[00:14] how can i get my server to host remote desktop sessions? [00:15] like a windows remote desktop-terminal server does [00:41] Hi, there seems to be an "admin" group on ubuntu-server which is preventing me from creating an "admin" user that I need for my admin@mydomain email address [00:41] Would it be safe to remove that group or is there some way around it? [00:47] you could just call the initial user admin? [01:02] wo0f, oh.. what do you mean by initial user? [05:13] im getting errors like this: http://pastebin.com/m19871e7a.... whats wrong? [05:18] New bug: #300151 in openssh "ssh slow during logon, when ENTERING INTERACTIVE SESSION" [Undecided,New] https://launchpad.net/bugs/300151 [08:02] New bug: #300615 in mysql-dfsg-5.0 (main) "package mysql-server-5.0 None [modified: /var/lib/dpkg/info/mysql-server-5.0.list] failed to install/upgrade: subprocess pre-installation script returned error exit status 1" [Undecided,New] https://launchpad.net/bugs/300615 [08:15] https://help.ubuntu.com/8.10/serverguide/C/samba-ldap.html - not a draft anymore? [08:15] it still does contain mistakes [08:15] as an example: ldapadd -x -D cn=admin,cn=config -f /tmp/ldif_output/cn\=config/cn\=schema/cn\=\{12\}misc.ldif [08:16] missing -W and misc should be replaced with samba [08:16] btw kerberos configuration is a bit messy too === ssd7 is now known as Guest93322 [08:22] nme_: COuld you file a bug about it here: https://bugs.edge.launchpad.net/ubuntu-doc/+filebug ? [08:22] Thanks! [08:23] moin [08:26] soren: i will in the spare time, right now im trying to get things work ;) btw. it would be much simpler to allow people to comment entries on docs page - Like it is possible at django online book - take a look at http://www.djangobook.com/en/1.0/chapter06/ [08:27] nme_: I've never really liked that sort of thing. [08:28] so maybe you will consider hiring at RedHat? ;) [08:28] nme_: It makes sense on articles that are expected to be immutable, but for documentation that's expected to be refined all the time, it's just confusing. [08:28] nme_: Eh? [08:28] as Ubuntu says its is soo comunity oriented distribution - this way of extending documentation would be very natural [08:29] Oh, we're happy to accept comments. [08:29] I just don't think that's the right way to do it. [08:29] take a look at php docs page - comment resources are great [08:29] See, the problem is this. [08:29] : [08:29] If there's a section that has problems, someone makes a comment to that effect... [08:29] ...then someone makes an attempt to fix it... [08:29] What to do with the comment? [08:29] Remove it? [08:29] But what if the commenter doesn't find his complaints were properly addressed? [08:30] Leave them? [08:30] Well, that's just confusing to anyone who reads them later. [08:30] give me a sec (im at work, phone..) [08:30] Sure, take your time. [08:30] I'm here all day :) [08:30] nme_: Dear, oh dear, no. The PHP pages are even worse. [08:31] In fact, for the php ones, I think they make the docs worse. [08:31] ..because there's apparantly not enough incentive to fix the docs, because "hey, comment number 97 explains how to do it properly". [08:32] In the case of php, for instance, you can download the docs in various formats, but they don't (and shouldn't) contain the comments... [08:32] ...but without the comments, the docs suck. [08:33] The only right way to go is to fix the docs themselves. [08:33] there are two options - one guys writes docs - ten testers "do" actual tests... he corrects things... and so 11 guys get paid... on the other side - one guy writes docs - community points mistakes - the other guy interactive'ly administers web page comments - merging them to docs, updating docs, filtering some content... so 2 guys get paid more... not 11. [08:33] ...and comments on page like that makes for a lousy bugtracker. [08:34] Anyone is free to join the documentation team. [08:34] (noone on the documentation team gets paid, by the way, as far as I know) [08:34] Ok, a random example from the django page... [08:34] Someone filed this comment: "I don't understand this section at all. I thought the previous section already was customizing the look and feel of the Django admin index page. Also, I don't see the available applications listed anywhere on the page. What is the URL that you're referring to? " [08:35] how to handle that? It's not like he's proposing an alternate wording. [08:35] He's just dissatisfied with the current content. [08:35] When is it ok to remove such a comment? [08:35] The answeR: [08:35] When the commenter feels his complaints have been addressed. Not sooner nor later. [08:36] ...but how to tell? [08:36] That's why this stuff belongs in a bugtracker. [08:36] Any set of docs that only feels complete with the comments attached to it is broken, IMO. [08:37] The alternative is to leave the comments there indefinitely. [08:37] That's troublesome in two ways: [08:38] If it's attached to a specific paragraph as in the django case, what happens if that paragraph ceases to exist, because the docs are refactored somehow? [08:39] in such situation when comment does not suggest any changes and only shows dissatisfaction about some paragraph - we still got options - one is to (allowing only registered users to comment) drop his comment sending it back to his mail with an answer, another is to hide this paragraph and open new one with better content [08:39] If the comment has been adressed, but it's still there it's either just noise (since it doesn't add anything that's not currently in the docs) or plain confusing because it makes a fuss about something that's been reworded. [08:40] I don't see a problem with integrating the server guide with the bug tracker. Such as putting links here and there that automatically fill in parts of the bug report and makes that part easier... [08:40] but the ubuntu's bugtracker is a bit dead... [08:40] ...but using a comment system in place of a bugtracker? I'm sorry, but I think that's a horrible idea. [08:40] nme_: Dead? How so? [08:41] as I reported 2 weeks ago a problem which concerns casper and kvm, still noone looked at it [08:41] That's a manpower problem. [08:41] It only makes things worse to add *another* system for tracking bugs. [08:42] after some time I've just placed my suggestion how to solve it, where the problem lies... and it is still, opened... [08:42] nme_: What's the bug number of the casper+kvm thing, by the way? [08:43] nme_: Please appreciate that we have tons and tons of stuff to do. [08:43] nme_: Like, say, answering stuff on IRC. [08:44] [Bug 296089] [NEW] casper boot=nfs ipconfig: eth0: SIOCGIFINDEX: No such device [08:44] Launchpad bug 296089 in casper "casper boot=nfs ipconfig: eth0: SIOCGIFINDEX: No such device" [Undecided,New] https://launchpad.net/bugs/296089 [08:44] thats why I'm trying to help :) [08:45] You see - I've already suggested how to change 11 people working on a issue to 2 people... [08:45] foolano: ono any better than timofonica where you are? [08:46] And I've tried to explain why you're wrong. [08:46] The docs aren't written in the way you clearly think they are. [08:46] and how about maintaining debian packages? [08:46] How about that? [08:47] I don't understand the question? [08:47] that where the problem of launchpad is [08:47] some guys are developing a code... [08:47] The problem of launchpad is maintaining debian pakcages? [08:47] they know whats the best as a default setup... [08:48] then comes a Debian maintainer and he says "no, in my opinion it will be like this..." [08:48] and he prepares the package [08:48] like for instance infamous openssl maintaniner did ;) [08:48] but then [08:49] this package is repackaged and called ubuntu one [08:49] when some guys have a problem he reports it where? at ubuntu launchpad. [08:49] Do you understand what "repackaged" means? [08:49] where nobody knows anything about it, because given package was prepared by a Debian maintaniner [08:50] Look, this discussion is pointless. It's very obvious that you're not familiar with your development model, and you're making lots of assumptions that are just plain wrong. [08:50] im not sure - something like sed -i s/Debian/Ubuntu/gc ... [08:51] *sigh* [08:51] Why would we do that? [08:51] The only right thing to do is to report it to the distribution where you got it. [08:52] in fact, if you went on to report a bug in Debian's BTS about a package you got from Ubuntu, you'd better be prepared to handle the consequences. Lots of Debian developers get *very* annoyed by that. [08:52] I did not want to argue, just wanted to point a few things [08:52] Well, if you just want to say things and not hear the response, I don't see the point. [08:53] I read Your answers [08:53] Look, if you get a package from Ubuntu and something fails, you should either report it to Ubuntu or be familiar enough with the package itself, its dependencies and all the other relationships it has with other packages, before you just go ahead and assume that it's a problem in the Debian package as well. [08:53] Most users aren't. [08:53] Heck, some developers aren't. [08:53] ...so treating it as an Ubuntu bug first is the right thing to do. [08:54] ok then [08:54] so [08:54] ...if we look at it, and can see that it must be a bug in Debian, we forward it to them. [08:54] when there is a Debian package in Ubuntu distribution [08:54] Often with a fix. [08:54] and You get a problem [08:54] and you report it to ubuntu launchpad [08:54] and You will not get the answer [08:55] Ubuntu is a community effort. [08:55] (which is probably) [08:55] So is Debian. [08:56] Why is it that you think that failure to keep up with the bug volume is a problem that's unique to Ubuntu? [08:56] no [08:57] all I want to tell is that lets say 50% of packages are unmaintained by anyone at Ubuntu side, there are just taken from Debian giving community no possible way to get any support for them [08:58] huh. [08:58] nme_: the fact that the packages are in sync with Debian doesn't mean they are unmaintained. [08:59] nor developers nor maintainers are seers... Ubuntu is growing extremally strong currently... There are a LOT of people... Most of Linux desktops are Ubuntu... [08:59] most ubuntu installs are desktops as well [08:59] as my part job is to administer a student campus network of over 1200 hosts... and if anyone asks about Linux I say Ubuntu. [08:59] but there are dark sides I already see... [08:59] recommend os x, they'll appreciate it ;) [09:00] os x is illegal at every hardware which is not apple originated. [09:00] dark sides [09:00] I'll explain it in a minute (phone) [09:02] if by "dark sides" you mean "i submit a bug and noone seems to be looking at it" I'm just not sure you wouldn't find those dark sides anywhere [09:02] nme_: if you have suggestions on how ~100 people (only half of which work full time on Ubuntu) can better keep up with 10000 new bugs reported each month, we'd be happy to hear it. [09:02] Trust me, though: Adding alternate means for "tracking" bugs is not the right approach. [09:03] soren: maybe his point is that choosing a distro with a higher developer/user ratio might help in getting timely response [09:03] it doesn't help in getting better packages in general though :) [09:03] im back. are you familiar with avahi ? [09:04] sounds like he needs a much smaller distro with only a handful of users, whom are the developers? ;) [09:04] nme_: Yes. [09:04] Deeps: speaking about me? speaking about Slackware? [09:05] soren: do You know that avahi in hardy caused arp flood at some kind of networks? [09:05] the problem was reported only by Ubuntu users [09:05] nme_: No. [09:05] i dont know anything about slackware so i'm certainly not speaking of that [09:05] it did not happen on any other distribution at the networks I administer [09:06] nme_: I think I'm missing your point. [09:06] and the problem lasted for above halp a hear [09:06] half a year [09:07] how to make life easier for those 100 people? trust the community a bit more ;) [09:07] Can you elaborate? [09:07] about trust? [09:08] soren: Create less bugs ;) That way it's easier to keep up. :P [09:08] Jeeves_: Thanks, I'll make a note of that. [09:08] :p [09:09] nme_: Yes. [09:09] nme_: by sitting here bitching to the people that actually do track the bugs and help resolve them, you're actually slowing things down more. if you want it to improve, get involved more yourself. [09:09] mental note: Create better code, will create less bugs [09:09] * Koon tries to fix more bugs than he reports, to be part of the solution ;) [09:09] Koon: That's a good start :) [09:09] I'll just try to ignore bugs [09:09] That should save you guys work too :) [09:10] Deeps: I explain - Slackware is simple distro. Its very old one, maintained for a long time by one single guy. All he did was preparation of init scripts and making simple package system which are automatically build on default settings which project developer set. Belive me or not, it really works. If someone dislikes default - he rebuilds given package by his own. [09:11] nme_: Why not join #slackware [09:11] Or #swaret [09:11] gee only tried to help [09:11] suggest few approaches [09:11] all i see is an anger and hate now [09:11] nme_: By telling developers to stop developing and trhowing people in the deep [09:11] I wont bitch to people anymore, sorry [09:12] Jeeves_: where I said this? who is this developer you mantioned? [09:12] nme_: Slackware users are people that enjoy screwing around with themselves, enjoying pain and agony fixing a simple box. [09:12] Ubuntu users like it to 'just work' [09:12] That needs more development, and thus more bugs [09:13] nme_: you can do that in ubuntu too [09:13] Thats why I suggest it to desktop users [09:13] y I know [09:13] nme_: I administer a few hundred Ubuntu servers [09:14] And we did Slackware before [09:14] And we're all very happy that slackware doesn't exist anymore :) [09:14] nme_: the flipside to the slackware model is that the packages wont conform to any standards, each package will do things differently as each developer will do things differently. ubuntu standardises a number of things aross all packages to make it easier to maintain. bugs occur in the process. and the joys that come with slackware allowing you to rebuild a package in the way you like also exist in ubuntu. apt-get source and you're half way th [09:15] nme_: so i dont see your point. slackware does less to make life easy for you and offers feature A. ubuntu does more to make life easy for you and also offers feature A. [09:16] nme_: I still don't understand what you mean about trusting the community. [09:16] nme_: you point out that slackware's great because you can fix stuff to work how you want it yourself, yet moan that ubuntu doesn't fix stuff quickly enough for you? and then suggest that slackware's better? [09:16] there is a point [09:17] Slackware actually originates from base tar.gz from developers project page [09:17] So does Ubuntu packages.. [09:17] but go on. [09:17] if You got a problem with specific package, you can go to the project developers - not the distro maintainers [09:18] You're free to do that with Ubuntu packages, too. Noone's forcing you to use Launchpad. [09:19] no. the guys will tell you to compile it as default - then check if a problem exists and then - report it :) [09:20] well duh [09:20] k guys, lets have a break, got a bit work to do :/ [09:20] So what you're saying is that most bugs in Ubuntu are because we break things that were perfectly functional when the upstream developers shipped it? [09:20] If that's truly what you believe, I'd advise you to find another distro. I wouldn't use a distro I didn't trust. [09:21] soren: "So does Ubuntu packages" -> apt-get source iptables [09:21] Yes? [09:21] What does that give you? [09:21] It gives you the original tarball + the code it takes to make it into an Ubuntu package. [09:22] What? You think Slackware just downloads the tarballs and stares at it until it turns into a binary package? [09:23] I also can testify that upstream prefer a distro with somewhat-unique compile flags than one where... every user uses a different set [09:23] heheh :) [09:23] soren: No! You type 'pkg_install iptables' or 'swaret install' and than stuff happens, you hope [09:23] (if you see what I mean) [09:24] i think nme_'s point in this case is that iptables comes from the debian package [09:24] soren: execute this command and see it for yourself [09:24] nme_: Look. I *made* that package. I know what that command does. [09:24] no [09:24] No??!?? [09:24] im just pointing how many patches it got [09:24] its not pure iptables [09:24] Yes. TO FIX BUGS! [09:25] What do you think happens? We download the upstream tarball and start looking around for random ways to break it? [09:26] soren: There are no bugs in upstream packages, Patrick says so! [09:26] paddy's always right [09:26] http://www.youtube.com/watch?v=9OydvSCGxt0 [09:27] the main difference is that as a "distribution" we feel responsible for the bugs in the software we ship [09:27] nme_: You're patrick? [09:27] If that's what you think of us, I strongly suggest you find another distro. I wouldn't sleep well at night at all if I thought the people who made my distribution were monkeys on crack who spent all day harvesting bad patches from the internet and applied it to perfectly good packages. [09:27] soren: wow, take IT seriously much? lol [09:27] i sleep perfectly well knowing my OS is developed by monkeys on crack [09:27] Jeeves_: good one ;] [09:28] * Deeps <3 winxp [09:28] nme_: Really, what the **** are you doing in here? [09:28] Jeeves_: Patrick? [09:28] Patrick J. Volkerding ;) [09:29] good sense of humor ;) [09:29] soren: http://en.wikipedia.org/wiki/Patrick_Volkerding [09:29] Ah. [09:29] Jeeves_: wanted to help [09:29] nme_: How? [09:30] Jeeves_: I think he really wants to help [09:30] i only came to ask why there are no way to put comments on docs (which yesterday were marked as DRAFT) [09:30] because i did make step by step and found few problems [09:30] nme_: You mean documentation bugs? [09:31] y i know that i should put it on launchpad [09:31] it's just that he has difficulties scaling community processes from small projects to something like Ubuntu [09:31] as commercial approach [09:32] there are many ways to get community work at Your profit [09:32] making less costs, letting you guys earn more [09:33] * Kamping_Kaiser blink [09:33] wtf is happening here? [09:33] Jeeves_, evening mate [09:33] Kamping_Kaiser: Hi there! [09:34] :) [09:34] Speaking of proprietary software [09:34] * Kamping_Kaiser worries about the Kamping_Kaiser -> proprietary software mental jump [09:35] Kamping_Kaiser: [09:35] Just because Canonical produces proprietary support software doesn't [09:35] justify other companies doing it (or making it an ok thing to do). [09:35] What do you mean by that? :) [09:36] Jeeves_, it sounded like the argument was 'canonical who own ubuntu do $FOO, so $FOO is ok for others to do' (which i think is a falacy). It may or may not have been relevant to the convo ;) [09:37] Kamping_Kaiser: My point is that to make (good) money with open source, you need the open source to attrack people, and closed source as an extra to make money [09:37] Exactly like Canonical does [09:37] Nothing wrong with it [09:38] The only thing that's wrong about Landscape, for instance, is that you cannot host the server part yourselve [09:38] Jeeves_, thats entirely possible - I dont know if its true, but thats not an argument I can take part in - I dont know enough about such things. [09:38] It's very much ok that it's closed source, and payware [09:46] Jeeves_: you think it's good that there's a tool being used to administer your server that you have no idea how it works? [09:47] Deeps: I trust Canonical. [09:47] They create the entire OS the server runs on [09:47] the only part I'm interested in is easy upgrades [09:47] No, they produce a bunch of packages. [09:48] Deeps: And they patch and compile the whole bunch [09:48] Well, it's a bit more than that, but you get my drift [09:48] So the create is [09:48] s/s/t [09:48] "To the create is" ? [09:48] What? [09:49] soren: There isn't a /i! [09:49] "So the create it"? What does that mean? [09:49] okok [09:49] +y somewhere :) [09:49] "Soy the create it". Oh, ok. [09:49] "So they create it." [09:49] :p [09:49] grmbl [09:49] hehe [09:49] while it's a great business move for canonical, i think it's terrible that it's closed source. cant praise open source with one hand and then shoot it down with the other [09:50] Anywho, I do trust Canonical and the Ubuntu community [09:50] If I didn't, I would be running Gentoo [09:50] massively hypocritical [09:50] not Slackware? [09:50] Deeps: No it isn't. [09:50] ;) [09:50] nme_: No, I don't trust Patrick [09:50] speaking of Landscape, do You guys know why DELL does not sell Ubuntu Server, saying that it does not work with their SAN arrays? [09:51] Jeeves_: NO U [09:51] nme_: Dell hardware usually doesn't work at all. [09:51] ?!?? [09:51] So they only ship it with Redhat [09:51] Which also doesn't work :) [09:51] RHEL or SUSE [09:51] they told us yesterday. [09:51] What does landscape, Dell and SAN's have to do with one another? [09:52] soren: leave him alone! :) [09:52] All these hard questions to answer! [09:52] to have Landscape access one need to have commercial Ubuntu, right? [09:52] But I want to understand! [09:52] nme_: There is no commercial Ubuntu [09:52] nme_: I think you can buy it seperately, but let's say "yes" and see where this goes.. [09:52] :P [09:52] when we buy more hardware, we buy software too that will support the hardware. [09:53] Right, there's no "commercial Ubuntu". There's only one Ubuntu. [09:53] ...but you can buy a support contract, and when/if you do, you get landscape as well. [09:54] to have complete support we need it for hardware and software [09:55] ok, now I know :) [09:58] I'm still curious what Dell said. [09:59] Did they say it wouldn't work or that it wasn't supported. That's not the same thing. [10:06] there were two guys - one was sale oriented, the other - technical oriented, the second one - while chat, when I asked why RH, can't we run Ubuntu? He said that SAN array we will buy requires multipathing which does work on RH and SUSE and it DOES NOT work on Ubuntu because of some problems. [10:10] nme_: Hmm [10:13] when we do the deal with Dell I'll run Ubuntu to check and solve it for myself anyway... but it might be possible to do it with IBM. Dont really know how they will react to my suggestion about Ubuntu instead of proposed RH at first shot. [10:17] nme_: No, but I do know that for sanity and reliability, avoid any storage solution from dell at all cost. :) [10:18] Or at least have a "make it work or take it back" paragraph in your contract, so you have something to threaten with when it doesn't work [10:18] maswan: You're serious? [10:19] we are currently during conversation with IBM and Sun too... [10:19] but I dont like Sun anymore and they will fail with their offer anyway... [10:20] nme_: I have several collegues at other sites that have had to go to the contractual "ok, then you'll have to take it back" before they got as far as getting someone that could actually give them some "oh, but you need this and that firmware, etc, etc" to make it work. [10:22] nme_: I'd go for Sun. :) [10:22] Dell sucks, bigtime [10:22] maswan: what other SAN solutions You consider better? we are running Sun storage which works quite good but soon the hardware wont be supported anymore. [10:22] nme_: Keep on running Sun :) [10:22] nme_: What kind of stuff do you need? [10:23] SAS? FC? iScsi? [10:23] all in one would be the best [10:23] to have primary servers on FC but still have possibility to connect some server through iscsi [10:26] Dell says that they got FC and iscsi cards for their array but only one standard can be currently used at the time. You need to choose. They also say that in the future after some upgrade - two will be able to work concurrently [10:27] nme_: That seems logical [10:28] what seems logical? [10:29] soren, you're a virtualization guy, right? When you make a virtual disk, that's very fast. Any idea how they do that? I mean, an 8GB disk was created in only a few seconds. [10:29] qemu-img create -f qcow2 sda.qcow2 8G [10:30] less than a second [10:30] joerlend: if it's not allocating the entire diskspace at creation time, then it's gonna be very quick [10:30] right. But how does qemu-img do that? [10:30] Deeps, again, that's what I want to do. :) [10:30] qcow2 data format [10:30] if you copy that file, then it's still 8GB being copied? [10:30] joerlend: every app can open a file, seek to +8GB and write file [10:31] but qcow2 seems a better solution ;) [10:31] seek is being used? [10:31] not in qcow2 [10:31] joerlend: you want to allocate all the space at creation time? [10:31] that is, you could use seek in dd to achieve the same effect? [10:31] Deeps, someone wanted to make a large file in order to test something, but in a quick manner. [10:31] joerlend: exacly [10:32] nme_: Oh, well, personally I would avoid SANs, in favour of less complexity. :) [10:32] joerlend: if you want the file to be empty, then it's relatively easy, if you want the file to have random data to minimize the effect of compression screwing up your tests, different matter [10:33] joerlend: or rather, it's still relatively easy, but not quick [10:33] we have some ibm stuff though, and it kind of works. [10:33] Deeps, thanks. :) [10:34] if in the case of wanting random data, dd and /dev/[u]random are what you'll be most interested in, i suspect [10:37] Deeps, lets say you wanted to create a 50GB file very quickly. What would you do? :) [10:37] joerlend: It's a sparse file. [10:37] right. [10:38] joerlend: google for some C code that will open a file descriptor, seek +50G and then write the file, or use dd if=/dev/zero [10:38] joerlend: I'd use qemu-img. [10:38] dd if=/dev/zero of=file seek=FILESIZE_HERE bs=1 count=0 [10:38] what nme said [10:38] qemu-img create somefile 50G [10:39] Or kvm-img, but you get the idea. [10:39] * soren runs to get his power supply [10:40] phew... Nick of time. [10:40] Deeps, soren: if you'd try to copy that file to another partition, then it would actually copy the filesize of nulldata? [10:40] joerlend: Depends. [10:40] on..? [10:40] Stuff. [10:40] :p [10:40] hehe [10:41] Probably not. [10:41] whatever you're using may realise it's all null data and compress it on the fly, artificially speeding up the transfer process [10:41] that was the usecase here. He wanted to test copying speeds with a large file. [10:41] It would probably create another sparse file on the destination partition. [10:41] hense why you'd probably want to use a file filled with random data [10:41] Oh. Just do "cp --sparse=never" [10:41] thanks! :) [10:42] or that [11:07] soren: you're virtualization guy, have you read any papers about how many virtual machines could work on specified numer of cpu / cores? im interested in some optimal amounts, You work on kvm only? [11:07] Mostly kvm. [11:07] There's no such number. It depends entirely on what the virtual machines are doing. [11:08] I tried Xen for a while but then I switched to kvm, works pretty good... I love qcow2 :) [11:08] I have a machine that runs 11 vm's. [11:08] and even then it's difficult to choose solutions, since it's difficult to find real numbers. [11:08] Those VM's aren't very heavy, so it's not a problem at all. [11:09] You might run 2 you might run 20 on the same hardware. [11:09] In other cases I wouldn't run more than a single virtual machine on a physical machine. [11:12] I got some ancient "terminal servers" solution here... they work under Windows Server, their load is high. Primarily those servers worked only on physical machines 2x Xeon 2 GHz (one core)... currently placed one of them as a virtual on 2x Xeon Dual 1.86 GHz each, but seems that load is higher... and Im thinking about placing 2 same vm on single hardware dividing clients to those two virtual terminal servers, how you think - how will servers load behave? [11:14] Let me see if I understand correctly.. [11:14] You used to have two of these machines, each running on a Xeon 2 GHz? [11:14] some old xeon single core machines [11:15] Right. [11:15] now i got spare 2 x Xeon dual core 1.8 GHz [11:15] running Ubuntu [11:15] Ok. [11:15] $SOME load [11:16] And how many cpus have you assigned to the guests? [11:16] and Im running there kvm with exact copy of system from physical machine above [11:16] ok. [11:16] I have 2 cores to VM (physical machine have 4 of them) [11:16] I gave 2 cores... ^ [11:17] Ok. The guest is running Windows? [11:17] y, Windows 2003 Server [11:17] And it's using both cores? [11:18] dont know how to check it - currently machine runs DRBD, kvm and 1 vm (this Windows 2003 Server), /proc/interrupts shows every single core is used in same manner... [11:18]