1124150477 M * Bertl you might want to play with the token bucket adjustments .. they allow for really good fine tuning ... so if you do it properly you can reduce latency (which is probably of interest for game servers) 1124150527 M * eugenesan and Gbit NIC on PCIE works fine, but kernel says: "pcie_portdrv_probe->Dev[005d:10de] has invalid IRQ. Check vendor BIOS" 1124150552 M * Bertl hmm, so it might be a bios issue ... 1124150556 M * eugenesan well tomorrow i wil play with it 1124150605 M * Bertl excellent, you might consider adding yourself/your company to the list of vserver hosting providers/happy users once everything is working fine ... 1124150629 M * eugenesan maybe, but googling showed that all have that messege. 1124150662 M * eugenesan i hope i will be one of happy users of vserver :-) 1124150690 M * Bertl eugenesan: yeah, could easily be, x86_64 is not x86 and PCI-E is not that common either ... btw, I thought PCI-X is going to make the race on x86_64 1124150791 M * eugenesan on my MoBo there are two NICs one is on PCIE and one on PCI-X but 32bit and runnign 100mhz. Nforce4 is realy weird beast. 1124150819 M * Bertl guess they are still experimenting ... 1124150832 M * eugenesan Nforce4Pro is using PCI-X 1124150904 M * eugenesan but building servers on Nforce4 is much cheaper while is fast enough. 1124150948 M * Bertl and how many servers/guests do you plan (with linux-vserver) 1124151042 M * Bertl (if I may ask :) 1124151047 M * eugenesan for now i am planning couple for 32bit tasks (shoutcast streaming and gaming) and one for 64bit for experiments. 1124151076 M * eugenesan there are still many applications available in 32bit only 1124151139 M * eugenesan in future i am planning separate server for vserver. 1124151165 J * Aiken ~james@tooax6-208.dialup.optusnet.com.au 1124151174 M * eugenesan btw, what distro are you using? 1124151180 M * Bertl morning Aiken! 1124151213 M * Aiken hello 1124151249 M * Aiken looks like the vserver I have setup on the alpha will get used soon, a friend wants access to it. 1124151264 M * Bertl eugenesan: I personally use a heavily patched version of Mandrake, somewhere inbetween 9.0 and 10.2 .. Enrico (the tools guy) uses Fedora (probably Core 3 or 4) and a lot of Folks here use Debian and Gentoo ... 1124151343 M * Bertl and I have to say that Gentoo seems pretty well maintained recently ... 1124151364 M * Bertl Aiken: great! 1124151501 M * eugenesan Bertl: gentoo seems to be miss perspective of "Fast Updating", building packages on live systems is utopia. 1124151533 M * eugenesan Bertl: gentoo is good for desktop/dev-station not production. 1124151568 M * Bertl well, I'm not talking about the distro itself, just about how up-to-date they are with linux-vserver support :) 1124151616 M * eugenesan yes, source-based distro always supports better projects in develement 1124151722 M * Bertl maybe, but I think it more depends on the people (the folks doing the packaging and this stuff) 1124151739 M * eugenesan btw, few months ago i was playing with Virtuozzo on w2k3, it seems like it it uses "vserver like" mechanism, am i wrong? 1124151791 M * Bertl guess not, it's the commercial version of linux-vserver ... slightly more expensive ... similar features ... 1124151802 M * Bertl s/version/pendant/ 1124151907 M * eugenesan also i found two projects dealing with vserver management (by molesoft and openvps) do you guys somehow cooperate? 1124151935 M * eugenesan i am looking for some kind of "Control Panel" or so. 1124151967 M * Bertl openvps does utilize the linux-vserver framework ... from molesoft I've not heard until yet ... 1124152024 M * eugenesan they have comercial control panel for vserver (i think) and screenshots looks very impressive 1124152040 M * Bertl really? you got an url for that? 1124152048 M * eugenesan i sec 1124152120 M * eugenesan http://molesoftware.com/www/index.php?i18n=en&pid=1&title=Produkte&sub=3 1124152126 M * Bertl tx 1124152289 M * eugenesan in past they released their project as GPL, maybe this one they will also... 1124152482 M * Bertl I guess mole soft is not using linux-vserver, maybe freevps or virtuozzo ... maybe something different ... 1124152516 M * eugenesan maybe freevps, but isn't it based on vserver? 1124152550 M * Bertl it was an early fork (somewhere around 1.0 for 2.4) I guess we do not share much code nowadays ... if any at all 1124152587 M * Bertl they went commercial, and they emphasize on the frontend (H-Sphere, IIRC) 1124152593 M * eugenesan but the idea is the same, i think control panel could be easely ported for vserver 1124152610 M * Bertl probably, but I guess that's not their interest :) 1124152639 M * Bertl (because they sell the control panel) 1124152735 M * eugenesan they were selling Hosting Panel also, and 6 months ago released it under GPL. 1124152751 M * Bertl really? 1124152766 M * eugenesan yes, i am using it :-) 1124152773 J * lilo_ tor@lilo.usercloak.oftc.net 1124152778 M * Bertl do you have an url for that too? 1124152794 M * eugenesan it's vhcs.net 1124152815 M * Bertl ah, that's from the molesoft folks ... 1124152821 M * eugenesan yes 1124152868 M * Bertl freevps is positive software ...http://www.freevps.com 1124152904 M * eugenesan but for now it's only monitoring software :-( 1124152926 Q * lilo Ping timeout: 480 seconds 1124153038 M * eugenesan I think Project as Control Panel for vserver should be started. There is a demand for that. Maybe i will have invest some of my spare time in it. 1124153065 M * Bertl check out the openvps stuff, IIRC, they aimed to provide something like that 1124153107 M * Bertl and sure, something like that is definitely a good project ... 1124153144 M * eugenesan is there any url for IIRC? 1124153194 M * DaCa dict.org ? 1124153197 M * Bertl hehe, IIRC = If I Remember/Recall Correctly .. but here is your url ... 1124153201 M * Bertl http://www.openvps.org/Plone 1124153266 M * eugenesan :-) 1124153413 M * eugenesan i probably have to sleep, if i get IIRC as project... 1124153428 M * eugenesan good night 1124153463 M * Bertl have a good one eugenesan! 1124153469 M * eugenesan 10x 1124154431 J * stefani ~stefani@c-24-19-46-211.hsd1.wa.comcast.net 1124155828 J * Johnsie ~john@acs-24-154-53-111.zoominternet.net 1124156047 Q * Doener` Quit: Leaving 1124158626 Q * Loki|muh Read error: Connection reset by peer 1124158732 Q * meebey Read error: Connection reset by peer 1124158836 J * Loki|muh loki@satanix.de 1124159498 Q * Johnsie Ping timeout: 480 seconds 1124159882 Q * logger Ping timeout: 480 seconds 1124159903 J * logger ~rs@vds.pas-mal.com 1124159984 Q * dsoul Remote host closed the connection 1124160204 J * dsoul darksoul@vice.ii.uj.edu.pl 1124162606 J * Aiken_ ~james@tooax8-167.dialup.optusnet.com.au 1124162934 Q * Aiken Ping timeout: 480 seconds 1124163941 M * Bertl okay, off to bed now ... have a good whatever everyone! cya tomorrow! 1124163965 N * Bertl Bertl_zZ 1124164000 M * litage can anyone recommend a security consultant/consultancy company in Australia? 1124164351 Q * flock Remote host closed the connection 1124169289 P * stefani parting (is such sweet sorrow) 1124171385 Q * tchan Read error: Operation timed out 1124172241 J * tchan ~tchan@c-24-13-81-164.hsd1.il.comcast.net 1124174617 J * schoobidoo ~drue@p548BA2C1.dip0.t-ipconnect.de 1124176144 M * lownoize good morning 1124176590 Q * monski Quit: Leaving 1124176770 J * Pazzo ~Pazzo@host130-250.pool8172.interbusiness.it 1124177820 Q * OliverA Ping timeout: 480 seconds 1124178379 J * OliverA ~kvirc@ti200710a080-6581.bb.online.no 1124180899 J * Aiken__ ~james@tooax6-071.dialup.optusnet.com.au 1124181228 Q * Aiken_ Ping timeout: 480 seconds 1124181930 J * ata ~ata@gw-swip.djingis.se 1124182032 M * ata anyone awake? 1124182145 M * eyck 0nope 1124182151 M * ata :) 1124182152 M * Hollow kinda 1124182155 M * Hollow ;) 1124182186 M * ata i've set up three vservers, and all is working lovely (i run debian on the host and the vservers) 1124182212 M * ata my only problem is vserver-2-vserver networking. this doesn't seem to work at all 1124182259 M * ata I use tap-devices on the vservers, and host-2-vserver communication is working just fine. I can also connect from a vserver and out 1124182728 M * TheSeer i don't see what you need those devices for? 1124182878 M * schoobidoo I'm using the dummy device for that and evrything is running out of the box. I only created DNS entries for the vservers. 1124183101 M * TheSeer i only use dummy for "internal" vservers 1124183112 M * TheSeer e.g. vservers that are not supposed to be accessable from outside 1124183128 M * TheSeer or if i have to dnat them due to lack of public IPs 1124186204 Q * albeiro Ping timeout: 480 seconds 1124186476 J * erwan_taf ~erwan@konilope.dyndns.org 1124186599 J * albeiro ~albeiro@graffias.estrefa.pl 1124189173 J * DuckMaster ~duckx@195.75.27.158 1124189194 P * md` 1124189339 Q * Aiken__ Ping timeout: 480 seconds 1124189839 Q * albeiro Ping timeout: 480 seconds 1124190062 J * albeiro ~albeiro@graffias.estrefa.pl 1124191842 J * revenger ~joe@bulldog.infosys.de 1124192233 Q * revenger_ Ping timeout: 480 seconds 1124192523 M * ata TheSeer: i removed all tap devices and use dummy-nics, and now everything works just fine. thank you! 1124192532 Q * ata Quit: Leaving 1124195492 M * wibble hmm 1124195513 M * wibble hmm, is there a new ctx patch for vs2.0? 1124195816 Q * erwan_taf Remote host closed the connection 1124196089 J * erwan_taf ~erwan@konilope.dyndns.org 1124196441 M * wibble never mind! I believe its included! 1124196955 Q * Vudumen Ping timeout: 480 seconds 1124197091 J * MrRagga ~Jonas@p54A52C3B.dip0.t-ipconnect.de 1124197152 J * Vudumen vudumen@perverz.hu 1124197308 Q * MrRagga Quit: 1124197311 J * Doener ~doener@p548742B2.dip.t-dialin.net 1124197618 J * lonewolff ~lonewolff@host86-128-17-74.range86-128.btcentralplus.com 1124198290 Q * albeiro Ping timeout: 480 seconds 1124198501 J * albeiro ~albeiro@graffias.estrefa.pl 1124198889 J * monrad ~monrad@213083190134.sonofon.dk 1124198912 J * MrRagga ~Jonas@p54A52C3B.dip0.t-ipconnect.de 1124199518 Q * MrRagga Quit: Verlassend 1124199609 M * wibble hmm. there appears to be an issue with newvserver and vlan interfaces 1124199633 M * wibble when does vbuild configure the network interfaces? 1124199685 M * TheSeer vbuild doesn't do anything 1124199700 M * wibble oh ok 1124199707 M * TheSeer it just write a config file 1124199711 A * wibble re-reads newvserver 1124199729 M * TheSeer the interfcae will be "created" when the guest is started 1124199743 M * wibble ah I see 1124199828 M * TheSeer vserver servername build -m --hostname=servername.mydomain.tld --interface eth0:1.2.3.4/24 -- -d 1124199854 M * wibble yup, doing that but it doesnt config the vlan interface correctly 1124199854 M * TheSeer where could be apt or apt-rpm for a fedora box 1124199883 M * TheSeer define correctly 1124199899 M * wibble ok for example, it creates eth0.1030 1124199901 M * wibble fine 1124199911 M * TheSeer or the lack of it 1124199917 M * wibble creates a 127.0.0.1 address on the interface 1124199923 M * TheSeer erm.. 1124199927 M * TheSeer that'll conflict ;> 1124199930 M * wibble indeed 1124199948 M * wibble unless, it doesn't like not having an IP on eth0 1124199959 M * TheSeer it doesn't 1124199960 M * wibble sorry lemme reword 1124199970 M * TheSeer you need ot have an ip 1124199974 M * TheSeer or no interface at all 1124199980 M * TheSeer you cannot have an interface without ip 1124200002 M * wibble ok, ideally I don't want an IP on the "root" interface where all the vlans are stacked 1124200013 M * TheSeer then use dummy0 1124200090 M * wibble never heard of dummy0 1124200094 A * wibble reads 1124200152 M * TheSeer well.. it's a dummy interface ;> 1124200160 M * wibble obviously ;) 1124200181 M * TheSeer works pretty much like lo 1124200222 A * TheSeer hopes not to get killed for that simplification by network-nerds 1124200254 M * wibble heh 1124200384 M * wibble well what I am trying to achieve is a bunch of vservers on different vlans but with no default route 1124200408 M * wibble although I haven't figured out how I can keep them up to date 1124200410 J * MrRagga ~Jonas@p54A52C3B.dip0.t-ipconnect.de 1124200424 M * wibble maybe via a proxy or local mirror 1124200433 M * TheSeer well.. you can run vapt-get update on them 1124200440 M * TheSeer from the host 1124200461 M * TheSeer and if you use dummy0 to bind the vhosts on using let's say 10.0.0.x as ip range 1124200472 M * TheSeer their ips won't be routable 1124200490 M * TheSeer unless you provide a dnat/snat with iptables for them 1124200509 M * TheSeer or general nat for that matter 1124200516 Q * MrRagga Quit: 1124200611 M * wibble does vapt-get work with context quotas? 1124200817 M * TheSeer guess so.. 1124200845 M * wibble its pretty cool though 1124201023 Q * VooDooMaster Quit: Nettalk6 der Freeware IRC-Client 1124201298 J * pzYsTorM schak@dsl-082-082-148-106.arcor-ip.net 1124201400 M * pzYsTorM morning... i have vs1.9.3.14 for 2.6kernel... where can i get a new one?... this version is sometimes very unstable... and the vserver are shutting down themselves 1124201416 M * wibble vs2.0 is released 1124201422 M * pzYsTorM wow 1124201423 M * wibble stable 1124201427 M * pzYsTorM great stuff 1124201468 M * pzYsTorM but where to find?... on 13thfloor.at there is only old stuff 1124201508 M * wibble http://www.13thfloor.at/vserver/s_rel26/overview/ 1124201521 M * wibble or 1124201523 M * wibble http://www.13thfloor.at/vserver/s_rel26/v2.0/patch-2.6.12.4-vs2.0.diff.gz 1124201549 M * pzYsTorM ah... its in the stable branch... ok 1124201553 M * wibble werks with 2.6.12.5 but just change the makefile 1124201588 M * pzYsTorM ok, great 1124201659 M * pzYsTorM is there a feature list of the changes/added stuff in relation to the 1.9.x ? 1124201694 M * wibble http://linux-vserver.org/index.php?page=Release+FAQ 1124201750 M * pzYsTorM thx 1124202109 J * keyser_soze ~cimarron@host197.201-252-9.telecom.net.ar 1124202122 M * pzYsTorM do someone know a script to install sarge (debian) to a new vserver?... i have debian-newvserver.sh, but this only works with woody and produces some errors if i use it for sarge 1124202141 M * wibble host is sarge right? 1124202146 M * pzYsTorM yes 1124202154 M * wibble ok i have some backported scripts that might help 1124202159 M * wibble .deb rather 1124202202 M * pzYsTorM cool, could you send/upload/mail/icq me these? 1124202255 M * wibble http://www.wibblewobb1e.co.uk/vserver/ 1124202270 M * wibble you need iproute, vlan etc 1124202319 M * wibble dont know if you doods want to host them permanently somewhere, or just link to mine 1124202326 M * wibble well the ones provided by zobel 1124202667 N * Bertl_zZ Bertl 1124202685 M * Bertl morning folks! 1124202690 M * wibble hello! 1124202698 M * Bertl pzYsTorM: is something wrong with util-vserver build? 1124202763 M * Bertl pzYsTorM: http://linux-vserver.org/alpha+util-vserver (see Vserver creation) 1124202883 M * pzYsTorM wibble: thx for the debs! 1124202911 M * pzYsTorM Bertl: i havent spent much time to this util-vserver-stuff... because i had this script... 1124202933 M * wibble hmm 1124202935 M * wibble its still borked 1124202973 M * wibble it creates the vlan interface but gives it an ip 127.0.0.1 1124203051 M * Bertl util-vserver build is borked? 1124203060 M * wibble not sure 1124203079 M * Bertl well, what is your build command? 1124203087 M * wibble vserver testsrv build -m debootstrap --hostname test.uksolutions.net --interface eth0.1031:10.0.31.254/24 -- -d sarge 1124203182 M * Bertl trying now ... 1124203191 M * wibble eth0.1031 Link encap:Ethernet HWaddr 00:11:D8:BA:54:C6 1124203191 M * wibble inet addr:127.0.0.1 Bcast:127.255.255.255 Mask:255.0.0.0 1124203614 M * wibble ifdown: shutdown eth0.1031: Permission denied 1124203614 M * wibble ERROR: trying to remove VLAN -:eth0:- error: Operation not permitted 1124203615 M * wibble hmm 1124203650 M * wibble removes the ip though, but doesnt take down the vlan interface 1124203694 M * wibble removing it by hand 1124203695 M * wibble Removed VLAN -:eth0.1031:- 1124203703 M * wibble so it looks like its trying to take down the interface? 1124203711 M * wibble physical interface rather 1124204162 M * Bertl okay, yes, looks like a bug, did you file a bug report on savannah? 1124204176 M * wibble no, dont even know what savannah is! 1124204226 M * Bertl http://savannah.nongnu.org/projects/util-vserver 1124204253 M * wibble ok, going through the process now 1124204272 M * Bertl don't bother, I file it for you ... 1124204277 M * wibble ah ok 1124204289 M * wibble just mention my name !! :) 1124204294 M * wibble first bug i ever found lol 1124204306 M * Bertl okay, then go ahead, file it yourself ... 1124204334 M * Bertl just file it for the 0.30.208 utils ... 1124204370 M * wibble doh 1124204375 M * wibble im running 0.30.207-11 1124204383 M * wibble shouldn't make any difference? 1124204406 M * Bertl don't file it as 0.30.207-11, the result will be, debian brokenness and the bug will get ignored 1124204413 M * wibble i c 1124204422 M * Bertl (get the mainline version, 0.30.208 is latest IIRC) 1124204438 M * Bertl try with that, it fails in the same way (I'm using it) 1124204456 M * wibble ok cool. The debs provided were 207-11 and backported (which i dont know how to do) 1124204464 M * wibble im not really a linux engineer, but im learning 1124204472 M * Bertl vserver test103 start 1124204472 M * Bertl Added VLAN with VID == 1031 to IF -:eth0:- 1124204488 M * Bertl there are debs for 208 too ... but in general it's better to use mainline 1124204494 M * wibble im sure this worked with the old utils 1124204502 M * Bertl ifconfig 1124204511 M * Bertl eth0.1031 Link encap:Ethernet HWaddr 00:B0:D0:68:8D:28 1124204511 M * Bertl inet addr:127.0.0.1 Bcast:127.255.255.255 Mask:255.0.0.0 1124204527 M * Bertl build was your line except the name which I changed to test103 1124204541 M * wibble yup 1124204559 M * wibble i have a sneaking feeling that this worked with the old debian utils - when using 1.9.5 1124204576 M * Bertl ERROR: trying to remove VLAN -:eth0:- error: Operation not permitted 1124204599 M * Bertl and the vlan interface remains after shutdown ... 1124204603 M * wibble yup 1124204611 M * wibble exactly the same thing 1124204644 M * Bertl okay, so you are going to file it, right? (just making sure that it doesn't get lost) 1124204661 M * wibble yes, I am just waiting for my login details! 1124204674 M * Bertl perfect, TIA! 1124204684 J * MrRagga ~Jonas@p54A52C3B.dip0.t-ipconnect.de 1124204688 M * wibble well I just got to find a work around now 1124204753 M * wibble i might have to go back to the old utils, but then I loose some features I think 1124204869 Q * Greek0 Quit: leaving 1124205031 J * Greek0 ~greek0@81.189.246.175 1124205055 M * Bertl wibble: you can easily work around this 1124205080 M * Bertl just do not specify the interface at all (for now) and create the interfac with ip by hand 1124205096 M * Bertl specify the ip to the guest with the nodev option ... 1124205165 M * Bertl okay, I'm off now ... back later this evening ... 1124205171 N * Bertl Bertl_oO 1124205311 Q * MrRagga Quit: Verlassend 1124205637 M * wibble bug id 14148 :) 1124205649 M * wibble hope I didnt give too much/too little information 1124205945 Q * lownoize Ping timeout: 480 seconds 1124206564 J * lownoize ~lownoize@p54ACAADD.dip0.t-ipconnect.de 1124206934 M * Hollow gentoo users: a nice bunch of new scripts: http://phpfi.com/74348 1124206958 M * Hollow if you want to test: http://home.xnull.de/work/gentoo/vserver/tools 1124208197 M * pzYsTorM is there a (simple) possibility of calculating the incoming and outgoing traffic of each vserver? 1124208272 M * wibble iptables 1124208463 M * wibble hmm minor issue 1124208464 M * wibble CTX PROC VSZ RSS userTIME sysTIME UPTIME NAME 1124208464 M * wibble 0 42 60.3M 4.8K 1m18s10 0m20s39 1h52m11 root server 1124208464 M * wibble 49154 5 11M 968 0m00s00 0m00s00 0m05s17 test12 1124208474 M * wibble didnt they used to have [] round the name? 1124208491 M * wibble 49153 6 16.5M 1.1K 3m54s44 1m34s23 4d59h41 [ftp] 1124208491 M * wibble 49154 42 2G 239.8K 1h05m48 4m48s53 4d59h41 [support] 1124208597 M * pzYsTorM iptables... hmm hmm... well... but: works only, if each vserver has its unique ip, doesnt it? 1124208621 M * wibble surely they do have their own unique IP? 1124208687 M * daniel_hozac [] is just for legacy guests. 1124208718 M * wibble ah ok 1124208833 M * pzYsTorM yes, they should have their own ip... but they dont have to 1124208872 N * Bertl_oO Bertl 1124208876 M * Bertl short visit ... 1124208897 M * Bertl pzYsTorM: how would you tell which incoming packet will hit what guest then? 1124209064 M * Bertl wibble: is the [] stuff an extension? 1124209133 M * pzYsTorM dont know :) ... may be it doesnt work... but... one day i forgot to add this /etc/vserver/bla/interfaces/0/ip file and every serversocket on the vserver was opened with the ip of the _host_ 1124209169 M * Bertl yes, it works from the guest/kernel point of view 1124209180 M * Bertl you can easily assing a single ip to three guests 1124209199 M * Bertl they will share the ip, but you have no way to tell who will be able to bind to what port 1124209215 M * Bertl the first one binding will succeed 1124209224 M * pzYsTorM yes, first come first serve ;) 1124209247 M * Bertl but what I mean, from the accounting point of view that's the pure horror, no? 1124209282 J * mikmu mikmu@h64-5-199-35.gtcust.grouptelecom.net 1124209303 M * Bertl welcome mikmu! 1124209348 M * pzYsTorM yes of course... and i dont want to think about it... because in my case, all vservers have their own ip ;) 1124209454 M * Bertl if you really want to 'share' a single ip, you usually decide which guest will be allowed to use what ports at the administrative level 1124209489 M * Bertl and if you do that, you are easier off with assigning a separate private ip to each guest, and s/DNATing the ports/addresses with iptables 1124209515 M * Bertl which will automatically allow the traffic to be accounted properly ... 1124209886 M * wibble is this a new thing: 1124209887 M * wibble /dev/hdv1 72G 1.2G 67G 2% / 1124209887 M * wibble none 16M 0 16M 0% /tmp 1124209897 M * wibble having a 16mb /tmp ?| 1124209952 M * daniel_hozac yeah, that's default. 1124210024 M * Bertl but you can change/adjust it ... 1124210127 M * wibble ah I see. I am so used to using newvserver 1124210136 M * wibble seams to just work for vs1.12 1124210140 M * wibble 1.2 rather 1124210188 M * wibble what I am basically trying to do is create a number of small debian sarge vservers, each bound to their own vlan interface 1124210200 M * wibble with context quotas, just for good fun 1124210258 M * Bertl per context disk limits work with 2.0, per context quota on a shared partition do not work yet ... 1124210287 M * Hollow hey Bertl 1124210295 M * wibble ok I keep getting confused with disk limits v quota 1124210320 M * wibble what I want to do is over commit a disk 1124210333 M * Hollow Bertl: what about scheduling my vienna trip this or next week? 1124210414 M * Bertl sounds good, should be around this weekend, but not sunday ... 1124210434 M * Hollow saturday would be fine? 1124210457 M * Bertl yes, guess so ... 1124210488 M * Hollow ok, and we'll keep the schnitzelwrit as meeting point? 1124210815 M * wibble /usr/sbin/vserver: line 79: 1360 Killed "${NICE_CMD[@]}" ${USE_VNAMESPACE:+$_VNAMESPACE --enter "$S_CONTEXT" -- } $_VCONTEXT $SILENT_OPT --migrate --chroot --xid "$S_CONTEXT" -- "${INITCMD_STOP[@]}" 1124210819 M * wibble hmm 1124210820 M * wibble that don't look right either 1124210841 M * mikmu Before I start compiling everything, the latest vserver is for 2.6.12.4, right? 1124210854 M * wibble werks for 2.6.12-5 1124210863 M * wibble well, apart from the bug i found 1124210878 M * wibble but it might be good to go for -4 :) 1124210900 M * mikmu wibble: serious bug? 1124210913 M * wibble issue with vlan interfaces being created/deleted - not that serious 1124210916 M * Bertl mikmu: not kernel related ... 1124210927 M * mikmu ok, greak 1124210937 M * wibble well in fact, not really serious at all 1124210939 M * wibble minor ! 1124210960 M * wibble i prolly should go now, i am sure I am getting on your nerves lol 1124211013 M * mikmu I updated from debian woody to debian sarge last night, and now for some reason, one of my vservers is binding port 25 from the main host instead of it's own. I have postfix on .56 with is the vserver, and exim on .58, the main host, when I telnet to port 25 on .56, I get exim instead of postfix 1124211020 M * mikmu Not quite sure what's going on 1124211090 M * wibble postfix & exim in vservers 1124211092 M * wibble correct? 1124211097 M * wibble what about exim on the host vserver? 1124211122 M * mikmu no, justpostfix in a vserver, exim on the host. When I telnet into to vserver, I get exim from the host 1124211132 M * wibble netstat -an on the host? 1124211138 M * wibble see if it is binding to 0.0.0.0 ? 1124211150 M * Bertl mikmu: the solution is simple: 1124211172 M * Bertl sarge, for some reason, installs the exim4 mailer (regardless if you want or not) 1124211188 M * Bertl so you now have a new exim4, which binds to all ports (on the host) 1124211193 M * pzYsTorM solution: MY_IP = 12.12.12.12 in exim4.conf 1124211203 M * Bertl precisely ... 1124211211 M * wibble ok well I kind of got there 1124211213 M * mikmu Great 1124211228 M * mikmu makes sense 1124211406 M * mikmu hmm, I imagine I should look into upgrading the vservsers to sarge too 1124211804 M * wibble actually 1124211810 M * wibble i think my vserver is broken 1124211816 M * wibble no interfaces come up now! 1124212451 Q * eugenesan Ping timeout: 480 seconds 1124212645 M * Vudumen Bertl: moon is up again 1124212649 M * Vudumen sorry for the late 1124213336 Q * erwan_taf Remote host closed the connection 1124214510 Q * schoobidoo Remote host closed the connection 1124215717 Q * cryo Ping timeout: 480 seconds 1124216188 Q * keyser_soze Quit: Abandonando 1124216364 J * stefani ~stefani@c-24-19-46-211.hsd1.wa.comcast.net 1124216779 J * W4RrI0R W4RrI0R@133-165.tr.cgocable.ca 1124216861 P * W4RrI0R 1124217222 P * mikmu 1124218566 M * pzYsTorM wibble: can you help me with the Makefile.rej ? 1124218909 J * renihs ~renihs___@193.170.52.70 1124220057 J * eugenesan eugenesan@bzq-219-189-212.cablep.bezeqint.net 1124220623 J * monski ~monrad@213083190134.sonofon.dk 1124220710 Q * monrad Ping timeout: 480 seconds 1124220740 Q * albeiro helium.oftc.net oxygen.oftc.net 1124220740 Q * Doener helium.oftc.net oxygen.oftc.net 1124220740 Q * Pazzo helium.oftc.net oxygen.oftc.net 1124220740 Q * brc helium.oftc.net oxygen.oftc.net 1124220740 Q * Beave helium.oftc.net oxygen.oftc.net 1124220740 Q * hetche helium.oftc.net oxygen.oftc.net 1124221327 J * Pazzo ~Pazzo@host130-250.pool8172.interbusiness.it 1124221328 J * Doener ~doener@p548742B2.dip.t-dialin.net 1124221439 J * albeiro ~albeiro@graffias.estrefa.pl 1124221473 J * brc bruce@201008071140.user.veloxzone.com.br 1124221747 Q * lownoize Remote host closed the connection 1124222121 J * Beave ~beave@vistech.org 1124222187 J * lownoize ~lownoize@p54ACAADD.dip0.t-ipconnect.de 1124222214 M * lownoize Bertl, fixed the userspace nfs server problem in my vserver 1124222256 M * lownoize was a problem with unfsd ;( now i use the one from the debian project with the patches from debian and everything looks fine, no unkown error 524 1124222329 M * lownoize so it was not a vserver problem ;) 1124222408 M * eyck good. 1124222494 M * lownoize so i recomend to anybody who wants a nfs server in his vserver to use the debian version with their patches 1124222620 M * lownoize thanks for this nice piece of software, and the good support here ;) 1124222623 A * lownoize is happy 1124223058 M * Hollow Bertl: does the ctx_rlimit.id field accept the values from resource.h too, or only the VLIMIT ones from vserver/limit.h? 1124223289 J * stephenM ~steve@83.151.227.52 1124223294 M * stephenM evening all 1124225190 Q * monski Ping timeout: 480 seconds 1124225394 J * monski ~monrad@213083190134.sonofon.dk 1124225615 M * _ag_ Bertl: is there any plans for ngnet now 2.0 is released? 1124226045 M * stephenM !seen Bertl 1124226093 Q * Doener Quit: Leaving 1124226229 J * Doener ~doener@p548742B2.dip.t-dialin.net 1124226304 Q * mcp Read error: Connection reset by peer 1124226305 J * _mcp ~hightower@wolk-project.de 1124226490 N * _mcp mcp 1124226679 J * ntrs ~ntrs@Dardeene-68.188.50.87.charter-stl.com 1124227276 M * Bertl hmm, back now ... 1124227369 M * Bertl _ag_: yes there are! 1124227401 M * Bertl Hollow: the vlimit ones are _extensions_ to the RLIMITS ... 1124227446 M * Bertl lownoize: thanks for the feedback .. basically 99% of all issues (brought here) are not vserver related and I really like it :) 1124227454 M * Hollow Bertl: i.e. i use the ones from resource.h and vserver/limit.h in set_rlimit? 1124227467 M * albeiro hey Bertl :) ! 1124227471 M * albeiro hey Hollow ! 1124227475 M * Hollow hey albeiro 1124227497 M * albeiro long time not seen. how's going ? 1124227511 M * Bertl okay, short dinner break, back in 20 minutes ... 1124227562 N * Bertl Bertl_oO 1124227597 Q * stupidawy Ping timeout: 480 seconds 1124227608 J * stupidawy foo@198.77.239.131 1124228097 M * stephenM anybody know how I impose cpu timelimits on vservers? 1124228114 M * stephenM for example a vserver can burst for a few seconds, say 5, and then is punished thereafter 1124228136 M * Hollow stephenM: look at the hard cpu scheduler (aka token bucket) 1124228182 M * Hollow http://linux-vserver.org/Linux-VServer-Paper-06 section 06.3 1124228205 M * stephenM thanks 1124228443 M * stephenM uhm 1124228444 M * stephenM See Example [XX] 1124228453 M * stephenM could you give me an example pls? 1124228680 J * Aiken ~james@tooax8-013.dialup.optusnet.com.au 1124228732 N * Bertl_oO Bertl 1124228742 M * Bertl okay, finally back now ... 1124228804 M * Bertl stephenM: you have an example in the default ... 1124228827 M * Bertl if you enable the hard CPU scheduler the guests will default to 1124228901 M * Bertl sched->fill_rate = 1; 1124228901 M * Bertl sched->interval = 4; 1124228901 M * Bertl sched->tokens_min = HZ >> 4; 1124228901 M * Bertl sched->tokens_max = HZ >> 1; 1124229033 M * Hollow Bertl: are the state change defines meant to be used in user-space? 1124229054 M * Bertl example/file? 1124229069 M * Hollow vserver/context.h VSC_STARTUP 1124229110 M * Hollow well, they are enums.. 1124229113 M * Hollow not defines 1124229128 M * Bertl yes 1124229180 M * Bertl long vs_state_change(struct vx_info *vxi, unsigned int cmd) 1124229186 M * Bertl snprintf(cmd_buf, sizeof(cmd_buf)-1, "VS_CMD=%08x", cmd); 1124229205 M * Bertl so cmd gets passed into userspace ... 1124229270 M * Hollow i mean, if a app like libvserver or util-vserver should use VSC_* 1124229298 M * Bertl well, the 'helper' app probably could make good use of it ... 1124229301 M * Hollow because util-vserver doesn't use them, and i'm currently adding defines to vserver.h so we don't depend on the kernel headers 1124229359 M * Bertl look, the files in include/linux/vserver _are_ dual use headers, so you can and should include them in userspace apps 1124229404 M * Bertl you do not need to use the ones the kernel brings, make your 'snapshot' copy of vs2.0 and use that one, until you need new ones ... 1124229408 M * stephenM Bertl: could you give me an example command to change those values/ 1124229411 M * stephenM *pls? 1124229420 M * Hollow hm.. ok 1124229466 M * Hollow an what about e.g. CAP_ defines from linux/capabilities.h? 1124229469 M * Bertl stephenM: vsched --xid 42 --fill-rate 10 1124229485 M * stephenM thanks 1124229491 M * Bertl Hollow: those should be part of the glibc headers 1124229510 M * Bertl (ultimately the linux-vserver headers would also be part of glibc headers :) 1124229517 M * stephenM does the vserver have to be stopped to change those values Bertl? 1124229544 M * Bertl stephenM: no, only a few things require the guest to be stopped, most changes can be done on the fly 1124229549 M * Hollow ah, i see.. there is an sys/capability.h too.. 1124229622 M * Bertl Hollow: basically you should find everything you need in recent glibc + the linux/vserver dir ... of course, some 'older' libc headers might require things like the personality or cpu affinity stuff to be defined elsewhere ... 1124229655 M * Hollow ok, thx 1124229659 M * Bertl yw 1124229747 M * stephenM Bertl: is --tokens the command to set the "bucket size"? 1124229813 M * Bertl no, tokens sets the 'current' amount of tokens 1124229828 M * Bertl what you are looking for is the token-max and min 1124229864 M * stephenM right, so if I wanted to give a vserver ~20 seconds of full CPU power, and then throttle them for a further 20 seconds, how would I do that? 1124229878 M * stephenM sorry for all the questions but this token stuff really confuses the hell out of me :P 1124230078 M * Bertl np, did you have a look at the TB description yet? 1124230091 M * stephenM yeah 1124230098 M * stephenM I kind of understand it, but... 1124230105 M * stephenM kind of don't :P 1124230107 M * Bertl so you understand how it works .. good ... 1124230115 M * stephenM yeah I get the basic principle 1124230128 M * Bertl let's make a few considerations here ... 1124230147 M * stephenM but I don't understand how to allow a vserver to run for 20 seconds with full cpu, and then get punished for the next 20 seconds, after which time it would be reset 1124230161 M * Bertl assuming that we have HZ=1000, how many jiffies will pass by in 1 second? 1124230167 M * stephenM uh, 10? 1124230188 M * Bertl hmm ... no, actually 1000 each second ... 1124230191 M * stephenM ah ok 1124230212 M * Bertl so let's further assume we consume one token each second (for a running process) 1124230217 M * stephenM yep 1124230231 M * Bertl how many tokens would it need to run for 20 seconds straight? 1124230238 M * stephenM 20 1124230245 M * stephenM 20 tokens, 1 per second right? 1124230256 M * Bertl okay, and if it consumes a token each jiffie? 1124230273 M * stephenM no idea, sorry 1124230284 M * stephenM I don't know how to relate tokens and jiffies 1124230311 M * Bertl hmm, you've been bad with math? 1124230322 M * Bertl np, we will fix that :) 1124230354 M * Bertl 1000 HZ -> 1000 jiffies each second -> 1000 tokens each second 1124230361 M * stephenM right 1124230369 M * stephenM got it 1124230375 M * Bertl so for 20 seconds straight, we need 20.000 tokens 1124230383 M * stephenM ah, right 1124230415 M * Bertl now it would sound reasonable to set the max to 20.000 but that would probably be wrong :) 1124230431 M * Bertl do you have any idea why? 1124230442 M * stephenM because it doesn't relate to cpu used? 1124230447 M * stephenM i.e. it could just be idling 1124230451 M * stephenM not using excessive cpu time 1124230451 P * stefani parting (is such sweet sorrow) 1124230466 M * stephenM or does an idle process not use tokens 1124230479 M * Bertl yeah, well, that's a different story .. if it is idle, it's not using tokens, but it's not running either ... 1124230493 M * Bertl so that's not the thing you want to limit (as burst, no?) 1124230519 M * stephenM I just want to allow a vserver to max out the cpu for 20 seconds, and then get punished for a further 20 seconds after that 1124230524 M * Bertl we started out with '20 seconds full cpu' 1124230527 M * stephenM yep 1124230542 M * Bertl so it's not idling when it's using full cpu, is it? 1124230545 M * stephenM nop 1124230549 M * stephenM certainly not 1124230561 M * Bertl okay, but we forgot the refill ... 1124230583 M * stephenM refill? 1124230592 M * stephenM so when they use their tokens, the bucket is empty 1124230594 M * Bertl the tokens usually will be 'refilled' (you read the TB description, sure?) 1124230601 M * stephenM yeah 1124230603 M * stephenM I get you 1124230615 M * stephenM but only after 20 seconds, I guess? 1124230625 M * stephenM so once it uses all its tokens, it will wait 20 seconds before resetting the bucket 1124230632 M * Bertl no, they will be refilled regardless of any state ... 1124230636 M * stephenM ah 1124230642 M * stephenM so how do I punish the vserver for X seconds? 1124230646 M * stephenM before allowing it to max the cpu again 1124230676 Q * ntrs Remote host closed the connection 1124230687 M * Bertl basically the token bucket has 3 states ... 1124230714 M * Bertl well, actually 4 but the first one is not different in behaviour 1124230721 M * Bertl from the second one ... 1124230757 M * Bertl 1/2 Token bucket is full, and token bucket is not empty ... 1124230777 M * Bertl basically you start with a full (filled to max) bucket 1124230791 M * stephenM yep 1124230797 M * Bertl when the processes are consuming tokens, the buckets starts to get emptier ... 1124230801 M * stephenM yep 1124230818 M * Bertl at a certain point, the bucket is empty, and no process can be run anymore ... 1124230824 M * stephenM yep 1124230834 M * Bertl this is 3) token bucket recovery ... 1124230845 M * stephenM right 1124230849 M * Bertl the token bucket has to fill up until the min is reached 1124230858 M * Bertl before that, no process will be run ... 1124230882 M * Bertl once the min is reached, processes might use up tokens again 1124230887 M * stephenM right 1124230900 M * stephenM so is there an interval I can set between the bucket becoming empty and it being refilled? 1124230904 M * Bertl so let's get back to the example with 20 seconds cpu burst 1124230908 M * stephenM ok 1124230936 M * Bertl we already calculated that we need 20.000 tokens to run for 20 seconds straight (at HZ=1000) 1124230940 M * stephenM yep 1124230957 M * Bertl now let's take a look at what happens after that ... 1124230973 M * Bertl the token bucket is empty, and processes are not scheduled ... 1124230993 M * stephenM yep 1124231002 M * Bertl for example, we would like to allow only 1/10th of the cpu resources to the context now 1124231008 M * stephenM yep 1124231021 M * stephenM I still want the processes to be able to run, but veeery slowly 1124231027 M * Bertl so we decide to 'refill' 1 token each 10 jiffies 1124231039 M * stephenM ahh, I see now! 1124231052 M * Bertl or 2 tokens each 20 jiffies, or 10 tokens each ?? jiffies? 1124231063 M * stephenM yep 1124231065 M * stephenM I get you 1124231083 M * Bertl okay, so let's assume we decide to go for 10 tokens each 100 jiffies 1124231093 M * Bertl (would be rate = 10, interval = 100) 1124231096 M * stephenM yep 1124231111 M * Bertl but this would also refill while we are bursting ... 1124231116 M * stephenM so it's not calculated based on MHZ, but by jiffies (cputime)? 1124231126 M * Bertl so we have to compensate on this ... 1124231144 M * Bertl (no it's not based on MHz because that's not really a measure for CPU resources :) 1124231148 M * stephenM ah ok 1124231173 M * Bertl assumed we have 20 seconds bursting, a rate of 10 and interval of 100, how many tokens would get refilled? 1124231200 M * stephenM 2000 1124231224 M * Bertl excellent 1124231236 M * Bertl so we have to set the maximum to 20.000 - 2.000 1124231244 M * stephenM right 1124231254 M * Bertl and we can decide to leave the minimum at 0 for example 1124231271 M * Bertl so now you've got your 4 magic values together 1124231282 M * stephenM the only option I don't understand is the --token argument 1124231289 M * stephenM tokens even 1124231290 M * Bertl max = 18.000, min = 0, rate = 10 and interval = 100 1124231300 M * Bertl the token value is basically your start value ... 1124231305 M * stephenM ah I get yiou 1124231306 M * stephenM you* 1124231314 M * Bertl (it's not very useful, unless you want to test something) 1124231327 M * stephenM [root@vs1 ~]# vsched --xid 1000 --fill-rate 10 --interval 100 --tokens-max 18000 --tokens-min 0 --tokens 20000 1124231327 M * stephenM [root@vs1 ~]# 1124231345 M * Hollow Bertl: very insightful, thx :) 1124231356 M * Bertl yw 1124231368 M * stephenM yeah thanks a lot Bertl, I really didn't think I could understand this but you have a very good way of explaining it :) 1124231378 A * Hollow nods 1124231387 M * Bertl thanks a lot ... 1124231477 M * stephenM hmm, I've applied that sched info to the vserver, but still my cpu-intensive bittorrent process is still using 90% cpu after 20 seconds, is that normal? 1124231499 M * Bertl did you enable the hard cpu scheduler? 1124231529 M * stephenM ah, possibly not 1124231530 M * Bertl kernel config and you need the sched_hard flag to be set 1124231554 M * stephenM CONFIG_VSERVER_HARDCPU=y 1124231565 M * stephenM where do I set that flag? 1124231593 M * Hollow Bertl: btw, any progress on the start/stop howto? 1124231602 M * Bertl stephenM: vattribute --set --xid 42 --flag sched_hard 1124231612 M * stephenM ty 1124231635 M * Bertl Hollow: no, I somehow managed to forget it again ... 1124231648 M * Hollow maybe we can talk about it on saturday? 1124231658 M * Bertl yeah, probably a good idea ... 1124231658 M * stephenM LOL now it's working 1124231661 M * Hollow good 1124231728 M * Hollow Bertl: you may want to think about your ideas for the future utils, too 1124231750 M * Hollow like to start them asap 1124231756 M * Bertl will do so ... great! 1124231804 M * Hollow fine, so i'll take a look at my mattress now.. good night everyone! 1124231833 M * Bertl night Hollow! 1124231883 M * stephenM Bertl: if a process continually tries to use excessive CPU, I guess that it will basically never be refilled to 20000 again? 1124231892 M * stephenM because it will be eating tokens faster than they are being refilled 1124231899 M * Bertl exactly ... 1124231909 M * stephenM is it possible to reset it after, say 5 minutes? 1124231917 M * stephenM regardless of current usage 1124231923 M * stephenM or should I just increase the refill rate 1124231927 M * Bertl what would be the purpose? 1124231934 M * Bertl (of the reset) 1124231936 M * stephenM to give them a chance with full cpu again :P 1124231953 M * stephenM otherwise they will be punished permanently, but I guess that's only because they are trying to use a lot of cpu 1124231968 M * Bertl would not improve anything, would it? but you could use a cronjob or similar to set the token value back to max 1124231985 M * stephenM mm 1124232298 M * stephenM oh btw, I wanted to ask - why is the default policy of vserver to block /proc for guests? is it some kind of security risk? 1124232320 M * Bertl yes, some proc entries are not properly secured 1124232347 M * stephenM can we block the bad ones? because it's really totally impossible to run a vserver without process info at least 1124232356 M * stephenM but at the same time, I don't want to compromise security 1124232369 M * Bertl that's what vprocunhide is doing, no? 1124232374 M * stephenM ah I see 1124232380 M * stephenM so it only unblocks the safe entries? 1124232430 M * Bertl yep, that's the idea :) 1124232442 M * stephenM ah right 1124232474 M * stephenM also, is it possible to restrict RAM usage in the same way as we've done with cpu? 1124232506 M * Bertl hmm, not really, but I guess a token bucket would not fit the ram usage scenario very well ... 1124232525 M * stephenM what's the equivelant? 1124232530 M * Bertl we have resource limits for VM (AS) and RSS 1124232542 M * Bertl (as well as the mostly unused VML) 1124232550 M * stephenM is it possible to limit a user to 500MB of RAM/swap? 1124232553 M * stephenM vserver rather 1124232576 M * albeiro VML ? 1124232580 M * Bertl VM (AS) is the adress space, which can be somewhat compared to Swap + RAM 1124232603 M * Bertl RSS is the in memory pages, so its the actually _used_ ram 1124232617 M * stephenM right 1124232643 M * Bertl but we do not force paging out yet ... 1124232654 M * Bertl (and I'm not convinced that it would be a good idea) 1124232678 M * stephenM so it's not possible to restrict a user to a certain amount of ram space? 1124232759 M * Bertl it is ... even stricter than on any real system 1124232765 M * stephenM oh, cool 1124232769 M * stephenM how? :P 1124232794 M * Bertl setting the RSS to 64MB for example, will not permit more than 64MB of RAM to be allocated by the guest 1124232822 M * Bertl trying to do so, will result in OOM ... 1124232838 M * stephenM that sounds bad :P 1124232845 M * stephenM how do I set that limit for a guest? 1124232849 M * albeiro that sounds correctly 1124232888 M * albeiro Bertl: why it is even better than on any real system ? say, i have set user rss limit, how does it compare ? 1124232916 M * Bertl stephenM: vlimit --xid 42 -H --rss ... 1124232936 M * stephenM thanks 1124232943 M * stephenM --rss's argument would be M? 1124232945 M * Bertl albeiro: you are right, with user rss set, it's identical ... 1124232945 M * stephenM like, 512M? 1124232980 M * Bertl albeiro: except for the small detail that we do not allow forks when over limit 1124233008 M * Bertl stephenM: last time I checked it was in pages ... (and a page is 4k on x86) 1124233071 M * stephenM for 512M then, I guess it would be 131072 pages? 1124233089 M * Bertl yep, exactly ... 1124233107 M * stephenM ok thanks 1124233121 M * Bertl you're welcome! 1124233201 M * stephenM oh also, with multi-processor systems, does vserver 'work' with them? I mean, are processes still equally shared across the processors? 1124233211 M * stephenM or is that not something vserver gets involved with 1124233303 M * Bertl SMP is working fine and is very well tested with linux-vserver 1124233329 M * Bertl but you have to 'adjust' your calculations if you allow guests to use more than one cpu ... 1124233395 M * stephenM so they aren't allowed to use both cpus by default? 1124234009 Q * albeiro Ping timeout: 480 seconds 1124234267 M * lownoize re 1124234314 J * albeiro ~albeiro@graffias.estrefa.pl 1124234795 J * tanjix tanjix@office.star-hosting.de 1124234801 M * tanjix hello together 1124234907 M * stephenM 'hello together'? 1124234938 M * lownoize stephenM, here my vservers use both cpu's by default 1124235102 J * ntrs ~ntrs@Dardeene-68.188.50.87.charter-stl.com 1124235109 M * Bertl hey tanjix! lownoize! 1124235260 M * lownoize works so nice I'm so happy, and my parents will be happy too because only one server not 3 anymore ;) 1124235323 M * Bertl great, time for the Happy Users page? 1124235547 M * stephenM hmm, just out of interest waht is the difference between the vserver and UML projects? 1124235551 M * stephenM are they identical? 1124235577 M * Bertl UML allows you to run a linux kernel in userspace 1124235586 M * lownoize uml is more like xen every vm has it's own kernel 1124235605 M * lownoize vserver is more like bsd jails the vm's share one kernel 1124235625 M * stephenM ahh right 1124235640 M * stephenM which one is better? :P 1124235648 M * stephenM pretty daft question! lol 1124235651 M * lownoize i tested all 1124235652 M * Bertl depends on what you want to do ... 1124235679 M * lownoize performance in vserver was the best, on my old dual p3 500 1124235722 M * Bertl not talking of performance the main differences are probably: 1124235753 M * Bertl Xen and UML are basically working with virtual machines, while linux-vserver is absed on sharing resources 1124235771 M * stephenM I guess with virtual machines, the overhead is higher 1124235781 M * lownoize yes 1124235788 M * lownoize but xen is very fast to 1124235790 M * Bertl Xen is working on x86, somewhat on x86_64 and ppc, linux-vserver and UML support most kernel archs (basically all of them) 1124235816 M * lownoize and with xen you have the ability to run netbsd .... 1124235824 M * lownoize but for my needs vserver is the right thing 1124235854 M * Bertl in speed comparison probably they are linux-vserver >= Xen >= UML 1124235887 M * stephenM oh also, how do I get user passes in /etc/shadow instead of /etc/passwd? 1124235889 Q * Snow-Man Remote host closed the connection 1124235896 M * Bertl from the virtualization point of view I'd sort them (and QEMU) like this: 1124235896 M * stephenM that seems to be the default 1124235914 M * Bertl QEMU >= UML >= Xen >= Linux-Vserver 1124235948 J * Snow-Man ~sfrost@snowman.net 1124235957 M * daniel_hozac how does QEMU fare against UML in performance comparisons? 1124235974 M * lownoize stephenM, you have passwords in passwd? must be a very old distribution or a strange setup 1124235981 M * stephenM lownoize: it's centos 4 1124236003 M * stephenM it defaults to /etc/shadow in the main host, but in guests it all goes in /etc/passwd 1124236014 M * lownoize I'm a happy slackware user 1124236018 M * daniel_hozac stephenM: /etc/pam.d/system-auth, add shadow where appropriate. 1124236147 M * Bertl daniel_hozac: QEMU is between 3-8 times slower than the real thing 1124236266 Q * Snow-Man Remote host closed the connection