian xen package have support for vnc |
| 12:19 | <Guy-> | I didn't know there was any relation between xen and vnc |
| 12:19 | <Guy-> | can't you use remote desktop? |
| 12:20 | <MDC> | have to install it first ;-) |
| 12:20 | <MDC> | from the (xen) hvmexample file: "enable VNC library for graphics, default = 1" |
| 12:20 | <Guy-> | ah |
| 12:20 | <Guy-> | I didn't know about that |
| 12:21 | <MDC> | i've just got a amd X2 4200 AM2 socket processor and wants to test all the goodies that xen has :-) |
| 12:22 | <Guy-> | well, the debian xen packages build-depend on libvncserver-dev |
| 12:23 | |-| | jasonmc [jason@spoon.netsoc.tcd.ie] has quit [Quit: leaving] |
| 12:23 | <Guy-> | so my guess is vnc is supported |
| 12:23 | <murb> | Guy-: which means pretty much nothing.. |
| 12:24 | <Guy-> | sure |
| 12:24 | <mastermind> | ~# xm destroy 2783 |
| 12:24 | <mastermind> | Error: an integer is required |
| 12:24 | <mastermind> | ah yeah ... |
| 12:24 | <MDC> | ok, so why doesnt it starts listening on 590X? |
| 12:25 | <MDC> | how can I find out if it has vnc support then? |
| 12:25 | <mastermind> | MDC: just check if qmu-dm is linked against libvnc |
| 12:26 | <MDC> | mastermind; hmm.. i don't have the qemu-dm binary.. could be that then.. |
| 12:27 | <MDC> | mastermind; i'm using the debian packages, but it looks like its missing a few important things... |
| 12:27 | <MDC> | but how come it start anyway (start = no error on xm create) |
| 12:27 | <mastermind> | MDC: naaah - it's just me :-) |
| 12:27 | <mastermind> | the binary is called qemu-dm ;-) |
| 12:27 | <MDC> | hehe |
| 12:28 | <MDC> | don't have it.. |
| 12:28 | <MDC> | will copy it from xen binary install |
| 12:29 | <mastermind> | MDC: i find it hard to believe that it is missing |
| 12:29 | <mastermind> | are you really using xen3 packages ? |
| 12:29 | <MDC> | have i missed a deb package? |
| 12:29 | <MDC> | yes |
| 12:30 | <mastermind> | MDC: no idea on the debian packages - i use a source install on my box |
| 12:30 | <MDC> | packages installed; xen-hypervisor-3.0-i386 xen-utils-3.0 linux-image-2.6.16-2-xen-k7 |
| 12:30 | <MDC> | ok.. will try copy a few things and see what happends, else i have to use the source.. |
| 12:35 | |-| | visik7 [~visi@87.7.47.104] has joined #xen |
| 12:38 | |-| | mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has joined #xen |
| 12:45 | <MDC> | ok, so i have all the so files, no error in /var/log/xend-debug.log, after xm create, xm says the domain is started.. but using no cpu at all, and no vnc listener.. |
| 13:18 | |-| | zwane_work [~zwane@24.85.144.101] has quit [Ping timeout: 480 seconds] |
| 13:22 | |-| | jimix [~jimix@yktgi01e0-s4.watson.ibm.com] has quit [Quit: jimix] |
| 13:23 | |-| | visik7 [~visi@87.7.47.104] has quit [Remote host closed the connection] |
| 13:30 | |-| | jimix [~jimix@yktgi01e0-s4.watson.ibm.com] has joined #xen |
| 13:40 | |-| | bartman_ [~bart@CPE0014bfaf982f-CM0011ae8a8ba6.cpe.net.cable.rogers.com] has left #xen [] |
| 13:40 | |-| | mastermind [~mastermin@mastermind.kaltenbrunner.cc] has quit [Read error: Connection reset by peer] |
| 13:41 | |-| | mastermind [~mastermin@mastermind.kaltenbrunner.cc] has joined #xen |
| 13:43 | <MDC> | ok, vncserver seems to be running, but the domain isnt doing anything.. some tips? |
| 13:47 | <MDC> | cpu time is 0.0 and i cant connect to the vnc.. have setup cd, kernel (hvmloader) and a image as hd.. anything else? |
| 14:16 | |-| | zwane_work [~zwane@24.85.128.227] has joined #xen |
| 14:48 | |-| | mejlholm [~mejlholm@port79.ds1-abc.adsl.cybercity.dk] has quit [Quit: Client exiting] |
| 14:51 | |-| | cableman [~cableman@3e6b5b45.rev.stofanet.dk] has quit [Quit: Leaving] |
| 15:01 | |-| | mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has left #xen [] |
| 15:01 | |-| | mdday [~mdday@cpe-024-163-120-222.nc.res.rr.com] has joined #xen |
| 15:08 | <MDC> | ok, it's working now, thanks |
| 15:09 | <Guy-> | what was the solution? |
| 15:24 | |-| | jerone [~jerone@pixpat.austin.ibm.com] has quit [Quit: Leaving] |
| 15:36 | <MDC> | installed the pre-built binaries from xen.. |
| 15:37 | <MDC> | but now the win2003 installations hangs at approx 13 min left.. |
| 15:49 | |-| | lazyb0y_ [~henning@85.176.127.157] has joined #xen |
| 15:50 | |-| | zwane_work [~zwane@24.85.128.227] has quit [Ping timeout: 480 seconds] |
| 15:56 | |-| | lazyb0y [~henning@e176103102.adsl.alicedsl.de] has quit [Ping timeout: 480 seconds] |
| 15:59 | |-| | jerone [~jerone@adsl-71-145-148-37.dsl.austtx.sbcglobal.net] has joined #xen |
| 16:02 | |-| | aliguori [~anthony@cpe-70-116-9-243.austin.res.rr.com] has joined #xen |
| 16:16 | |-| | zwane_work [~zwane@24.85.144.101] has joined #xen |
| 16:42 | |-| | bernarde [~bernarde@201.44.214.48] has quit [Remote host closed the connection] |
| 17:00 | |-| | hollisb [~hollisb@pixpat.austin.ibm.com] has quit [Quit: leaving] |
| 17:19 | |-| | JViz [Anomaly@adsl-065-013-131-023.sip.int.bellsouth.net] has quit [Read error: Connection reset by peer] |
| 17:32 | |-| | tom [~tom@nonserviam.net] has joined #xen |
| 17:32 | |-| | tom changed nick to PerlSaiyan |
| 17:32 | <PerlSaiyan> | woohoo! |
| 17:32 | <PerlSaiyan> | Who's used mcelog and Xen 3.0.2! |
| 17:33 | <PerlSaiyan> | more ? and less ! |
| 17:34 | |-| | Basic_py [~Basic@65.193.16.100] has quit [Quit: Leaving] |
| 17:37 | |-| | plars [~plars@pixpat.austin.ibm.com] has quit [Remote host closed the connection] |
| 17:39 | |-| | visik7 [~visi@87.7.47.104] has joined #xen |
| 17:41 | |-| | jimix [~jimix@yktgi01e0-s4.watson.ibm.com] has quit [Quit: jimix] |
| 17:41 | |-| | rharper [~rharper@pixpat.austin.ibm.com] has quit [Quit: Leaving] |
| 17:49 | <visik7> | what's the differences of perfomance of a modified domU kernel such as linux and a plain linux with vmx support ? |
| 17:50 | <aliguori> | a bunch |
| 17:50 | <visik7> | 10% or less ? |
| 17:50 | <aliguori> | i'm not sure that specifics have ever been published and it will depend greatly on your workloda |
| 17:51 | <aliguori> | depends on your workload |
| 17:51 | <aliguori> | if it's IO intensive, expect very large performance hits |
| 17:51 | <aliguori> | brb |
| 17:51 | |-| | aliguori [~anthony@cpe-70-116-9-243.austin.res.rr.com] has quit [Quit: Ex-Chat] |
| 17:55 | |-| | Shoragan_ [~shoragan@d072.apm.etc.tu-bs.de] has quit [Quit: Leaving] |
| 18:04 | |-| | aliguori [~anthony@cpe-70-116-9-243.austin.res.rr.com] has joined #xen |
| 18:19 | |-| | jerone [~jerone@adsl-71-145-148-37.dsl.austtx.sbcglobal.net] has left #xen [Leaving] |
| 18:28 | |-| | JViz [Anomaly@adsl-065-013-131-023.sip.int.bellsouth.net] has joined #xen |
| 18:40 | |-| | ns [~nivedita@c-67-171-181-157.hsd1.or.comcast.net] has quit [Quit: The voices in my head are telling me to...] |
| 18:53 | |-| | ns [~niv@bi01p1.co.us.ibm.com] has joined #xen |
| 18:54 | |-| | tessier_ [~treed@69.43.173.46] has joined #xen |
| 18:54 | <tessier_> | Hello all! |
| 18:55 | <tessier_> | Anyone know if it is possible to get xen and a dom0 kernel booted onto a diskless compute node? That is what I am attempting at the moment. |
| 18:55 | <tessier_> | I intend to use AoE for the root. |
| 18:59 | <visik7> | livecd ? |
| 18:59 | <visik7> | boot xen through syslinux |
| 18:59 | <visik7> | should be possible |
| 19:01 | |-| | visik7 [~visi@87.7.47.104] has quit [Remote host closed the connection] |
| 19:02 | |-| | stekloff [~stekloff@bi01p1.co.us.ibm.com] has quit [Quit: Leaving] |
| 19:06 | |-| | ns [~niv@bi01p1.co.us.ibm.com] has quit [Quit: The voices in my head are telling me to...] |
| 19:14 | <tessier_> | livecd is not possible. These cpu nodes are totally diskless and in cases in which disks cannot be installed. |
| 19:15 | <tessier_> | hmm...cfservd is still complaining that it cannot resolve an old hostname. |
| 19:15 | <tessier_> | Jun 20 16:58:37 sdapp01 cfservd[5841]: Unable to lookup hostname (dj-int.int.drjays.com) or cfengine service: Name or service not known |
| 19:15 | <tessier_> | But that hostname is mentioned nowhere anymore. |
| 19:55 | |-| | jimix [~jimix@ip139.152.45.216.suscom.net] has joined #xen |
| 20:11 | |-| | horms [~horms@vagw.valinux.co.jp] has joined #xen |
| 20:13 | |-| | horms [~horms@vagw.valinux.co.jp] has quit [Quit: ] |
| 20:13 | |-| | horms [~horms@vagw.valinux.co.jp] has joined #xen |
| 20:22 | |-| | jimix [~jimix@ip139.152.45.216.suscom.net] has quit [Quit: jimix] |
| 21:01 | |-| | ramsmith [~ram@211.30.173.81] has joined #xen |
| 21:04 | |-| | Z3R0`c00l [~zerocool@ip-36-216.sn2.eutelia.it] has joined #xen |
| 21:16 | |-| | ramsmith [~ram@211.30.173.81] has quit [Quit: Download Gaim: http://gaim.sourceforge.net/] |
| 21:41 | |-| | Basic_py [~Basic@warden.real-time.com] has joined #xen |
| 22:43 | |-| | womble [~mpalmer@220.233.135.106] has joined #xen |
| 22:43 | |-| | ns [~nivedita@67.171.181.157] has joined #xen |
| 22:59 | |-| | VS_ChanLog [~stats@ns.theshore.net] has left #xen [Rotating Logs] |
| 22:59 | |-| | VS_ChanLog [~stats@ns.theshore.net] has joined #xen |
| 23:09 | |-| | horms [~horms@vagw.valinux.co.jp] has quit [Quit: Leaving] |
| --- | Log | closed Wed Jun 21 00:00:31 2006 |
