1120262666 J * Aiken ~james@tooax6-143.dialup.optusnet.com.au 1120262881 Q * monrad Ping timeout: 480 seconds 1120263787 J * rs ~rs@imhotep.rhapsodyk.net 1120263811 Q * rs Quit: 1120265013 J * monrad ~monrad@213083190130.sonofon.dk 1120265575 J * badiane ~chatzilla@user-0ccerh5.cable.mindspring.com 1120265586 M * badiane ehlo everyone 1120265658 M * badiane I have a question concerning setting routing, rules with tables 1120265690 M * badiane is anyone familiar 1120266586 Q * Aiken Quit: Leaving 1120266615 J * Aiken ~james@tooax6-143.dialup.optusnet.com.au 1120268938 M * kestrel is there a tarball somewhere of an example config directory for the 2.x/1.9x util-vserver stuff? 1120269080 M * daniel_hozac vserver example build -m skeleton 1120269258 M * kestrel yeah i've done that 1120269264 M * kestrel i have a vserver up and running 1120269271 M * kestrel i'm just looking for examples of the rest of the options 1120269521 M * daniel_hozac so, use the options. 1120269530 M * daniel_hozac examine /etc/vservers/example after building. 1120269538 M * daniel_hozac or check the flower page. 1120270740 M * kestrel that is what i'm doing 1120270783 M * kestrel i just want verification that i'm doing it right..the flower page is good, but often doesn't fully document the options 1120270787 A * kestrel shrugs 1120271026 P * daniel_hozac Oops... 1120271055 J * daniel_hozac ~daniel@h56n2fls32o829.telia.com 1120271442 Q * sladen Ping timeout: 480 seconds 1120271766 Q * monrad Ping timeout: 480 seconds 1120271920 J * monrad ~monrad@213083190130.sonofon.dk 1120272231 J * sladen paul@starsky.19inch.net 1120272579 J * matti_ matti@linux.gentoo.pl 1120272586 Q * matti Ping timeout: 480 seconds 1120272593 N * matti_ matti 1120274316 J * eric__ ~eric@ool-182cef46.dyn.optonline.net 1120274324 N * eric__ jkl 1120274336 Q * monrad Ping timeout: 480 seconds 1120274377 M * jkl does anyone know if i can get logging or debugging info for starting vservers? all the sudden none of my vservers are starting up 1120274789 M * AprilDL Ok Bertl_zZ when you wake up you can tell me how easy it is to build the patched kernel NOT the debian way... after 24 + hours i admit defeat kernel 2.6.11 dell 2450 perc3si raid and adaptec7899. Every kernel panic I have to call the datacenter for a powercycle ugh. It booted fine to 2.6.11 before I try to build the kernel with the vserver patch. I'm not blaming the patch - more my skills at... 1120274790 M * AprilDL ...typing make-kpkg and dpkg apparently which is advertised as being very easy. 1120275077 M * kestrel jkl: the flower page seems to indicate that putting tty in /etc/vservers/vserver-name/apps/init will do what you want 1120275091 M * kestrel unfortunately i haven't been able to make this work 1120275506 J * monrad ~monrad@213083190130.sonofon.dk 1120275506 J * rs ~rs@imhotep.rhapsodyk.net 1120277746 Q * Aiken Quit: Leaving 1120278945 Q * badiane Ping timeout: 480 seconds 1120285480 Q * cryo Ping timeout: 480 seconds 1120285586 Q * complexho Ping timeout: 480 seconds 1120285986 J * _are_ ~are@dsl-084-056-151-209.arcor-ip.net 1120286234 M * _are_ Hi 1120286327 M * _are_ I run quite a few different vservers for different services, now I tried and move an old SuSE 6.2 box into a vserver on an opteron host. Other 32-bit vservers work, this one just doesn't start and I seem to be unable to get ... 1120286327 M * _are_ any debugging info from it. I am very sure I miss out on something, any hints? 1120288634 J * Doener_ ~doener@p54873DB3.dip.t-dialin.net 1120289066 Q * Doener` Ping timeout: 480 seconds 1120289720 J * SpaceDev ~vlad@72.29.99.216 1120289770 M * SpaceDev Is there any place I can obtain fully compiled kernel images (like the ones that used to be on solucorp.qc.ca) of the more recent 2.4 builds? 1120290153 M * SpaceDev anybody? 1120290388 M * id https://ghost.anime.pl/~eyck/log/Projects/bsd/ 1120290589 M * SpaceDev looks like just a bunch of patches... 1120290662 M * id SpaceDev, you are on debian ? 1120290669 M * SpaceDev no, redhat 1120290672 M * SpaceDev well, fc 1120290686 M * id ok there are only debs in the directory 1120290813 M * SpaceDev which directory is that?The site you pointed me to is a blog 1120290878 M * SpaceDev https://ghost.anime.pl/~eyck/Projects/bsd/deb/ i guess 1120290952 M * id right 1120290971 M * id maybe there is a tool called alien on rh/fc too 1120290995 M * id alien converts packages deb<->rpm 1120291011 M * SpaceDev yeah, I looked inside of that deb but they all seem to be just bunches of patches 1120291017 M * SpaceDev not a compiled image 1120291083 M * id -- btw - where is our problem to compile a kernel @ your own ? 1120291106 M * SpaceDev I'm not skilled enough to roll my own kernel 1120291209 M * id http://linux-vserver.org/FedoraHowTo 1120291221 M * id you can always try ;) 1120291236 M * id it is not that hard 1120291242 M * SpaceDev oh no, I've done it in the past 1120291262 M * SpaceDev but I don't have the skill to keep on rolling kernels every time a new patch set comes out 1120291271 M * SpaceDev and the associated troubleshooting.. 1120291283 M * id do you know the command "make oldconfig" ? 1120291328 M * SpaceDev ;) 1120291411 M * id sorry switched to 2.6 a while ago 1120291425 M * id (I would suggest to you too) 1120291470 M * id gotta go - cya laters 1120291699 Q * rs Quit: rs 1120292259 J * rs ~rs@imhotep.rhapsodyk.net 1120292869 J * complexho ~mark@funk.gotadsl.co.uk 1120293495 Q * rs Quit: rs 1120294537 Q * SpaceDev Quit: ircII EPIC4-1.0.1 -- Are we there yet? 1120295504 N * Doener_ Doener 1120295509 M * Doener morning! 1120295611 M * matti Hi Doener. 1120295612 M * matti ;] 1120295634 M * Doener hey matti, everything fine? 1120295650 M * matti Yep. 1120295676 M * Doener good :) 1120295691 M * eyck AFAIK, in those debs there is a ready kernel, not a patchset, 1120295742 M * matti Hi eyck. 1120296535 Q * _are_ Ping timeout: 480 seconds 1120297090 M * eyck hi matti 1120297271 J * maple ~maple@strace-nat0.ua-online.net 1120297682 M * Hollow moin Doener, i set up a svn repos for libvserver, you want access? 1120297757 M * Doener hm, i got little experience with userspace development under *nix (and just noticed that i basically broke the libvserver.h ;) 1120297786 M * Doener libvserver.h tries to include kernel/foobar.h, but after installation that's just wrong 1120297824 M * Hollow hm ;) 1120297844 M * Doener OTOH i libvserver.h should probably be splitted anyway and one should always include both (as needed) library and kernel headers 1120297879 M * Doener i.e. #include "vs_context.h"; #include "linux/vserver/context.h"; or something like that... 1120297880 M * Hollow yeah, that would prolly be the best way 1120297910 M * Hollow i also installed trac.. very nice app.. 1120297917 M * Hollow http://dev.croup.de/proj/libvserver/ 1120297935 M * Doener indeed it is, use that for a few things, too 1120297995 M * Doener if you don't mind that I'll probably fsck up the automake stuff from time to time, i'll gladly accept svn access ;) 1120298067 M * Hollow doesn't matter.. at least there are two who fsck the up ^^ 1120298074 M * Hollow *them 1120298136 M * Hollow Doener: query... 1120299893 J * _are_ ~are@dsl-084-056-151-209.arcor-ip.net 1120299933 M * _are_ hi (again, my provider is more off- than online) 1120302248 M * _are_ http://www.nongnu.org/util-vserver/doc/conf/configuration.html states you can link /etc/vservers/vserver-name/apps/init/tty to a tty you wnat the init-output redirected to, id the re a way to make it always the current tty? ... 1120302249 M * _are_ at least when I start it manually via vserver vservername start? 1120302307 M * Doener /proc/self/fd/1 should refer to your current stdout 1120302316 M * _are_ :-) 1120302331 M * _are_ i encountered my first vserver that won't boot and I'd like to know why 1120302482 M * _are_ well, actually after setting initstyle to plain it boots, but I'd still like to see the output. 1120302512 M * _are_ are the various init styles documented somewhere? I only know about fakeinit and plain and only because of some mails from the mailinglist 1120302561 M * kestrel i haven't been able to get the tty link to work 1120302607 M * Doener sysv (default): start the vserver via /etc/init.d/rc 3 (runlevel might be wrong ;) 1120302616 M * Doener plain: start init as the first process 1120302625 M * Doener gentoo: something gentoo specific 1120302629 M * kestrel hee 1120302631 M * kestrel hehe 1120302631 M * Doener those i know... 1120302650 M * kestrel i made a gentoo guest once, with the old tools, it was painful 1120302670 M * _are_ Doener: you know about any page in the wiki that already contains this info? If not, I will create one, suggesting name 'InitStyles' 1120302684 M * Doener don't know any... 1120302781 J * eXplasm explasm@p549F52F1.dip.t-dialin.net 1120303358 M * _are_ Page created, I'd say a link on http://linux-vserver.org/alpha+util-vserver would be suitable, but I am not sure abpout the correct placement in there, would be more suitable on the Flower Page (which is no wiki, afaik) 1120303470 M * Doener _are_: hm, I'd say below "vserver creation" or simply put it on the Documentation page... 1120303491 M * Doener on the alpha tools page, an own paragraph would probably be more suitable than a link 1120303589 M * _are_ well, basically I miss many detailed information on the Flower Page regarding what you can stuff into the files it describes, so a more specific doc page for alpha util-vserver would probably be a nice addition in the long ... 1120303589 M * _are_ run 1120304777 J * click click@dsl-static-122-208.aal.tiscali.no 1120307191 J * [maple] ~maple@strace-nat0.ua-online.net 1120308100 M * complexho hi Doener 1120308108 M * Doener hey complexho 1120308120 M * complexho did you take a look at that code yet? 1120308148 M * Doener glanced over it. a lot of empty directories ;) 1120308183 M * complexho really? the daemon was quite a sizeable project 1120308199 M * Doener doener@doener src $ ls home/ 1120308199 M * Doener CVS 1120308199 M * Doener doener@doener src $ ls home/CVS/ 1120308199 M * Doener Entries Repository Root 1120308212 M * Doener and there are quite a few directories like that 1120308232 M * complexho hmm 1120308259 M * Doener but the vsd* ones contain some stuff, didn't take a closer look yet 1120308272 M * complexho yes - that's where the daemon is (just unpacking now) 1120308314 M * complexho basically there is a c daemon plus a load of config modules, similar to what libvserver does but for freevsd. It had around 4 yrs of development up to the point of 1.4.9-2 1120308407 M * complexho we did do a port to ctx-12 + util-vserver back in 2002 but never managed to release before the company went under 1120308435 M * complexho sorry, not util-vser, it interfaced diretly to chbind, chcontext etc 1120308460 M * Doener probably it's been vserver-tools back then 1120308468 M * complexho yes 1120308473 M * Doener hm, what is/was freevsd? 1120308508 M * complexho it was one of the earlier vps projects... DSVR originally developed it back in 1997 and we stopped working on it as GPL project in about 2002 1120308529 M * Doener i c 1120308557 M * complexho in fact the company went bust, but the daemon was pretty powerful and flexible - network transparent and used ssl CA for auth and encryption 1120308584 M * Doener nice 1120308634 M * complexho the daemon was populated with functionality modules which advertised their features through the vsd daemon - so you could do vsdadm --help and get a complete lists of modules, features and syntax 1120308679 M * complexho and you could issue a command centrally and the daemon would execture the actual task remotely through a tcp socket 1120308712 M * complexho might give you some ideas if you are writing a daemon yourself - we had some good minds working on that 1120308776 M * complexho and I have about 15 server still running it without a bugfix since 2002 ;) 1120308807 M * Doener as of now, i plan to have the daemon do all the vserver stuff, communicating over a unix domain socket with client applications 1120308839 M * complexho we are looking into the same thing for a paid project, hence why I dusted off the old code recently ;) 1120308868 M * Doener those clients can then implement various useful functions. f.e. a control shell, some RPC/SOAP wrapper, the basic tools that come with util-vserver, or whatever 1120308973 M * complexho It's a very useful thing to do if you need to setup some sort of interface to the vserver system... Our daemon went much further and did app config (apache, bind, sendmail, users etc) as well, within the vserver itself... 1120308994 M * complexho I will download your system and check it out tonight 1120309043 M * Doener doesn't do much as of now, trying to get familiar with userspace programming atm... did mostly kernel stuff till now 1120309062 M * complexho what have you implemented so far? 1120309107 M * Hollow we seriously need some more devs for userspace utils.. 1120309111 M * Hollow imo 1120309153 M * Hollow libvserver is my first C stuff ever... 1120309221 M * _are_ well, I can read C, the last C I wrote has been 12 years ago 1120309245 M * complexho well like I said we are looking at it and I do have a couple of coders at hand who will be doing our stuff... We would be happy to discuss our aims with you if there is crossover 1120309257 M * Doener complexho: just some hackish unix domain socket communication, functions for context creation (setup-less) and context 'entering' 1120309276 M * complexho Myself personally, I am limited at bash and a n00b at Perl, but 7 yrs workng with vservers ;) 1120309321 M * Doener the last part work via a pseudo terminal whose master is given to the client via the uds, the client then forwards between the pseudo terminal inside the vserver and the user's current terminal 1120309362 M * Doener thus the "enter" part avoids the pty problems that you have with util-vserver's "vserver xxx enter" 1120309364 M * complexho sounds interesting 1120309420 M * Doener the current client application is what could evolve into a control shell, stand-alone tools will come later 1120309435 M * complexho ok how do you mean, pty problems? 1120309465 M * Doener util-vserver does not create a pty inside the vserver when you enter it, you simply keep using your pty/tty from outside the vserver 1120309483 M * complexho of course, you don't show up on 'w' 1120309491 M * complexho I quite like that ;) 1120309513 M * Doener and because ptys are protected across context boundaries, you can't do some things 1120309536 M * Doener for example there the good old "mesg: /dev/pts/7: Operation not permitted" 1120309556 M * Doener and also screen does not work inside the vserver if you used "vserver xxx enter" to get inside 1120309558 M * complexho maybe I never noticed that, my usage is generally just setting up LAMP 1120309589 M * complexho I noticed the screen issue but not the /dev/pts/7 error 1120309620 M * Doener maybe your distro doesn't do "mesg n" in your bashrc or wherever that is called here... 1120309620 M * complexho is that the same reason my ssh agent won't come into a vserver when I do enter? 1120309629 M * complexho (FC1-3) 1120309663 M * Doener no, that's because the ssh-agent works over a unix socket (usually in /tmp IIRC) and that is, of course, not available inside the vserver 1120309703 M * complexho is there any nice 'hack' method like symlinking on the way in to get around that, or is that a security issue? 1120309769 M * complexho to be honest though, these days I tend to just ssh directly to the vs... 1120309781 M * Doener i do that all the time ;) 1120309789 M * Doener it's the preferred way 1120309810 M * complexho even from the root of the box, I have an alias ;) 1120309828 M * Doener same here 1120309940 M * complexho the other little thing that bugs me, which have never got round to resolving... When syslogd starts it always tries to start klogd (which always dies of course). Do you have this issue 1120309971 M * complexho is it possible to disable the klogd component? 1120309984 M * Doener usually you don't need klogd inside the vserver, but recent versions of the vserver patch should also virtualize that IIRC 1120310007 M * Doener i.e. klogd doesn't get any messages, but keeps running without errors 1120310024 M * complexho right... just seems a little untidy but hey it's minor really 1120310086 M * complexho I have also been working on some scripts to manage the local firewall based on the vserver config, but I'm too embarrassed about my code to show anyone... 1120310088 M * complexho lol :) 1120310133 M * complexho hoping to do a new version when I get time thatis more suitable for stas gathering 1120310140 M * complexho *stats 1120311671 M * Hollow off for now, back tomorrow.. 1120311682 M * Doener cya! 1120311701 M * Hollow cya Doener and happy commiting :P 1120311704 M * Doener hehe 1120311722 M * Doener Hollow: hm, why is the uts stuff in vnamespace in libvserver? 1120311746 M * Hollow you mean vhi? 1120311786 M * daniel_hozac i'd be happy to help with the userspace stuff. 1120311825 M * Doener Hollow: yep 1120311830 M * Doener daniel_hozac: great :) 1120311892 M * Hollow daniel_hozac: ping me tomorrow (CET) so i can give you write access 1120311897 M * Hollow but i'm really afk now ;) 1120311904 M * daniel_hozac ok 1120312608 M * kestrel .win shr 4 1120314380 M * Hollow Doener: where should the vhi stuff else be? 1120314383 N * Bertl_zZ Bertl 1120314396 M * Bertl morning folks! 1120314397 M * Hollow hey Bertl 1120314409 M * complexho hi Bertl 1120314410 M * Doener util-vserver has vuname... after all the uts stuff has nothing to do with fs namespaces... 1120314413 M * Doener morning Bertl 1120314430 M * Hollow Doener: but it's defined in kernel/namespace_cmd.h 1120314447 M * Bertl yup, there is no separate command section yet 1120314457 M * Hollow eh.. namesapce.h 1120314485 M * Doener Bertl: hm, vx_uts_virt_handler is in cvirt.h... 1120314489 M * Hollow Bertl: what is cvirt supposed to do? 1120314495 M * Hollow *g* 1120314505 M * Doener (well, it was me who put it there ;) 1120314533 M * Bertl yep, sec 1120314615 M * Bertl we could move the vhi/uts stuff into cvirt ... 1120314633 M * Bertl or make a separate utsname include? 1120314649 M * Bertl the commands in namespace are misplaced anyway 1120314674 M * Bertl let me know your preference, I'll do the cleanup of the headers 1120314683 M * Hollow i'd prefer cvirt 1120314705 M * Hollow and namespace commands like namespace_{create,migrate} or so.. 1120314710 M * Hollow just like ctx 1120314717 M * Doener hm, it is context virtualization and we won't get much more uts stuff so I'd also go for cvirt 1120314744 M * Bertl Hollow: well, the namespaces do not work this way, at least kernel side 1120314761 M * Bertl i.e. namespace_create is something which happens in userspace 1120314769 M * Hollow ah 1120314817 M * Bertl we can talk about the 'migrate' part but I guess it's not worth the changes (which require compatibility stuff anyways) 1120314860 M * Doener what would namespace_migrate do? 1120314867 M * Doener sounds like namespace_ente... 1120314872 M * Doener s/ente/enter/ 1120314872 M * Bertl the same as namespace enter ... 1120314922 M * Doener hm, guess i didn't get the 'question' then... 1120314942 M * Bertl 16:31 < Hollow> and namespace commands like namespace_{create,migrate} or so.. 1120314961 M * Bertl 16:33 < Bertl> we can talk about the 'migrate' part but I guess it's not worth the changes (which require compatibility stuff anyways) 1120314966 M * Doener yeah, but what did he mean? creating those? 1120314988 M * Doener "get" like "understand", not "receive" ;) 1120314990 M * Hollow no rename 1120314992 M * Hollow ;) 1120314992 M * Bertl I interpreted this as a request to make the commands more smilar 1120314997 M * Doener ah, i see 1120315062 M * Hollow Bertl: btw, there is some progress wrt libvserver ;) http://dev.croup.de/proj/libvserver/timeline 1120315096 M * Doener Hollow: any objections against moving the uts stuff to vuname? 1120315097 M * Hollow we've a svn repos and write access for everyone who wants to help 1120315110 M * Hollow Doener: no 1120315113 M * Doener ok 1120315151 M * Hollow Doener: newheader-layout++ 1120315193 M * Hollow so... around 6pm CET i'll be away to a friends... 1120315469 M * Bertl okay, so from the kernel header side, we move the uts stuff to cvirt, yes? 1120315492 A * Doener agrees 1120315515 M * Hollow jup 1120315575 M * Bertl and of course, we get a namespace_cmd.h too, right? 1120315587 A * Hollow nods 1120315589 M * Doener yep 1120315783 M * Bertl and of course we change the vx_set_vhi ... stuff 1120315791 M * Hollow to..? 1120315798 M * Bertl http://linux-vserver.org/ToDo+List+Kernel 1120315801 M * Bertl (last entry) 1120315812 M * Hollow ah, yep.. 1120315923 M * Hollow is all this required to release 2.0? 1120315974 M * Bertl hehe, no, but if we get stuff done, the better for us :) 1120316003 M * Hollow ;) 1120316025 M * Doener ok, uts stuff is splitted in libvserver 1120316094 M * Hollow Doener: shouldn't uts.c be named cvirt.c? 1120316151 M * Doener feel free to rename it... i tend to separate stuff too much 1120316195 M * Hollow i mean just be consistent with the header namings ;) 1120316367 M * Doener got no problem with it... I'm the type of guy that might have a .h/.c combo for each function 1120316393 A * Doener .oO( hmm... someone said Solaris has a .c file for each syscall... ) 1120316409 M * Hollow heh 1120316564 M * albeiro not that bad actualy ;) 1120316585 M * albeiro (after going throught all linux syscalls for some review ;) 1120316598 M * albeiro well, ok, cscope can substitute it ;p 1120316840 M * Doener Hollow: is the context field in vhifields reserved for anything? 1120316854 A * Hollow shrugs 1120316867 M * Hollow isn't this the context name? 1120316878 M * Doener context's don't have a name in the kernel ;) 1120316886 M * Doener s/context's/contexts/ 1120316894 M * Bertl okay, dinnertime ... back in a few ... 1120316900 N * Bertl Bertl_oO 1120316906 M * Hollow so vps shows what? the nodename? 1120316919 M * Doener no, it looks up the name in the vserver configuration 1120316928 M * Hollow ah 1120318153 M * Doener ah, i c, you got that from the kernel headers... 1120318189 M * Hollow ? 1120318198 M * Doener the context field 1120318204 M * Hollow ah 1120318218 M * Hollow i just cleaned up stuff a bit ;) 1120318272 M * Doener there actually _is_ a field for the context name in the kernel... but it's unused... 1120318383 M * Hollow tztztz ;) 1120318519 M * Bertl_oO to clarify, the VHIN_CONTEXT was intended to store the guest name in the kernel 1120318534 M * Bertl_oO (to allow for consistent xid -> name lookups) 1120318556 M * Bertl_oO I really don't know if enrico uses it, but he originally requested it, IIR 1120318612 N * Bertl_oO Bertl 1120318614 M * Doener ah, the vx_vhi_name thingy is in both functions, so you actually can get the name, missed that 1120318626 M * Hollow and you can set it? 1120318658 M * Doener VHIN_CONTEXT 1120318668 M * Hollow yup 1120318685 M * Doener hm, i misread that as "how can you...", sorry 1120318691 M * Doener yes, you can set and get it 1120318696 M * Hollow great! :) 1120318739 M * Doener hum, hum... that strsep thing is evil ;) 1120318746 M * Hollow imo we should make vuname compatible with uname (wrt cmd line parameters) + add a switch for ctxname 1120318758 M * Hollow Doener: right you are 1120318775 M * Doener segfaults all over the place *g* 1120318794 M * FaUl kann winamp unterdessen ogg-streams abspielen? 1120318796 M * Doener and a switch to set those values would also be needed 1120318797 A * Hollow nods 1120318797 M * FaUl oh, ww 1120318847 M * AprilDL Upgraded to 2.6.11 and machine booted fine - using initrd image. Now I need to build a custom kernel so that the vserver patch can be included. How can I do this so that initrd image will still be the boot method? Using make-kpkg then dpkg -i, lilo.conf changes to not using initrd image and I probably don't have a new initrd image. I wind up with kernel panic. It's a dell 2450 with raid... 1120318849 M * AprilDL ...which people say they have trouble getting to work with 2.6 kernel. 1120318889 M * DaPhreak AprilDL: you're using debian, right ? 1120318894 M * AprilDL yes 1120318910 M * Bertl AprilDL: usually folks have a tool called mkinitrd (from your distro) which does build the initrd 1120318942 M * Bertl AprilDL: you should make sure that some things like ramdisk, the filesyste, some compressions, etc, are compiled into the kernel 1120319025 M * Doener the things that uname knows about to not completely fit. it knows some additional things and it misses at least the domainname 1120319031 M * Doener s/to/do/ 1120319132 M * AprilDL ramdisk is compiled in the kernel - it makes 16 of them 4096 each. yes filesystem is there. /me wonders why i can't just add vserver to the "existing" kernel that works (the one created by upgrading to 2.6.11) without making all these changes. 1120319163 M * AprilDL trying mkinitrd now altho it scares me because my debian guru guy said he's never done it successfully 1120319246 M * Bertl AprilDL: well, 'adding' might be fine, did you 'rebuild' the debian kernel yourself? 1120319260 M * Bertl (the 2.6.11 kernel) 1120319383 M * AprilDL yes 1120319421 M * AprilDL well i don't know what you mean by "myself" - since it was make-menuconfig to pick what to build into the new kernel and then make-kpkg 1120319552 M * AprilDL the unpleasant thing is that I'm working thru a remote KVM which cannot power cycle or hard reset. So every time I get kernel panic I have to submit a trouble ticket for the data center people to go powercycle it and I have to standby to change the boot sequence back to the rescue cd. So i wish I could avoid kernel panic... 1120319608 M * Bertl okay, so you 'managed' to build a kernel (non-debian) with make-kpkg after configuring it with make menuconfig, right? 1120319722 J * random012 ~wwee@24-52-166-190.lndnnh.adelphia.net 1120319729 M * Bertl welcome random012! 1120319732 M * random012 hiya 1120319735 N * random012 mrplum 1120319834 M * mrplum I'm really stressing my brain trying to compensate the differences between vserver, UML, and Xen. My intent is to run a virtual linux machine inside my current Debian install to host a VoIP server which needs access to a PCI modem card. 1120319867 M * mrplum But I can't rseem to figure out how everything is seperated and if any real hardware access is permitted, so my question is can I install Debian inside of debian using vserver and have access to my pci modem card? 1120319872 M * Bertl well, then you are wrong here, because linux-vserver is no virtual machine :) 1120319884 M * mrplum see that's me confused :) 1120319891 M * mrplum what is it ? 1120319903 M * mrplum an advanced jail ? 1120319912 M * Bertl it is a virtualization technique similar to jails 1120319922 M * AprilDL bertl: I built the kernel using the debian tools make-kpkg then dpkg -i 1120319953 M * Bertl AprilDL: okay, then you should be able to apply the vserver patches, make oldconfig and build a new one ... 1120320029 M * Bertl mrplum: how do you access the PCI modem card? as ttySX 1120320066 M * mrplum no it's something like /dev/zap/modem0 1120320070 M * mrplum software based modem 1120320077 M * Bertl yeah, okay, fine too ... 1120320089 M * Bertl you can provide that to a linux-vserver guest quite fine 1120320101 M * DaPhreak well just create the device inside the guest, right ?! :) 1120320119 M * Bertl (from the host, into the guest, yes) 1120320186 M * mrplum really hmm.. and it's not like the chroot jails I read are so easy to break out of ? 1120320223 M * Bertl if you manage to break out of a secure guest, you get a place in our Hall'o'Fame 1120320231 M * mrplum :-P 1120320367 M * AprilDL Bertl: The only way I know to apply vserver patches is using make-kpkg --added-patches vserver. I will study about make oldconfig to see if that somehow would use the old initrd image that did boot ok as the basis for everything. 1120320406 M * Bertl AprilDL: no, you need the make oldconfig to configure the new options for the vserver patch 1120320414 M * mrplum Support for 2.6 kernels? 1120320424 M * Bertl AprilDL: it's equiv to making menuconfig (as you did before) 1120320437 M * Bertl mrplum: better than for 2.4 (feature wise :) 1120320456 M * mrplum oh my smile keeps getting bigger and bigger 1120320489 M * Bertl about 12 different archs are supported (6-8 of them are tested) 1120320539 M * mrplum now I need to find a doc to help me dive into all this, cause I'm horribly lost 1120320552 M * Bertl linux-vserver.org (starting point) 1120320584 M * Bertl http://linux-vserver.org/Linux-VServer-Paper 1120320592 M * Bertl (explanation of the principles) 1120320715 M * [maple] flower page the best. creator must be crazy ;) 1120320868 M * DaPhreak heh 1120320869 M * Bertl [maple]: well, 'de gustibus non disputandum est' and there are different stylesheets :) 1120320888 M * DaPhreak [maple]: just use the boring style *g* 1120320920 M * [maple] using lynx for that ;) 1120320956 M * DaPhreak ah .. thats even better ;) 1120320988 M * Doener mrplum: http://linux-vserver.org/Step-by-Step+Guide+2.6 1120321018 M * Bertl ah, yeah, forgot about that one, thanks! 1120321132 M * Doener np 1120321607 M * Doener dinner time, back later 1120321810 J * rs ~rs@imhotep.rhapsodyk.net 1120321816 M * Bertl evening rs! 1120321816 M * mrplum thanks for all your help, i'll visit again 1120321821 Q * mrplum Quit: Leaving 1120321830 M * rs hi 1120322116 M * Bertl rs: how is life, the universe, and everything? 1120322179 J * shuri shuri@64.235.209.226 1120322197 M * shuri hi 1120322314 M * Bertl evening shuri! 1120324191 M * _ag_ AprilDL: did you read the manpage of make-kpkg? you have to pass it --initrd for the initrd.gz image to be create in the .deb package 1120324304 Q * rs Quit: rs 1120324465 M * _ag_ s/create/created/ 1120325284 M * AprilDL _ag_ yes I did read that. I was swayed to try making a non-initrd kernel image as that was the instruction of the newbie kernel build page I followed and I didn't know that possibly the dell 2450 would require the initrd image possibly due to the raid drivers. 1120325331 M * AprilDL now when I do mkinitrd it says it cannot find the module for the scsi driver. so maybe thats a clue 1120325362 M * AprilDL going to the debian on dell page to see if it will tell me how to get that module into my source tree or something 1120325405 M * Bertl IIRC, some of those scripts try to _include_ the currently loaded modules 1120325439 M * Bertl if you have build them into the kernel, or they are named differently, then the detection might have changed 1120325449 M * Bertl s/changed/failed/ 1120325966 M * AprilDL the source that i'm building from is not the source that the successful kernel was built from. i didn't know where to find it so i went and downloaded it again. 1120326002 M * Bertl well, if you 'built' the debian kernel from a debian source, I'd be careful ... 1120326005 M * AprilDL the successful kernel came from using apt-get upgrade distribution or similar command 1120326041 M * AprilDL yes it was from a debian source. i can tell how much you love debian maintainers :) 1120326046 M * Bertl because the debian kernels differ somewhat from the mainline kernels 1120326096 M * Bertl AprilDL: you said you have some kind of remote console, no? 1120326112 M * AprilDL yes an IP based kvm 1120326128 M * Bertl so there should be no problem then, no? 1120326159 M * AprilDL well my biggest problem is that i cannot reset or reboot the machine after a kernel panic 1120326175 M * Bertl just add the option panic=60 1120326178 M * AprilDL and that happens every time i unsuccessfully try a new kernel build 1120326185 M * Bertl the kernel will then reboot 60 seconds after a panic 1120326188 M * AprilDL oh thank god i knew there had to be something like that 1120326195 M * AprilDL where do i put it 1120326210 M * Bertl add it to the kernel command line (depends on your bootloader) 1120326218 M * AprilDL lilo 1120326245 M * Bertl add it to the append=" statement 1120327292 J * Loki|muh_ loki@satanix.de 1120327292 Q * Loki|muh Read error: Connection reset by peer 1120327292 Q * _mountie Read error: Connection reset by peer 1120327319 J * _mountie ~mountie@CPEdeaddeaddead-CM000a739acaa4.cpe.net.cable.rogers.com 1120327361 Q * Loki|muh_ uranium.oftc.net keid.oftc.net 1120327361 Q * shuri uranium.oftc.net keid.oftc.net 1120327361 Q * eXplasm uranium.oftc.net keid.oftc.net 1120327361 Q * sladen uranium.oftc.net keid.oftc.net 1120327361 Q * Pazzo uranium.oftc.net keid.oftc.net 1120327361 Q * SNy uranium.oftc.net keid.oftc.net 1120327361 Q * meebey_ uranium.oftc.net keid.oftc.net 1120327361 Q * maharaja uranium.oftc.net keid.oftc.net 1120327361 Q * tchan uranium.oftc.net keid.oftc.net 1120327361 Q * neofutur uranium.oftc.net keid.oftc.net 1120327361 Q * albeiro uranium.oftc.net keid.oftc.net 1120327442 J * tchan_ ~tchan@c-24-13-81-164.hsd1.il.comcast.net 1120327442 J * shuri shuri@64.235.209.226 1120327442 J * eXplasm explasm@p549F52F1.dip.t-dialin.net 1120327442 J * sladen paul@starsky.19inch.net 1120327442 J * neofutur ~neofutur@neofutur.net 1120327442 J * maharaja maharaja@ipax.at 1120327442 J * albeiro ~albeiro@albeiro.usercloak.oftc.net 1120327442 J * meebey_ meebey@booster.qnetp.net 1120327442 J * Pazzo ~Pazzo@host130-250.pool8172.interbusiness.it 1120327442 J * SNy ~mfr@bmx-chemnitz.de 1120327443 Q * albeiro Read error: Connection reset by peer 1120327445 J * albeiro ~albeiro@graffias.estrefa.pl 1120327586 J * Loki|muh loki@satanix.de 1120327923 Q * Loki|muh uranium.oftc.net keid.oftc.net 1120327923 Q * albeiro uranium.oftc.net keid.oftc.net 1120327923 Q * SNy uranium.oftc.net keid.oftc.net 1120327923 Q * Pazzo uranium.oftc.net keid.oftc.net 1120327923 Q * meebey_ uranium.oftc.net keid.oftc.net 1120327923 Q * maharaja uranium.oftc.net keid.oftc.net 1120327923 Q * sladen uranium.oftc.net keid.oftc.net 1120327923 Q * eXplasm uranium.oftc.net keid.oftc.net 1120327923 Q * shuri uranium.oftc.net keid.oftc.net 1120327923 Q * tchan_ uranium.oftc.net keid.oftc.net 1120327923 Q * neofutur uranium.oftc.net keid.oftc.net 1120327994 J * tchan ~tchan@c-24-13-81-164.hsd1.il.comcast.net 1120328045 J * meebey meebey@booster.qnetp.net 1120328206 J * Loki|muh loki@satanix.de 1120328206 J * albeiro ~albeiro@graffias.estrefa.pl 1120328206 J * sladen paul@starsky.19inch.net 1120328206 J * shuri shuri@64.235.209.226 1120328206 J * eXplasm explasm@p549F52F1.dip.t-dialin.net 1120328206 J * neofutur ~neofutur@neofutur.net 1120328206 J * maharaja maharaja@ipax.at 1120328206 J * Pazzo ~Pazzo@host130-250.pool8172.interbusiness.it 1120328206 J * SNy ~mfr@bmx-chemnitz.de 1120329468 M * _are_ uff, anyone in here with a clue about oracle and vservers? I am pretty sure I have already heard of that comination running, but the one I have to run here (some hardware crashed -> move it to vserver now instead of in 4 ... 1120329468 M * _are_ weeks) won't start the database same IPs ofc, it claims: 1120329470 M * _are_ SQL> Connected to an idle instance. 1120329470 M * _are_ SQL> ORA-27123: unable to attach to shared memory segment 1120329470 M * _are_ Linux Error: 22: Invalid argument 1120329470 M * _are_ SQL> Disconnected 1120329513 M * Bertl you are trying to attach across contexts` 1120329521 M * Bertl +? 1120329539 M * _are_ erm, doubt it, I start the vserver and in there the oracle should start 1120329554 M * _are_ I entered the vserver with vserver oracle enter 1120329556 M * Bertl hmm, are we talking about 2.6? 1120329575 M * _are_ kernel 2.6 on dual opteron, 64bit host 1120329592 M * _are_ I first assumed it is the IPC* CAPs, but seems I am wron there 1120329598 M * Bertl oracle is 64bit, and/or it works on the host? 1120329638 M * _are_ nope, oracle is 32 bit, it is a very old version I'd love to get rid of, but can't atm 1120329660 M * _are_ it runs on an old suse 6.2, but the hardware crashed so I had to move it into a vserver from backup 1120329686 M * Bertl okay, so we do not know why it fails ... 1120329693 M * Bertl (vserver or 64bit machine) 1120329704 M * _are_ well, it is a 32 bit vserver 1120329744 M * Bertl I'd suggest to try with chroot only (so that the vserver guest is not activated) 1120329750 M * _are_ k 1120329871 M * _are_ ouch, happens with chroot, too 1120329917 M * _are_ guess I have a real problem then 1120329919 M * Bertl so it's very likely an incompatibility with the kernel or with kernel settings ... 1120329959 M * _are_ ah well, will try running an UML with the original kernel 1120329985 M * Bertl have you googled for the error 1120329997 M * Bertl UML with the original kernel is probably not that easy, btw .) 1120330015 M * _are_ uff, sory, seems I'm getting old, no I have not googled 1120330038 M * _are_ no, but with the original kernel version 1120330063 M * Bertl was it UML compiled? 1120330097 M * _are_ no, ofc not, it is suse 6.4, some 2.2 kernel I assume 1120330122 M * Bertl hmm, and you think you will get a 2.2 suse 6.4 kernel to compile for UM? 1120330130 M * _are_ will try google first, if that fails, i will try and find out the actual kernel version if I can UML it 1120330150 M * Bertl http://searchoracle.techtarget.com/originalContent/0,289142,sid41_gci933283,00.html#group9 1120330154 M * _are_ oh, I could qemu it if that fails 1120330262 M * _are_ oi, they want me to reg for reading the answers 1120330369 J * Loki|muh_ loki@satanix.de 1120330471 Q * SNy jupiter.oftc.net keid.oftc.net 1120330471 Q * Pazzo jupiter.oftc.net keid.oftc.net 1120330471 Q * maharaja jupiter.oftc.net keid.oftc.net 1120330471 Q * eXplasm jupiter.oftc.net keid.oftc.net 1120330471 Q * shuri jupiter.oftc.net keid.oftc.net 1120330471 Q * sladen jupiter.oftc.net keid.oftc.net 1120330471 Q * Loki|muh jupiter.oftc.net keid.oftc.net 1120330471 Q * neofutur jupiter.oftc.net keid.oftc.net 1120330471 Q * albeiro jupiter.oftc.net keid.oftc.net 1120330553 J * albeiro ~albeiro@graffias.estrefa.pl 1120330553 J * sladen paul@starsky.19inch.net 1120330553 J * shuri shuri@64.235.209.226 1120330553 J * eXplasm explasm@p549F52F1.dip.t-dialin.net 1120330553 J * neofutur ~neofutur@neofutur.net 1120330553 J * maharaja maharaja@ipax.at 1120330553 J * Pazzo ~Pazzo@host130-250.pool8172.interbusiness.it 1120330553 J * SNy ~mfr@bmx-chemnitz.de 1120331204 M * Doener back now... 1120331230 M * Bertl wb Doener! 1120331244 M * Doener thx 1120331340 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-2.0/delta-vhi-clean01.diff 1120331344 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-2.0/delta-kconfig-clean01.diff 1120331867 M * _are_ gnnnn, oracle technet has the answer 'check the manual, you can download it here' -> oracle signon -> create account -> email, password, password confirm, proceed -> duplicate fields, your request can't be processed: (and ... 1120331867 M * _are_ then the empty list of fields that are duplicate) 1120331942 M * _are_ hmmm, ok, the error message translate to: 'you fool already signed up 4 years ago, go remember your password' 1120332706 M * Bertl yeah, that's support as we love it, no :) 1120332720 M * Bertl the advantage of proprietary products *G* 1120332734 M * Doener looks good 1120332832 M * _are_ well, if I wozuld be the one to decide it would be a postgres, but in this case I am not 1120332843 M * Bertl Doener: I assume you meant the patches ... 1120332852 M * Doener right 1120332875 M * Doener well, the support is probably at least as good ;) 1120334153 Q * complexho Quit: using sirc version 2.211+KSIRC/1.3.12 1120334939 Q * eXplasm Ping timeout: 480 seconds 1120334974 J * eXplasm explasm@p549F52F1.dip.t-dialin.net 1120335170 M * Bertl wb eXplasm! 1120335411 M * eXplasm :) 1120336031 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-2.0/delta-debug-clean01.diff 1120336043 M * Bertl (jsut an update to the debug.txt) 1120336073 M * micah I have a question -- can you limit the amount of memory a vserver sees? Say the host has 1gig, can you make a vserver only have 512meg? 1120336079 M * micah my guess is vsched? 1120336088 M * Bertl interesting guess ;) 1120336089 M * micah Bertl: btw... am trying your updated patch in about 30 minutes 1120336111 M * micah (the debian kernel ext3 chattr fix) 1120336115 M * Bertl you can limit RSS, and VM ... 1120336132 M * micah ok, so when you limit a vserver to 512meg, they are only able to see 512meg available using things like free? 1120336201 M * Bertl if you enable the virtualization for that, yes 1120337004 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.12.2-vs2.0-rc5.diff 1120337013 M * Bertl (final testing state) 1120337025 M * _are_ uiii, rc5 1120337047 M * Bertl is anybody willing to verify that _all_ FOR-2.0 patches are included? 1120338093 M * _are_ gnnnnn, the opteron is a NUMA architecture and oracle 8.1.6 seems not to work on NUMA architectures. :-/ 1120338141 M * _are_ and to my understanding the vserver concept won't allow pretending to be a single cpu architecture 1120338161 M * Bertl NUMA != Single CPU 1120338174 M * _are_ yes, it is 2 CPUs 1120338194 M * Bertl I meant: don't confuse NUMA with SMP or SMT 1120338217 M * _are_ NUMA = Non Unified Memory Architecture, doe sthat make any sense without SMP? 1120338287 M * Bertl yes, for example fast vs. slow memory? 1120338331 M * _are_ uhm 1120338389 M * Bertl but I guess oracle is just 'confusing' SMP/SMT with NUMA or it chokes on some NUMA support in the kernel 1120338486 J * rs ~rs@imhotep.rhapsodyk.net 1120338852 M * _are_ well, whatever, I guess I have to try either qemu or uml :-/ 1120339374 J * are|lunch ~are@dsl-084-056-146-027.arcor-ip.net 1120339785 Q * _are_ Ping timeout: 480 seconds 1120339934 N * are|lunch _are_ 1120340425 M * Bertl wb _are_, enjoyed your lunch? 1120340767 M * Doener damn it... util-vserver also comes with a libvserver.a and i wonder for almost an hour why i can't link my test program... 1120340823 M * Bertl lol 1120340987 M * Bertl http://www.osdl.org/plm-cgi/plm?module=patch_info&patch_id=4576 (sparc is impressive :) 1120341049 M * _are_ Bertl: this is just the next nick on the list, I still foght oracle 1120341051 M * Doener woah! 1120341343 M * micah Bertl: is this an automatic patch build farm? 1120341415 M * Bertl yep, Patch Lifecycle Manager (a service provided by osdl) 1120341568 M * micah thats really cool 1120341594 M * micah debian has a similar thing, the buildd network, where any package that you upload is built on many different architecutres 1120341598 M * Bertl yeah, especially as there are scripts to show deviations from the baseline kernels 1120341603 M * micah and then bugs are automatically filed on failures etc. 1120342318 J * Greek0 ~greek0@81.189.246.175 1120342319 M * Greek0 hi 1120342365 M * Doener welcome Greek0 1120342377 M * Bertl hey Greek0! 1120342379 M * Greek0 what are the default capabilities the guests have on 2.6? 1120342426 M * Doener CapPrm: 00000000d44c04ff 1120342427 M * Doener CapEff: 00000000d44c04ff 1120342434 M * Bertl tx 1120342641 M * Greek0 Doener: how did you get those values? or where can they be found in the source tree? 1120342647 M * Doener cat /proc/self/status 1120342687 M * Bertl the caps are configured by the tools ... 1120342691 M * Doener those can be found "somewhere" in the util-vserver code... grepping for secure (case insensitive) should help you finding them 1120342914 Q * rs Quit: rs 1120343070 M * Doener hmm, we have a macro called SEXIT in libvserver... ;) 1120343131 M * Bertl so? :) 1120343291 M * Doener hm, well, maybe my mind splits the word in a different way than yours does ;) 1120343441 M * Greek0 hehe 1120343477 M * micah Bertl: your patch seems to work 1120343490 M * micah (naturally) :) 1120343517 M * micah Bertl: I will upload a new version of the debian package with this new patch replacing the old one 1120343525 M * micah so other debian people will not come complaining 1120343527 M * micah ;) 1120343675 M * Bertl good ... 1120343699 M * Bertl maybe you want to work yourself through the FOR-2.0 patches and apply them? 1120344408 M * Doener Context: /etc/vservers/naucki 1120344424 M * Doener hm, that seems to be what the tools write into the name field in vx_info 1120344507 M * Doener the "tools" = util-vserver.. guess we need to use real names again ;) 1120344770 J * notasnark ~sam@mailgate.glendale.org.uk 1120344845 M * Bertl welcome notasnark! 1120344851 M * notasnark Hi 1120344879 M * notasnark Anyone here know anything about vserver? :-) 1120344896 J * rs ~rs@imhotep.rhapsodyk.net 1120344956 M * notasnark I've just installed it on Gentoo, and though it sort of appears to work, the network can't see it, and services don't seem to be starting correctly. I can't find any logs anywhere however. 1120345062 M * notasnark Can anyone point it at where a virtual server logs to when it is started? 1120345100 M * Bertl Doener: http://vserver.13thfloor.at/Experimental/2.6.12.2-vs2.0-rc5-report.txt 1120345130 M * Bertl notasnark: it usually logs to its own log files 1120345146 M * Bertl notasnark: but check with vserver-stat if your guest is running 1120345219 N * DaCa DaCa-fi 1120345221 M * notasnark Where are those logs? Within the virtual server or on the host system? vserver-stat says that something is running, and I can 'vserver gentoo enter' to enter the virtual. 1120345231 M * Doener Bertl: hm, those "i have no idea why it failed" vmlinux errors again 1120345276 M * Doener but quite a few warnings are gone... how come? 1120345288 M * Bertl notasnark: yes, the logs are the same as on a normal machine and they are all inside the guest 1120345294 M * notasnark I have posted a fuller description of what I'm seeing to the Gentoo forums, at http://forums.gentoo.org/viewtopic-t-355449.html 1120345326 M * Bertl looks good 1120345330 M * notasnark There are no logs in /var/log which is where I'd normally look. syslong-ng is one of the things refusing to start. 1120345358 M * Doener gtg, probably back in about an hour... 1120345361 M * Bertl I assume that your host has a different ip than 192.168.161.30? 1120345363 M * Greek0 ok, I think the capabilites available in the guests by default are the ones left back by reducecap --secure 1120345387 M * Bertl Greek0: yes, unless you specify others in the config, of course ... 1120345403 M * notasnark The host IP is 192.168.161.11. If I try to connect to 192.161.163.30 then I end up connecting to the host. 1120345420 M * Bertl conenct with what? 1120345450 M * notasnark The .30 address is only enabled if the virtual is running. ssh sees the host's sshd. nmap sees the host's ports. 1120345465 M * Greek0 Bertl: so /etc/vservers/*/ccap only extends the default caps? 1120345474 M * Bertl notasnark: there are only 'host' ports ... 1120345493 M * Bertl notasnark: did you restrict the hosts sshd (Listen directive) to use only host ips? 1120345500 M * Greek0 */capabilites of course 1120345519 M * Bertl Greek0: ccaps are something different than bcaps ... 1120345544 M * Bertl Greek0: and yes they 'modify' the secure default 1120345619 M * notasnark How do you mean only host ports? I may well be wrong, but I'm assuming that the virtual server would be running on .30, and if I was running a web server on the virtual, but none on the host, then connecting to .30:80 would see a web server, but .11:80 would not. So nmap against the host IP should show different results than against the virtual IP. Or have I got it all wrong? 1120345663 M * Greek0 Bertl: is there any in-depth documentation on it? the flower page isn't very verbose on that topic :-/ 1120345693 M * Bertl notasnark: the host is able to bind to all ips, the guests are restricted to their IP subsets ... 1120345741 M * Bertl notasnark: of course, if you bind an sshd to 0.0.0.0 on the host, the guest will not be able to bind one at the same port 1120345763 M * Bertl Greek0: the source? 1120345794 M * Greek0 already looking at it.. I just thought there might be a quicker way.. 1120345851 M * notasnark But an external machine will see 'two' machines - one at the host IP and the other at the virtual IP, yes? I've been testing the network from a seperate machine. sshd won't even start on the guest, so sshd configuration isn't an issue at the moment :-( 1120345925 M * Bertl it will not start, because the host blocks the port 1120345954 M * Bertl restricting the host's sshd to .11, will allow the guest to start the sshd 1120345970 M * Bertl the sshd of the guest does not need to be restricted 1120345981 M * Bertl (it is restricted by the chbind config) 1120346039 M * _ag_ Bertl: i haven't give it a try yet, but do you appear to know whether ipv6 is handled? 1120346054 M * Bertl on the host yes, for the guest no ... 1120346095 M * _ag_ Bertl: so... patch idea! ;) because i will need it ;P 1120346156 Q * monrad Ping timeout: 480 seconds 1120346178 M * Bertl _ag_: it is not done for a good reason ... it will have to wait for ngnet 1120346277 Q * rs Quit: rs 1120346378 M * _ag_ Bertl: the page on the wiki seems cool, is ngnet actually usable in ipv4? 1120346390 M * notasnark Bertl: Okay, I've changed sshd config on the host, and sshd now starts on the guest, and I can ssh to the guest server. Vserver isn't quite working how I expected it to work, but that's fine - I'll worry about that later. However... 1120346496 M * Bertl notasnark: yeah, folks usuall expect something they know from UML or similar 1120346505 M * notasnark ... when I start sshd with /etc/init.d/sshd start I get a whole load of errors: http://www.glendale.org.uk/~sam/vserver.txt 1120346544 M * notasnark Since sshd wasn't then starting, I assumed that these errors were causing it not to start. sshd now starts correctly despite these errors. 1120346569 J * mrplum ~m@digitalbrains.com 1120346578 M * Bertl those 'errors' are the result of the gentoo dependancy system for the sysv scripts 1120346597 M * Bertl basically you can not do hardware related stuff inside the guest 1120346616 M * Bertl (as it is no virtual machine, just an isolation) 1120346636 M * notasnark So they can be ignored? 1120346639 M * Bertl so if you clean up the guest scripts to _not_ attemt to do hardware related things 1120346651 M * Bertl then those errors will go away ... 1120346677 M * Bertl IIRC, the gentoo folks here have different solutions for that .. maybe you should have a look at the gentoo howto? 1120346690 M * notasnark The HOW-TO for vserver on Gentoo had me remove hardware related stuff from the syslog-ng config, but I guess it wasn't enough (or I did something wrong). 1120346706 M * notasnark I used http://www.gentoo.org/doc/en/vserver-howto.xml 1120346747 M * Bertl http://linux-vserver.org/GentooGuestHowto 1120346768 M * Bertl http://dev.gentoo.org/~hollow/vserver/docs/ 1120346805 M * Bertl (not sure they are up to date, just pasting from the linux-vserver page) 1120346837 M * notasnark I think the one I used is the uptodate version of the gentoo docs in hollow's homepage. 1120346939 M * mrplum In my Debian vserver I just setup, when I enter into it all my variable from the user I entered from get moved into the jail, is that normal? 1120346951 M * mrplum and how do I access cdrom devices? do I need to mount it outside the jail and someone link it in? 1120346970 M * mrplum tried making the hd* nodes, but it gave me not permitted errors 1120346981 M * notasnark Anyway, thanks for the tips. It's past my bedtime so I'm going to think again on it in the morning. At least I have a better idea of how things look from the network side. Cheers Bertl. 1120347015 M * Bertl have fun! 1120347026 M * notasnark :-) 1120347555 J * complexho ~mark@funk.gotadsl.co.uk 1120347575 M * Doener back again 1120347592 M * complexho hi Doener :) 1120347636 M * Doener Bertl: hm, Hollow created vinfo in libvserver, which looks like some central "get information" tool... do you think that's a good idea? IMHO the "get" stuff should be in those programs that also do the "set" stuff 1120347668 M * Doener i.e. "vinfo --xid 123 --vhi" vs. vuname --xid 123 --get 1120347699 M * Bertl well, are we talking about a tool or a library? 1120347725 M * Doener there are both in the current libvserver repo 1120347739 M * Bertl for the tools, I have no problem with a vinfo tool, but that doesn't mean that vuname should not 'support' the vuname specific gets 1120347767 M * Bertl IMHO the best solution would be a _single_ multiname tool ... 1120347774 M * Bertl (similar to busybox) 1120347797 Q * shuri Remote host closed the connection 1120347909 M * Bertl Vudumen_: got some issues (seemed scsi related) with the sun 1120347964 M * Bertl _ag_: you are editing the ngnet testing howto, right? 1120347972 M * Vudumen_ Bertl: hmm? i check it 1120347990 M * Vudumen_ hmm can you ssh to the machine? 1120348010 M * Bertl not atm, I had to reboot it via sysrq, because the fs was missing :) 1120348031 M * Bertl [ 2109.892247] scsi0 (0:0): rejecting I/O to offline device 1120348039 M * Vudumen_ Bertl: well i'll be back at the office at wednesday. i'll fix it or if my collegaue have time he will fix it 1120348047 M * Vudumen_ ahh fine :( 1120348072 M * Bertl np, it booted fine with the recent release ... so that seems okay 1120348091 M * Greek0 Bertl: could you perhaps put online the config of your example-vserver you had on your notebook at the presentation? 1120348111 M * Bertl the guest config? 1120348116 Q * mrplum Remote host closed the connection 1120348124 M * Greek0 the /etc/vserver/example directory 1120348140 M * Bertl it's the default config created by util-vserver 1120348155 M * Bertl (you can recreate it with the skeleton build method) 1120348166 M * Greek0 but you had stuff like /proc/version mangeling iirc? 1120348187 M * Bertl hmm, right ... did some echo commands ... 1120348225 M * Greek0 I already thought at your presentation that you used some nice features there. 1120348247 M * Bertl yes, I can upload it tomorrow evening (when I get back to my laptop) 1120348257 M * Greek0 would be great, thanks 1120348260 M * Bertl hmm, this evening actually 1120348280 M * Bertl what 'features' are you looking for? 1120348422 M * Greek0 the version mangeling was the thing I remembered right now, but the coolest thing was probably the various per-context limits, including the cpu limits via your scheduler 1120348511 M * Bertl ah, okay, for the version stuff: /etc/vservers//uts/release 1120348572 M * Bertl limits can be set with /etc/vservers//rlimits/{cpu, fsize, data, stack, core, rss, nproc, nofile, memlock, as, locks} 1120348623 M * Bertl and the cpu limits can be adjusted via /etc/vservers//schedule 1120348637 M * Bertl (see the flower page for further details) 1120348667 M * Greek0 yup, I saw it in the flower page now. I just thought it would probably be interesting to study your config and learn by example 1120348687 M * Bertl yeah, as I said, this evening I can upload it ...