1156809619 M * cehteh they have to, i didnt configured any other interface 1156809663 M * ntrs Bertl, can we have the named (bind) hack with the capabilities in stable? What would that involve if possible? 1156809682 M * cehteh well the 2nd is just a test if its possible to make a virtual circruit between 2 vservers without a physical nic 1156809724 M * cehteh while the first is really a bridge which bundles a tap and a eth and the vserver aliases the bridge .. but i can tell you later how well it works 1156809747 M * Bertl ntrs: not in the stable 2.0 branch, but you can hire somebody to backport that feature 1156809777 M * ntrs Bertl, is it too complicated to be available in stable branch as well? 1156809790 M * Bertl it will likely be part of the stable 2.2 branch once it is out 1156809822 M * cehteh Bertl: could the thing be faked for the named by a LD_PRELOAD lib? 1156809846 A * cehteh thought about such as general solution for broken binaries 1156809852 M * Bertl cehteh: simply recompiling bind helps 1156809885 M * cehteh yes i know .. but if we could ship a LD_PRELOAD lib with the utils it would be even easier for users 1156809910 M * ntrs Bertl, it is not practical to rebuild a gazzillion binds. 1156809960 M * cehteh and we are in control of fixing broken apps without relying on other upstream programmers fixing the packages 1156809994 M * Bertl preload would probably work if you block the cap syscalls 1156810005 M * cehteh block or fake 1156810023 M * Bertl well, yeah, basically overload and filter :) 1156810043 M * cehteh maybe i give it a try next time 1156810054 M * Bertl but I think 2.2 will provide a viable solution in the near future 1156810088 M * cehteh yeas but a lib can fix some more things 1156810111 M * cehteh like servers which bind 127.0.0.1 or other nastiness 1156810136 M * cehteh anyways do you have an idea when 2.2 will be ready? 1156810161 M * Bertl will take at least a month from now 1156810200 M * ntrs Bertl, ok, I can wait until end of september 1156810225 M * ntrs I'll use the stable in the meantime. 1156810238 M * Bertl testing the feature in development is a good idea 1156810263 M * Bertl because if it isn't tested/used, it cannot mature and get into stable .. 1156810388 A * cehteh think he wont like to use the full development version on his production server .. if it would be backported or such features would be developed in branches it would be much better 1156810575 M * Bertl we already have three branches, it doesn't scale if we make a new branch for every tiny feature 1156810589 M * cehteh which scm do you use? 1156810593 M * cehteh git? 1156810606 M * Bertl something like that, soon it will be git 1156810631 M * cehteh (well git isnt really a scm, but ok for kernel development) 1156810688 M * cehteh but branching is cheap there having small branches for single features could be possible ... 1156810820 M * cehteh oh fine, i found a german howto for all 3 kinds of interface mapping ... 1156810832 M * cehteh (modutild, udev, mac-address) 1156810936 M * cehteh 4th kind using 'nameif' 1156810938 M * Bertl yeah, but it's not worth the efford, unless you _know_ that some folks are going to extensively test that 1156811073 M * cehteh /etc/mactab ... sounds good 1156811165 M * cehteh man nameif .. cool 1156811601 M * cehteh ifconfig | sed -e 's/\(eth.\).*HWaddr \([0-9A-F:]*\)/\1 \2/p;d' | sort | uniq >/etc/mactab 1156811792 M * ntrs Bertl, we don't have to make new branches. It could just be another feature that is in the official 2.0.2 stable 1156812166 M * ntrs Bertl, is BME part of stable 2.0.2-rc31? 1156812341 J * ntrs_ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1156812341 Q * ntrs Read error: Connection reset by peer 1156812360 M * ntrs_ Bertl, are you there? 1156812491 M * Bertl yep 1156812507 M * Bertl BME was included some time ago 1156812539 M * Bertl ntrs_: features do not get stable by people wnating them to be in stable :) 1156812567 M * ntrs_ Bertl, I agree. Understood. But people can have wishes, too. 1156812571 M * Bertl features get stable by folks _testing_ them extensively in devel, and providing reasonable feedback 1156812591 M * Radiance BerI, you're visiting ccc december ? :) 1156812591 M * ntrs_ Ok, then devel is having some serious problems right now. 1156812600 M * Bertl yes, you can _always_ cook up your own branch and add patches from devel 1156812617 M * Bertl Radiance: not sure yet, locations? 1156812647 M * Bertl ntrs_: have you checked that? i.e. did you compare it to a stable branch on the same machine/setup? 1156812679 M * Bertl ntrs_: the only thing I saw so far is a process (qmail) which did encounter some issues with locks 1156812681 M * ntrs_ Bertl, yes the stable I used previously 2.0.2-rc20 did not have any problems. 1156812705 M * Bertl well, that was a completely different kernel back then 1156812723 M * Bertl don't forget about mainline changes, and you adjusted the scheduler stuff too, no? 1156812752 M * Bertl and aside from that, we do not even know if there is a real issue 1156812773 M * Bertl could as well be that 'something' changed in the guest's behaviour 1156812785 M * Radiance BertI, i hope you do would be nice to say hi in real :) 1156812836 M * Bertl Radiance: sure, maybe I'll be there ... 1156812884 M * Radiance as far as i read the location will be Berlin 1156812910 M * Bertl ntrs_: for example, I have a devel guest running mysql quite fine, no issues, no problems so far, it's just a little faster and smother with devel than it was with stable 1156812926 M * Bertl ntrs_: doesn't mean there is no issue, after all it's devel :) 1156814655 P * nachoguy Leaving 1156815713 J * babyfish admytren@alumni.engin.umich.edu 1156815802 M * babyfish Hi matti 1156815818 M * matti Hi. 1156815830 M * Bertl welcome babyfish! 1156815840 A * matti is little amazed... 1156815855 M * babyfish Thanks! I found the channel through "step by step guide" 1156815864 M * matti No one never says hello to me at the first place :) 1156815877 M * babyfish :) 1156815894 M * matti babyfish: Good to hear :) 1156815923 M * babyfish Well, I've ran into some problems with vserver and was wondering if I could get some help or pointers 1156815935 M * matti babyfish: Just ask. 1156815982 M * matti ;) 1156815996 M * babyfish I'm trying to backport vserver 2.1.0 patch to 2.6.11 kernel. Started with 2.6.14 diff and manually patched rejected chunks 1156816019 M * matti I need to ask. 1156816026 M * matti Why to 2.6.11? 1156816029 M * matti It is not too old? 1156816064 M * babyfish Yeah, it's a bit too old 1156816087 M * babyfish But I have some patches that have not moved to 2.6.14 yet 1156816090 M * matti Problably not very secure also - due to numerous vulnerabilities since 2.6.11... 1156816156 M * matti babyfish: You may try to apply them on to 2.6.17.11 maybe? 1156816167 M * matti s/Problably/Probably/ 1156816230 M * babyfish I've tried that but ran into some problems - backporting vserver was easier... 1156816278 M * matti Hm. 1156816280 M * babyfish Up until getting vserver-tools to run :-( 1156816315 M * matti Please, be more specific :) What happen'? 1156816322 M * matti :) 1156816323 M * babyfish Ok: 1156816340 M * babyfish Basically chcontext and chbind fail 1156816356 M * babyfish chcontext complains it doesn't see kernel symbol: 1156816361 M * matti babyfish: Did you try Bertl's "testme.sh"? 1156816367 M * babyfish chcontext: vc_new_s_context(): Function not implemented 1156816379 M * babyfish However the symbol is there: 1156816388 M * matti babyfish: I assume, that you've latest utils, but old kernel, right? 1156816439 M * babyfish cat /proc/kallsyms | grep vc_new_s_context 1156816441 M * babyfish c014a920 T vc_new_s_context 1156816450 M * babyfish Yes - you can say so 1156816465 M * babyfish I'm using util-vserver-0.30.210 1156816473 M * babyfish Vserver patch 2.1.0 1156816479 M * babyfish From 2.6.14 kernel 1156816510 M * matti babyfish: Kernel compilation finished without any serious warnings? Error not occured I pressume. 1156816521 M * babyfish Yep 1156816547 M * babyfish I didn't bring in a bun of stuff from non-x86 arch's and no ext2/ext3 1156816553 M * babyfish Just x86 on reiserfs 1156816629 M * matti babyfish: http://vserver.13thfloor.at/stuff/script/testme.sh - try this, and paste output somewhere. 1156816647 M * matti Bertl: Any clue on babyfish's issue? 1156816709 M * babyfish just a sec... 1156816728 M * matti Take your time :) 1156816730 M * matti No rush. 1156816731 M * matti :> 1156816742 Q * s0undt3ch Ping timeout: 480 seconds 1156816778 M * babyfish bash-2.05b# ./testme.sh -v 1156816778 M * babyfish Linux-VServer Test [V0.15] Copyright (C) 2003-2006 H.Poetzl 1156816778 M * babyfish chcontext: vc_new_s_context(): Function not implemented 1156816778 M * babyfish chcontext failed! 1156816778 M * babyfish chbind: kernel does not provide network virtualization 1156816779 M * babyfish chbind failed! 1156816779 M * babyfish chcontext-compat 0.30.210 -- allocates/enters a security context 1156816781 M * babyfish This program is part of util-vserver 0.30.210 1156816781 M * babyfish Copyright (C) 2003,2004 Enrico Scholz 1156816783 M * babyfish This program is free software; you may redistribute it under the terms of 1156816783 M * babyfish the GNU General Public License. This program has absolutely no warranty. 1156816785 M * babyfish Linux 2.6.11.11-vs2.1.0 #2 Mon Aug 28 18:22:11 PDT 2006 i686 1156816785 M * babyfish Ea 273/glibc (DSa) 1156816787 M * babyfish VCI: 0001 273 03000116 (KtTbLgnP) 1156816787 M * babyfish (babyfish@babyfish-box) 1156816789 M * babyfish (gcc version 3.4.4 20050721 (Red Hat 3.4.4-2)) 1156816789 M * babyfish #2 Mon Aug 28 18:22:11 PDT 2006 1156816830 M * Bertl reconfig tools with --enable-apis=NOLEGAY 1156816838 M * Bertl *NOLEGACY 1156816875 M * matti Bertl++ 1156816905 M * babyfish Ok, trying that right now 1156816935 M * matti Bertl: I hope, I don't disturb you :) 1156816958 M * Bertl np 1156817105 M * matti :) 1156817257 M * babyfish Thanks Bert! But unfortunately still no go: 1156817280 M * Bertl let's upload the testme.sh output to paste.linux-vserver.org 1156817315 M * babyfish bash-2.05b# ./testme.sh 1156817315 M * babyfish Linux-VServer Test [V0.15] Copyright (C) 2003-2006 H.Poetzl 1156817315 M * babyfish chcontext: vc_new_s_context(): Function not implemented 1156817315 M * babyfish chcontext failed! 1156817315 M * babyfish chbind: vc_set_ipv4root(): Function not implemented 1156817317 M * babyfish chbind failed! 1156817317 M * babyfish Linux 2.6.11.11-vs2.1.0 #2 Mon Aug 28 18:22:11 PDT 2006 i686 1156817319 M * babyfish Ea 273/glibc (DSa) 1156817319 M * babyfish VCI: 0001 273 03000116 (KtTbLgnP) 1156817319 M * matti babyfish: No, no ;p 1156817321 M * babyfish ok 1156817325 M * matti ;) 1156817330 M * babyfish Sorry :) 1156817357 M * Bertl did you disable any legacy options on the kernel build? 1156817379 M * Bertl doesn't look so from the VCI, which looks a little broken, btw 1156817474 M * babyfish CONFIG_VSERVER_LEGACY=y 1156817494 M * babyfish CONFIG_VSERVER_LEGACYNET=y 1156817584 J * s0undt3ch ~s0undt3ch@bl7-240-60.dsl.telepac.pt 1156817823 M * babyfish I'm a noob to kernel programming, so pardon me if this is a silly question. If the symbol is available in /proc/kallsyms, shouldn't it be available as a system call? 1156817840 M * Bertl no, not related 1156817856 M * Bertl linux-vserver uses a single system call 1156817866 M * Bertl it is called sys_vserver 1156817881 M * Bertl the vc_* are syscall commands 1156817902 M * Bertl they are multiplexed into the one syscall 1156817918 M * babyfish Ok, I see 1156817931 M * Bertl the fact that the symbol is there in the ksyms just means that the legacy is enabled 1156817949 M * Bertl why did you backport that btw? 1156817961 M * Bertl and what are the other patches/base kernel? 1156817985 M * Bertl because I think you might have messed up in the process with the syscall itself 1156818010 M * Bertl to me it looks like the entire syscall is disabled 1156818049 M * babyfish Ugh, there is a bunch of patches for x86 board I'm working with (flash, diagnostics, etc). Not an easy port forward 1156818082 M * Bertl do you have a .diff for the vserver part you are using right now? 1156818103 M * Bertl (i.e. some unified diff I could look at :) 1156818202 M * babyfish At the moment it's a mess :) I've started with patch-2.6.14.4-vs2.1.0.diff and worked through rejects 1156818249 M * babyfish Let me double check that vserver syscalls work and get my diff ready. I should come better armed when asking for help :) 1156818259 M * Bertl are you sure that a recent kernel does not work on your hardware, btw? 1156818300 M * babyfish Yeah, it requires some of the device driver patches 1156818337 M * babyfish Unfortunately :-( 1156818430 M * babyfish Matti, Bertl - thank you for your help. I will double check that syscall itself works (I think all of my vserver commands are failing - so this is a good area to concentrate on) 1156818479 M * matti Hm. 1156818480 M * Bertl you can check with the vcmd tool quite easily 1156818489 J * DreamerC_ ~dreamerc@61-217-226-135.dynamic.hinet.net 1156818492 M * Bertl also enabling syscall debugging can help a lot 1156818536 M * matti babyfish: Welcome, but I do nothing exactly. You need to thank Bertl mostly :) 1156818581 M * Bertl well, you did ask a lot of initial questions, which helped to get through to the real issue 1156818601 M * Bertl so, at least from my side, your help was appreciated :) 1156818603 Q * DreamerC Ping timeout: 480 seconds 1156818631 M * babyfish Sorry, one more question - where can I get vcmd? 1156818660 M * Bertl http://vserver.13thfloor.at/Experimental/TOOLS/vcmd-0.05.tar.bz2 1156818698 M * babyfish Thanks! Matti Bertl you have both been of great help! 1156818708 P * babyfish 1156819742 Q * matti Ping timeout: 480 seconds 1156820416 J * matti matti@linux.gentoo.pl 1156820582 M * matti Eh. 1156821913 Q * id23 Ping timeout: 480 seconds 1156822441 J * id23 ~id@p508112D5.dip0.t-ipconnect.de 1156822789 Q * derjohn Ping timeout: 480 seconds 1156823156 J * derjohn ~derjohn@80.69.37.19 1156823395 M * Bertl off to bed now ... have a good one everyone! 1156823406 N * Bertl Bertl_zZ 1156823859 J * ntrs ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1156823859 Q * ntrs_ Read error: Connection reset by peer 1156823923 M * essobi_ Anyone used any of the web interfaces for Vserver? 1156824122 M * cehteh isnt there only one? :) 1156824163 M * cehteh anyways, i dont use it but i'd seen the demo on linuxtag and it looked very promising 1156824247 M * essobi_ nah there's two with very similar names.. 1156824358 M * essobi_ openvcp 1156824363 M * essobi_ and openvps 1156824530 M * essobi_ openvps seems to be primarily all python scripts 1156824535 M * essobi_ and mod_python 1156824640 M * cehteh ah ok 1156824679 M * cehteh do you want to setup a vserver farm or just administrate a single machine? 1156824757 M * cehteh i think the openvpc was the one which handles that and which i looked at Linuxtag 1156826777 M * essobi_ They both appear to do farms.. 1156826780 M * essobi_ Which I need.. 1156826792 M * essobi_ I'd rather not piss w/python ATM thou.. 1156826824 M * essobi_ vcp is wrote in php but appears to have a few less features.. 1156826850 M * daniel_hozac VCP is a combo of PHP and some horrible C. 1156826894 M * essobi_ Heh.. Yea? 1156826902 M * essobi_ The daemon is that busted? Heh. 1156826935 M * daniel_hozac at least the last time i looked at it. 1156826956 M * essobi_ nice... 1156827106 M * essobi_ Ehh... it's tiddly-winks C-code.. 1156827120 M * essobi_ Simple, and basic so it would appear. 1156827190 M * essobi_ Bejezus.. 1156827192 M * essobi_ while(1) 1156827209 M * essobi_ Are you supposed to do that in C? I do that in Bash. :) 1156827254 Q * derjohn2 Ping timeout: 480 seconds 1156827417 M * essobi_ Ah, it's a select loop. 1156827664 J * derjohn2 ~aj@dslb-084-058-215-186.pools.arcor-ip.net 1156829757 J * dna ~naucki@38-243-dsl.kielnet.net 1156829758 M * dna in 1156830015 Q * PowerKe Quit: Oops, wrong button 1156831886 J * cdrx ~legoater@cap31-3-82-227-199-249.fbx.proxad.net 1156833432 M * cdrx morning europe 1156834991 M * Wonka GOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOD MORNING VIETNAAAAAAAAAAAAAAAAAAAAAAAAAAAAM! 1156836394 M * nayco_work Hello, all ! 1156836844 J * meandtheshell ~markus@85-124-36-161.dynamic.xdsl-line.inode.at 1156837798 M * harry there is a new grsec for 2.6 kernels btw... 1156837880 M * harry where iptables stealth works ;) 1156839510 M * nib-nico I haven't found some documentation of the possible files in the /etc/vservsers// direcotry, especially limits (cpu, memory) do you know where I can find an overview or something? 1156839594 Q * FireEgl Ping timeout: 480 seconds 1156839676 J * ||Cobra|| ~cob@pc-csa01.science.uva.nl 1156839731 M * tanjix Dears, when I compile the vserver-uztils with --enable-apis=ALL then i get errors when running make 1156839738 M * tanjix is that a known issue ? 1156839947 M * tanjix nib-nico: the flower page has all that information for you: http://www.nongnu.org/util-vserver/doc/conf/configuration.html 1156839995 M * nib-nico http://www.nongnu.org/util-vserver/doc/conf/configuration.html 1156839999 M * nib-nico sorry 1156840023 M * nib-nico Ahh thank you, time to read 1156840044 M * coocoon tanjix: with the latest util-vserver version this is iknown, why are u using --enable-apis=ALL and why not --enable-apis=NOLEGACY 1156840431 M * coocoon tanjix: sorry i mean the known problems are with the api settings in the kernel ;-) 1156840471 M * tanjix coocoon: i did it with NOLEGACY and that worked through fine 1156840497 M * coocoon ok that i use also if i compile it by hand 1156840506 M * tanjix coocoon: But now when trying to start a vserver i get errors :( 1156840516 M * tanjix that /var/run(utmp could not be found although this exists 1156840540 M * coocoon u have disabled all api settings in the kernel 1156840552 M * tanjix mh 1156840564 M * coocoon that is important 1156840590 M * coocoon and does the guest work before 1156840604 M * tanjix there is nothing disabled? 1156840619 M * coocoon tanjix: or more exactly not set 1156840650 M * tanjix [*] Enable Legacy Kernel API 1156840662 M * coocoon ok there must set nothing 1156840677 M * coocoon wait 1156840781 M * tanjix coocoon: http://paste.linux-vserver.org/320 1156840851 M * coocoon tanjix: my settings int the kernel config http://paste.linux-vserver.org/321 1156840861 M * coocoon only have a look at the apis 1156841017 M * coocoon and after that maybe this will also help ./configure --prefix=/usr --mandir=/usr/share/man --sysconfdir=/etc --localstatedir=/var --with-vrootdir=/vservers --enable-apis=NOLEGACY 1156841054 M * coocoon tanjix: make && sudo make install && sudo make install-distribution 1156841184 M * coocoon nib-nico: here u will found something i think http://linux-vserver.org/some_hints_from_john and here http://linux-vserver.org/Documentation 1156841366 M * tanjix coocoon: am i right, that i do the kernel by make && make modules && make modules_install && make install? 1156841403 M * coocoon tanjix: yes thats the way i do it also 1156841442 M * tanjix ok, then i wll do a reboot of my machine now and lets see what happens then :) 1156841510 M * tanjix finished 1156841525 M * nib-nico coocoon: The flowers page is the right thing in this moment,, but thankls anyway, the FAQ seems also interesting. Tho Documenationpage I know already but I haven't found such an overview tehre 1156841530 M * tanjix now rebuilding the tools on your way 1156841661 M * tanjix coocoon: finished too 1156841661 M * tanjix coocoon: new errors again :( 1156841678 M * tanjix http://paste.linux-vserver.org/322 1156841954 M * coocoon what kind of guest is that which ut tried to starthave u build it by urself 1156841997 M * coocoon have reboot the system with the new kernel settings 1156842045 M * coocoon *u 1156842106 M * tanjix yes i rebooted 1156842125 M * coocoon and then u have compiled the util-vserver tool 1156842127 M * tanjix that is an ubuntu image i got from http://mirrors.sandino.net/vserver/images/ 1156842157 M * coocoon hm which distro u r using on the host 1156842195 M * tanjix i found the error :( i forgot to mount the lvm volume :) 1156842211 M * coocoon good if u found it ;-) 1156842251 M * tanjix but with volume mounted chcontext still says function not implemented 1156842268 M * tanjix when i try to start that server 1156842269 M * coocoon could u past it 1156842275 M * tanjix sure 1156842286 M * tanjix chcontext: vc_new_s_context(): Function not implemented 1156842323 M * coocoon tanjix: do u use dynamic contexts or fixed contexts 1156842366 M * tanjix coocoon: i think they should be dynamic 1156842384 M * coocoon tanjix: hm it is better to use fix contexts, sorry 1156842392 M * coocoon ;-) 1156842405 M * coocoon this will be obsolete in the next kernel patches 1156842414 M * tanjix S_CONTEXT="xxx" in the config file? 1156842424 M * coocoon also the api settings will go away 1156842489 A * harry wants 2.0.2 vserver releaaaaaaaaaaaase (please ;)) 1156842509 M * coocoon tanjix: which kernel version do u use 1156842519 M * coocoon and which patch 1156842527 M * coocoon could u paste vserver-info 1156842534 M * tanjix 2.6.14.3-vs2.0.1 1156842555 M * tanjix http://paste.linux-vserver.org/323 1156842593 M * tanjix based on the doc @ http://linux-vserver.org/Step-by-Step+Guide+2.6 1156842660 M * coocoon ah ok dunno it exactly but i think there must be also a setting in the kernel where u can set under vserver dynamic context, this must be disabled and then u set in /etc/vservers/foo/context --> e.g. 55 1156842670 M * coocoon after reboot it must work 1156842716 M * tanjix coocoon: that would work with the new config style method 1156842723 M * tanjix i am using the old style with only one config per vserver 1156842783 M * coocoon ah ok why????? 1156842810 M * coocoon have never used it sorry so it could be with the sett9ng u have said ;-) 1156842816 M * nayco_work Heyall, i'm writing a french article for 2.0.2 release, to be published on a large audience site... Understand that I musn't make any mistake :D ! Well, there are a couple of terms I cannot axplain : 1156842834 M * nayco_work "bind mount extensions" 1156842852 M * nayco_work "kernel helper imporvement" 1156842870 M * nayco_work Could I have details about these items, please ? 1156842913 M * tanjix coocoon: because i dont really understand the new config style and how to get it to work 1156843178 M * coocoon tanjix: ok for sure i dunno what the problem there is because of i haven't had probs with that ever and if i understand that u will, too ;-) 1156843202 M * coocoon but ok i have said it is better to use fixed cotextssaid 1156843213 M * coocoon *contexts 1156843410 M * coocoon must correct my sely not fixed context it is called static context 1156843846 M * tanjix with S_CONTEXT="xx" i aso get "not implemented" 1156843945 J * lilalinux ~plasma@h1-gw.of.net-lab.net 1156844235 J * lilalinux_ ~plasma@dslb-084-058-252-175.pools.arcor-ip.net 1156844237 M * coocoon could u paste it here http://paste.linux-vserver.org please 1156844274 M * tanjix the config ? 1156844282 M * coocoon ther error 1156844309 M * tanjix http://paste.linux-vserver.org/324 1156844342 M * coocoon and the config gile 1156844378 M * coocoon *file 1156844395 M * tanjix http://paste.linux-vserver.org/325 1156844403 Q * Johnnie Read error: Connection reset by peer 1156844476 M * coocoon what happens if the conf file is away 1156844480 Q * shedi Quit: Leaving 1156844509 M * tanjix Can not find a vserver-setup at '/etc/vservers/ubuntu/'. 1156844575 M * coocoon tanjix: hm sorry i think that it is more compilcated with the conf file but dunno can't help u with that as i said have never used it and will never used it ;-) 1156844666 Q * lilalinux Ping timeout: 480 seconds 1156844681 M * tanjix but you can tell me how to create a working vserver setup fot the new style ? 1156844694 M * tanjix i created a directory /etc/vservers/ubuntu/ 1156844697 M * tanjix and then` 1156844699 M * tanjix ? 1156844709 M * coocoon hm 1156844728 M * coocoon tanjix u will use a tarballed image 1156844799 J * Johnnie ~jdlewis@static-acs-24-154-32-33.zoominternet.net 1156844870 M * tanjix sorry ? 1156844934 M * coocoon tanjix: for example like this "vserver foo build -m skeleton --hostname foo --interface eth0:192.168.0.11--context 66 --initstyle plain", then u must extract the tarball into the tarball into /vservers/foo 1156844949 M * coocoon -tarball +dir 1156845098 Q * DreamerC_ Quit: leaving 1156845113 J * DreamerC ~dreamerc@61-217-226-135.dynamic.hinet.net 1156845316 Q * DreamerC 1156845404 Q * Johnnie Read error: Connection reset by peer 1156845412 J * DreamerC ~dreamerc@59-115-51-158.dynamic.hinet.net 1156845486 J * Johnnie ~jdlewis@static-acs-24-154-32-33.zoominternet.net 1156845562 M * coocoon tanjix: more examples how to setup vserver (e.g. ubunut,...) u will find here http://linux-vserver.org/Documentation #Distribution specific Howtos 1156845912 Q * kir Ping timeout: 480 seconds 1156846276 M * tanjix coocoon: i could solve my issue because of the error message using context 1156846288 M * tanjix i had changed an option in the kerne-rebuilt it and it works 1156846301 M * coocoon tanjix: fine 1156847493 J * kir ~kir@swsoft-mipt-nat.sw.ru 1156847735 J * shedi ~siggi@dsl-149-109-85.hive.is 1156850778 M * harry how do i enable cpu scheduler? 1156850806 M * harry boromir:/usr/local/etc/vservers/luditdb# cat flags schedule 1156850806 M * harry sched_prio 1156850806 M * harry 7 1156850806 M * harry 2 1156850806 M * harry 100 1156850809 M * harry 10 1156850811 M * harry 1000 1156850814 M * harry dummy 1156850825 M * derjohn 7 2 is noz good 1156850828 M * derjohn not 1156850845 M * derjohn you get new 7 jiffies all 2 jiffies 1156850853 M * derjohn so they never get empty := 1156851074 M * harry yeah... if i use 100% cpu, 4 times 1156851075 M * derjohn harry, read this night chatlog. Bertl_zZ explainted pretty good how the sched-foo works. 1156851085 M * harry it still gives me 100% on all 4 cpus 1156851094 M * harry but now i enforced hard cpu scheduling :) 1156851099 M * harry and it works fine :) 1156851106 M * derjohn harry, do you know what that is? 1156851121 M * harry i think so 1156851135 M * derjohn I means the guests will be completely _halted_ until min_bucks it reached again 1156851137 M * harry are there onlnie logs? 1156851160 M * derjohn sure 1 1156851161 M * harry derjohn: i know... i just dont want ANY vps to be able to lock up the system :) 1156851173 M * harry this was just to test the parameters 1156851189 M * derjohn harry, you can do that with prio too I think 1156851209 M * derjohn hard is shit, if you use ssh inside the guest 1156851220 M * harry juh ;) 1156851226 M * derjohn http://irc.13thfloor.at/LOG 1156851241 M * derjohn harry, how many HZ has you kernel ? 1156851267 M * harry CONFIG_HZ=250 1156851487 M * derjohn 10*7/2 = 140 at 250 HZ thats a 140/250=0.56 second pause. 1156851640 M * derjohn (harry, if you swapped the 7 with 2 already) 1156851985 M * harry http://paste.linux-vserver.org/326 1156851990 M * harry any clues ? :) 1156851996 M * harry ? 1156852000 M * harry swap? why should i swap? 1156852087 Q * michal_ Ping timeout: 480 seconds 1156852093 M * harry i want 3,5 of 4 cpus available to the host 1156852109 M * harry so every 2 jiffies, 7 units 1156852443 J * michal_ ~michal@www.rsbac.org 1156852774 M * harry Linux boromir 2.6.17.11-grsec2.1.9-vs2.0.2-rc31 #5 SMP Tue Aug 29 11:06:42 CEST 2006 i686 GNU/Linux 1156852777 M * harry wiiiiiiiiii :) 1156853038 M * Wonka Wii? 1156853860 Q * cdrx Ping timeout: 480 seconds 1156856099 M * nayco_work Bertl_zZ , Hollow : I've got a couple of questions for you, in order to finish my article for 2.0.2 release... 1156856436 J * cdrx ~legoater@242.32.96-84.rev.gaoland.net 1156856637 Q * cdrx Remote host closed the connection 1156856913 J * cdrx ~legoater@242.32.96-84.rev.gaoland.net 1156857684 M * cehteh derjohn: got my answer in the morning? 1156857697 M * derjohn cehteh, i saw it , yes 1156857705 M * derjohn did it work ? 1156857729 M * cehteh not yet rebootet .. but i guess it should work 1156857736 M * derjohn :) 1156857737 M * derjohn k 1156857765 M * derjohn harry, harrymir-rc31 ;) 1156857825 M * derjohn harry, why would you allocate 7 jiffies all 2 jiffies? 1156857855 M * derjohn (hm, multi cpu ? I dunno how the jiffies are handled there I admit9 1156860357 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1156861801 M * daniel_hozac nayco_work: BME lets you do bind mounts with options differing from that of the original mount, e.g. mount -o bind,ro ... 1156861965 M * Radiance hi guys 1156862002 M * Radiance i noticed that /tmp is mounted seperately in the vps ... i noticed it has only 16mb of space which is full and certain installations fail in the vps because there is no space on /tmp 1156862005 M * Radiance how is this possible ? 1156862015 M * Radiance i thought /tmp is part of the vps space 1156862037 M * harry derjohn: because of multi-cpu, yes 1156862056 J * FireEgl FireEgl@Sebastian.Atlantica.US 1156862072 M * derjohn Radiance, /etc/vservers//fstab 1156862105 M * derjohn harry, so we have "4 parallel working jiffies" on a quad cpu ... 1156862126 M * harry if i understand the scheduler correctly: yes 1156862134 M * harry kinda kewl, huh :) 1156862197 M * Radiance ah thanks ! 1156862216 M * Radiance guess i'll have to restart the vps to make the change active 1156862746 J * lilalinux__ ~plasma@dslb-084-058-206-238.pools.arcor-ip.net 1156862748 Q * s0undt3ch hydrogen.oftc.net kinetic.oftc.net 1156862748 Q * cehteh hydrogen.oftc.net kinetic.oftc.net 1156862798 J * s0undt3ch ~s0undt3ch@bl7-240-60.dsl.telepac.pt 1156862798 J * cehteh ~ct@cehteh.homeunix.org 1156862850 Q * cdrx Read error: Connection reset by peer 1156862860 J * cdrx ~legoater@242.32.96-84.rev.gaoland.net 1156862928 M * nayco_work daniel_hozac: Thanks, now I understand ! 1156862980 Q * lilalinux_ Read error: Connection reset by peer 1156863032 M * cehteh http://www.pipapo.org/pipawiki/PasteBin/NameIf ... still not rebootet/tested but if that works (which i suspect) then i tihnk i solved this mac/eth mapping thing really aautomagically now 1156863035 M * derjohn Radiance, if you a cool geek, you can do that on the fly. but restarting is more comfortable. 1156863051 M * nayco_work daniel_hozac: And for the kernel helper improvements ? 1156863387 M * derjohn Radiance, you can remount stuff within the guest's namespace with vnamespace cmd: # vnamespace -e XID mount -t tmpfs -o remount,size=256m,mode=1777 none /var/lib/vservers//tmp/ 1156863411 M * derjohn I might not update the mtab within the guest (i dunno remeber) 1156863815 M * Radiance ok thanks, i noted that for the next time hehe 1156864086 Q * cehteh charon.oftc.net kinetic.oftc.net 1156864086 Q * s0undt3ch charon.oftc.net kinetic.oftc.net 1156864098 J * s0undt3ch ~s0undt3ch@bl7-240-60.dsl.telepac.pt 1156864098 J * cehteh ~ct@cehteh.homeunix.org 1156864109 M * derjohn Radiance, it in the wiki, too. look for some_hints_from_john 1156864519 J * coocoon ~coocoon@p54A069AA.dip.t-dialin.net 1156864979 N * Bertl_zZ Bertl 1156864982 M * Bertl morning folks! 1156864991 M * tanjix morning Bertl ;) 1156864991 M * daniel_hozac morning Bertl! 1156865143 M * matti Hi Bertl. 1156865633 M * nayco_work 'Mornin', Bertl ! 1156865641 M * derjohn huhu Bertl ! 1156865660 M * nayco_work Bertl: Have you seen my previous messages ? 1156865678 M * ntrs Bertl, are you there? 1156865691 M * Bertl nayco_work: the 'Morning'? :) 1156865712 M * Bertl ntrs: yep, I'm here 1156865730 M * ntrs Is the memory split patch not in devel? 1156865744 M * Bertl the memory split was dropped, as it is now in mainline 1156865745 M * ntrs My kernel is only seeing 1 GB of ram and it should be seeing 2 GB 1156865771 M * Bertl i.e. you have to enable the intuitive 'EMBEDDED' option, then you can select that 1156865774 M * ntrs How do I enable it? 1156865792 M * ntrs Where is embedded 1156865828 M * Bertl 'make menuconfig' press '/' type EMBEDDED 1156865920 M * ntrs Then I want 3G/1G, right? 1156865944 M * ntrs or 2G/2G? 1156865949 J * stefani ~stefani@tsipoor.banerian.org 1156865972 M * matti ntrs: If you have 2 GB of memory, the VMSPLIT_2G. 1156865986 M * matti ntrs: If 1 GB, then probably VMSPLIT_3G_OPT. 1156865996 M * ntrs If I upgrade to 3GB however I need 3G/1G, right? 1156866027 M * matti ntrs: Well, then you probably will need to enable HIGHMEM and friends. 1156866035 M * ntrs What about 1G/3G? 1156866069 M * ntrs Previously I had CONFIG_VMSPLIT_1G=y 1156866074 M * ntrs What is the equivalent now? 1156866117 M * matti ntrs: If you do not want highmem and only precious lowmem, then VMSPLIT_* is handy. And remember, that setting other than VMSPLIT_3G on machine with PaX and segmexec can break things. 1156866166 M * ntrs I have no idea what PaX and segmexec is but ok. 1156866169 M * matti :) 1156866177 M * matti Oh, then we can skip this ;p 1156866185 M * ntrs Bertl, what is the equivalent now of the CONFIG_VMSPLIT_1G=y? 1156866218 M * daniel_hozac that is the option... 1156866243 M * matti ;-p 1156866256 M * matti I need to reboot with new kernel, back in the moment. 1156866260 M * ntrs Yes, but what is the equivalent option now? 1G/3G or 3G/1G? 1156866304 M * daniel_hozac what do you want? 3/1 is the default. 1156866307 M * matti ntrs: Well, in 2.6 with EMBEDDED enabled (or with vmsplit expose patch from -ck/-cks) all options should be available. 1156866313 M * daniel_hozac so 1/3 would be the opposite. 1156866317 Q * michal_ Ping timeout: 480 seconds 1156866349 M * ntrs Ok, I previously had CONFIG_VMSPLIT_1G=y with Bertl's patches. What would be the equivalent option now? 1156866366 M * ntrs daniel_hozac, are you saying that the equivalent is 3G/1G? 1156866376 M * matti ntrs: You have 2 GB of RAM, right? 1156866384 M * ntrs Yes, but I may upgrade to 3 soon. 1156866399 M * matti 3? Not 4? :) 1156866402 M * daniel_hozac so 1/3 and CONFIG_VMSPLIT_1G is what you want. 1156866403 M * ntrs I guess nobody knows what the equivalent option is then. 1156866411 M * matti Well, anyway. 1156866428 M * Bertl ntrs: you should listen carefully, daniel already said it twice 1156866429 M * ntrs CONFIG_VMSPLIT_1G is no longer valid, it was in the old kernels with Bertl's patches 1156866446 M * matti ntrs: Yes, Bertl got the point. 1156866450 M * Bertl 1G/3G user/kernel split is the closes to CONFIG_VMSPLIT_1G you get 1156866463 M * ntrs I am listening, it anyone would have answered directly. 1156866478 M * matti Bertl: IIRC 1/3 is not really a good choice? Correct me if I am wrong. 1156866483 M * ntrs Bertl, now that was a direct answer. Thank you very much. 1156866522 M * Bertl matti: for a larger x86 based server, it's the best choice you have without using highmem 1156866537 M * Bertl ntrs: also disable secure computing, it only eats up cpu 1156866546 M * ntrs secure computing? 1156866605 M * ntrs where is secure computing? 1156866611 J * michal_ ~michal@www.rsbac.org 1156866616 M * Bertl it's a (non)feature one lkml member got into mainline 1156866660 M * Bertl ntrs: '/' type SECCOMP 1156866664 M * ntrs ok 1156866671 M * ntrs Is 2G/2G also a safe choice? 1156866679 M * Bertl with 2GB of ram, yes 1156866703 M * ntrs seccomp is disabled 1156866723 M * ntrs one last question, will 1g/3g allow for an upgrade to 3 GB? 1156866731 M * ntrs with no highmem? 1156866748 M * daniel_hozac that's kind of the point, yes. 1156866757 M * ntrs OK, thanks daniel_hozac 1156866782 M * ntrs Oh, one more thing, what is the problem that matti mentioned about 1g/3g not being a good choice? 1156866795 M * daniel_hozac it's a grsec thing. 1156866803 M * daniel_hozac (or?) 1156866810 M * Bertl you have to ask matti for that .. I don't know ... 1156866819 M * ntrs Ok 1156867043 Q * Greek0 Quit: Lost terminal 1156867270 Q * ||Cobra|| Remote host closed the connection 1156867765 M * matti ntrs: BTW, why you don't want highmem? :) I am curious. 1156867802 M * Bertl matti: mainly because highmem causes non-uniform memory and additional overhead for the highmem swap-in/out 1156867805 M * ntrs As recommended by Bertl. It is supposed to be slow. 1156867815 M * matti Bertl: :-) 1156867850 M * matti Bertl: But, for >= 4GB only HIGHMEM can be used? 1156867869 M * Bertl well, even for >3GB you will not get around it on x86(32bit) 1156867876 M * cehteh for >4GB .. use a 64 bit system 1156867883 M * Bertl fortunately most never cpus are emt64/amd64 1156867892 M * Bertl *newer 1156867899 M * cehteh (actually even for >2GB) 1156867911 M * Bertl nah, 3GB works fine 1156867934 M * matti cehteh: Why? x86 have 4GB address space. 1156867945 M * matti cehteh: And for more memory x86 can use PAE. 1156867962 M * Bertl yes, but with 4GB you have to do special mappings for the kernel 1156867962 M * derjohn matti, PCI bla exclusion 1156867979 M * Bertl derjohn: humbug :) 1156867983 M * matti ? 1156867983 M * derjohn Bertl, cool 1156867994 M * cehteh matti: its no fun performance wise and the address space has already some things mapped in and some ugly programs have silly 2GB limits 1156868000 M * Bertl thing is, you divide the 4GB between userspace and kernel 1156868001 M * derjohn Bertl, how much must be remapped? 1156868013 M * derjohn (if any ...i mean over 3 GB ) 1156868014 M * cehteh sure 3GB and 4GB works .. but it is not really fun in all cases 1156868041 M * cehteh if you have the need to big address spaces then 64 bit is really the way to go 1156868051 M * Bertl 4GB does need an 'overlay' like mapping where kernel and userspace have both 4GB, but separate mappings 1156868065 M * Bertl this adds significant overhead to the kernel-userspace transitions 1156868068 M * cehteh those are all workarounds 1156868084 M * matti cehteh: Even PAE? 1156868098 M * Bertl the highmem and pa* extensions do similar but only for certain ranges (which get mapped in and out) 1156868110 M * matti cehteh: I know, that this will add not really good 3rd level pagetables, but.. 1156868116 M * cehteh hence i recommend if >2GB then 64 bit .. doesnt mean that it wont work in 32 bit .. 1156868163 J * Greek0 ~greek0@85.255.145.201 1156868179 M * matti I see. 1156868190 M * matti So, the best choice is 64 bit arch. 1156868210 M * Bertl derjohn: pci is normally I/O space, but if it maps memory, like vga cards, that's around 128MB or so 1156868214 M * id23 i second this 1156868220 M * id23 hi Bertl :) 1156868224 M * matti Hi id23 :) 1156868227 M * Bertl welcome id23! 1156868229 Q * Greek0 1156868244 M * cehteh (besides if u have a use case where u need big address spaces like databases or such, then the app might actually scale well if u add some more GB's above 4GB in 64 bit) 1156868268 J * Greek0 ~greek0@85.255.145.201 1156868283 M * id23 i just ordered to more Opteron boxes - did you know there will be no more socket 940 - the new one (which should support quad-core) is socket F aka socket 1207 1156868286 M * Bertl wb Greek0! testing your client? 1156868289 M * id23 to=two 1156868315 M * matti id23: :) 1156868320 M * Greek0 Bertl: "The dungeon collapses. You are dead." 1156868325 M * matti Ehehehe. 1156868342 M * matti Bertl: Quick! Use teleportation scroll. 1156868342 M * matti ;] 1156868345 M * derjohn Bertl, some told me (without scientfic precision) that I might not be worth to buy a 4th GB if you already got 3 GB, as it may become slow or even unusable with x86 [S.M.: if you ever read this in the chatlog, shame on you ;)]. 1156868356 M * id23 i love RPGs ;) 1156868357 M * matti Oh, dead... damn. 1156868358 M * Greek0 screen decided that it ran long enough without crash :-/ 1156868379 M * matti derjohn: LOL 1156868379 M * matti ;] 1156868404 M * Bertl derjohn: it's not worth to buy the 4th GB if you already have 3GB with x86-32 :) 1156868432 M * matti Bertl, derjohn, cehteh: Basically, HIGHMEM is the worst-possible choice for lot of memory. The semi-good will be vmsplit, and the best 64 bit arch? 1156868447 M * matti :) 1156868470 M * matti BTW, so. Why people buying Xeons with 16 GB of memory? 1156868473 M * matti ;] 1156868475 M * derjohn id23, socket 940 dies? well 939 died, now socket 1207 ?? OMG, next with next generation opterons you can play "schiffe versenken" on the pingrid 1156868523 M * FaUl derjohn: socket-f is only required for multi-socket-boards, singel-socket boards (for opteron 1xx) will use AM2 instead 1156868529 M * derjohn Bertl, sure , we did not talk abotu x86_64. So S.M. was right? [Please forgive that I did not trust you :)] 1156868603 M * derjohn FaUl, AM2 is only for single cpu, but multi core in favor os 939. socket F was new to me. I always have to take care to buy spare parte in time bevore they get unavailable. 1156868610 M * id23 i spoke to tyan and AMD today about the comming quad cores - as our customer wanrts to upgrade to quad-core next year - but he wants a server today - quad cores will not be supported by socket 940 - so i pray that the socket 1207 mobo will get an bios upgrade as soon the quad-cores are out 1156868611 M * matti derjohn: Who is S.M.? 1156868620 Q * cdrx Ping timeout: 480 seconds 1156868650 M * derjohn id23, we just bought 3* Gt20 (cool machine ,except the cooling block) 1156868667 M * id23 no manufactorer has a "working sample" of quad cores today - so tyan cannot test 1156868676 M * id23 we used to buy the gt24 barebones 1156868684 M * id23 there are now two of them 1156868687 M * derjohn matti, no one from here. a collegue of mine. maybe he'll find this somewhen with google ;) 1156868692 M * id23 one with socket 940 and one with socket F 1156868703 M * matti derjohn: Oh, hehe :) 1156868737 M * id23 i love tyan - the have a *real working* serial console build into the bios 1156868746 M * derjohn well, the old "transport gt" had (have) worse supported sata chipsets. 1156868760 M * derjohn id23, yes, tyan is fine! 1156868765 M * id23 thats why we spend 300 extra bucks on 3ware raid 1156868769 M * id23 per box 1156868789 M * derjohn id23, the new barebone for dual opteron is about 700+ VAT. 1156868810 M * id23 i think about 750+VAT 1156868811 M * derjohn id23, 3ware became the greatest bullshit within the last 24 month. 1156868822 M * id23 because amcc bought them 1156868825 M * id23 maybe 1156868835 M * derjohn id23, switch to acreca if you can live without mainline support, or LSI logic. 1156868835 M * id23 i had to fight hard to get working 64 bit drivers 1156868855 M * id23 ok - i will try them 1156868860 M * derjohn id23, yup. besides that the userspace tools are binary only AFAIR. 1156868860 M * id23 thanx 4 the tip 1156868878 M * derjohn id23, they have 16 port SATA RAID controllers ;) 1156868880 M * id23 they are - i only use 3ware tw_cli 1156868887 M * id23 to get the status of the raid array 1156868892 M * id23 via script 1156868895 M * derjohn the tipp is from are_ , he uses them for a longer time. 1156868905 M * id23 k 1156868934 M * derjohn id23, we gloablly decided to go for softraid/md. save a lot of spare-part-foo and installation issues. 1156869155 M * cehteh .. with partitionable md-devices you can even forget the lvm crap :P 1156869169 M * id23 partitionable md-devices ? 1156869192 M * id23 i have 2 boxes running softraid - no problems so far but RAID1 should make no problems 1156869198 M * cehteh was and old patch and i in the mainline kernel since some time 1156869198 M * id23 anyway 1156869217 M * cehteh md raid5 is just quite slow naturally 1156869222 M * derjohn id23, as long as Debian Sid don't )(/&/()&$%§$ around with the mds after dist-upgrade ;) 1156869239 Q * s0undt3ch Server closed connection 1156869240 J * s0undt3ch ~s0undt3ch@bl7-240-60.dsl.telepac.pt 1156869257 M * cehteh but md raid1 often outperforms cheap raid adapters 1156869393 M * id23 cya tomorrow 1156869408 M * derjohn cehteh, yup, it certainly does. 1156869432 M * derjohn cehteh, an exception maybe a very loaded machine, as the cpu had to build the raid 1156869441 M * derjohn _may_ 1156869462 J * Viper0482 ~Viper0482@p549771E6.dip.t-dialin.net 1156869620 M * cehteh sure .. but which machine is cpu-loaded nowadays ;) 1156869638 M * cehteh (at least when io because of the software raid lags behind) 1156869701 Q * cehteh Server closed connection 1156869727 J * cehteh ~ct@cehteh.homeunix.org 1156870215 J * ybanafa ~admin@198.69.12.8 1156870278 M * ybanafa hi all .. what is the best way to get cd-image for my host-server sarge (to DVD) ? 1156870303 M * daniel_hozac what? 1156870309 M * matti :) 1156870310 M * daniel_hozac you mean an installable media to create hosts? 1156870328 M * Bertl ybanafa: cdrecord? 1156870424 M * cehteh growisofs for DVD 1156870445 M * ybanafa I installed clean sarge-server+vserver b4 i do any thing I want get cd-image-iso-bootable-dvd for guick restore 1156870457 M * daniel_hozac if you need a DVD for a host, you're doing something wrong :) 1156870479 M * Bertl ah, now I understand ... wasn't there some rescue tool to do that ... 1156870484 M * cehteh there are tools which create bootable DVD's automatically 1156870501 M * cehteh yes .. just cant remember the name 1156870504 M * daniel_hozac m something? 1156870551 M * Bertl http://freshmeat.net/projects/g4l/ 1156870571 M * Bertl but I remember something else too, like part* 1156870572 M * derjohn well, why not using systemimager or simply dd ? 1156870614 M * daniel_hozac http://www.mondorescue.org/ 1156870851 M * daniel_hozac Bertl: have you added people.linux-vserver.org to DNS yet? 1156870870 M * Bertl nah, I forgot ... will do now 1156870901 M * Bertl http://vserver.13thfloor.at/Experimental/delta-init-fix01.diff fixes http://paste.linux-vserver.org/307 1156870925 M * Bertl http://vserver.13thfloor.at/Experimental/delta-remount-fix01.diff (untested, but supposed to fix the debian issue) 1156870945 M * daniel_hozac ok, i missed the legacy case. 1156870971 M * daniel_hozac question is, do we want to readd STATE_INIT if init disappears? 1156871010 M * Bertl not for now .. we can do similar tricks later, but we are missing proper triggers for the generic case 1156871015 J * bonbons ~bonbons@83.222.36.236 1156871022 M * Bertl welcome bonbons! 1156871031 M * bonbons Hey Bertl! 1156871079 M * ybanafa All : thanks++; systeimager good vserver (look like from my reading) 1156871149 P * ybanafa Leaving 1156871154 J * cdrx ~legoater@cap31-3-82-227-199-249.fbx.proxad.net 1156871172 M * Bertl wb cdrx! 1156871232 M * daniel_hozac Bertl: shouldn't the remount check for CAP_SYS_MOUNT instead? 1156871259 J * nayco ~nayco@lns-bzn-58-82-251-208-184.adsl.proxad.net 1156871320 M * nayco 'loo again ! 1156871450 M * nayco Bertl: Yes, it was was at least 17h00 in the morning when you turned back online ;-). 1156871590 M * Bertl daniel_hozac: do we have one now? 1156871654 M * daniel_hozac hmm, sorry, i thought there was one. 1156871670 M * Bertl funny thing is, me too when I changed it first :) 1156871744 M * Bertl okay, people was added, leaving now ... have to run! 1156871747 M * nayco Bertl: I'd like to talk with you about the aricle I'm preparing for 2.0.2 release... Have you got a couple of minutes to spend, please ;-) ? 1156871754 N * Bertl Bertl_oO 1156871762 M * Bertl_oO nayco: later in the evening! 1156871767 M * daniel_hozac hehe. 1156871773 M * nayco Ok, thanks. 1156871798 M * daniel_hozac nayco: anything i can do? 1156871924 M * nayco Yeah, maybe ;-) 1156871937 M * nayco ...second... 1156871938 M * derjohn daniel_hozac, nayco well, we put it on our comanpy's news site, at least a PR5 page :) 1156871955 M * derjohn (even if we dont really use stable) 1156871970 M * derjohn maybe other should do it, too? 1156872004 M * nayco ...checking somethig. 1156872010 M * daniel_hozac huh? 1156872042 M * nayco Ok. daniel_hozac : Yes, there is a couple of thing missing for my article : 1156872082 A * nayco wonders what are the "kernel helper enhancements", and what is the kernel helper 1156872109 M * daniel_hozac hmm, i guess that could mean a few things... 1156872134 A * nayco would like to know when the 2.0.2 will be released, I do not want the news to be published to early : This site has huge french audience 1156872178 M * daniel_hozac sometime after Hollow gets back from Berlin, which is saturday (right?). 1156872213 M * daniel_hozac nayco: do you have some context? 1156872241 M * nayco "http://home.xnull.de/work/vserver/press/20060828-announce-202-wiki.pdf" 1156872252 M * nayco http://home.xnull.de/work/vserver/press/20060828-announce-202-wiki.pdf <= sorry 1156872277 M * daniel_hozac right. 1156872288 M * daniel_hozac do you mean "Improved helper state/return code handling"? 1156872307 M * nayco daniel_hozac: Ok, I'll delay the article publication accordingly, so ;-) 1156872385 M * daniel_hozac nayco: i believe the plan is to release no later than monday, but we've been saying similar things for weeks :) 1156872395 M * nayco :D 1156872411 M * nayco daniel_hozac: And for the new website, any hint ? 1156872436 M * daniel_hozac they are all going to be released at once, i think. 1156872449 M * nayco great 1156872547 M * nayco daniel_hozac: 'do you mean "Improved helper state/return code handling"?' => Yes, But more precisely, as I summed up the intro, the sentence is : 1156872556 M * nayco " Today, the Linux-VServer project released a new stable version of its ker- 1156872556 M * nayco nel patch featuring support for Bind Mount Extensions, improved filesystem 1156872556 M * nayco support for JFS, kernel helper improvements and other minor enhancements. 1156872556 M * nayco " 1156872584 M * daniel_hozac ah. 1156872615 M * nayco So, I nearly kept this sentence as-is, but added 2 footnotes for " Bind Mount Extensions" and " kernel helper improvements" 1156872620 M * daniel_hozac i don't know exactly what that is referring to, probably the vshelper improvements. 1156872624 M * nayco To explain 1156872668 M * daniel_hozac which let vshelper dictate whether or not vc_ctx_create should fail. 1156872674 M * nayco Yes, I think so : But as I need a (very) short explaination of what is the "kernel helper"... And as I nearly do not know what it is ;-)... 1156872707 M * daniel_hozac i believe Hollow uses this to create contexts from vshelper, i.e. without a vserver script, and the daemon just calls vc_ctx_create. 1156872744 M * daniel_hozac (which spawns vshelper, which starts the guest, which can (with the fixes) tell the caller how things turned out) 1156872748 M * nayco So, the kernel helper is a daemon that runs on the host ? 1156872763 M * daniel_hozac no, the kernel helper is a program invoked by certain operations. 1156872786 M * nayco yes, I remember that "daemon" is the new tools... 1156872789 M * daniel_hozac i.e. when you start a guest, or a guest dies, or when you use reboot/halt -f inside the guest to shut it down from the outside. 1156872824 M * daniel_hozac basically, it lets userspace care about how to deal with guests, rather than the kernel. 1156873004 M * nayco Ok, that's what I had heard months ago... I knew that it was involved in halt/reboot from inside the guest. So, to sum up in simple words, what would you say to readers ;-) ? 1156873045 Q * coocoon Ping timeout: 480 seconds 1156873110 M * daniel_hozac i don't know... a program that helps the guest perform certain privileged actions? 1156873388 M * nayco In fact, it's a kind of proxy that permits the guest to send privileged commands to the host ? 1156873496 M * daniel_hozac i guess... privileged may be the wrong word though, as it's possible to do a restart/halt without using it. 1156873727 Q * FireEgl Ping timeout: 480 seconds 1156873779 M * nayco Well, in fact, i'm not sure to understand the purpose of it, so, because... Oh, ok, I think I see clearer... reboot/halting is possible within a vserver, because all one has to do is to run the good init scripts, but when done, the context is still alive... And I guess that a vserver cannot destroy its own context ;-). So, is vshelper there to accomplish this : Create and destroy contexts when a guest request it (for the guest's c 1156873779 M * nayco ontext, I mean) ? 1156873883 J * coocoon ~coocoon@p54A0624D.dip.t-dialin.net 1156873928 J * FireEgl FireEgl@Sebastian.Atlantica.US 1156874149 M * cdrx hi bertl ! 1156874177 M * daniel_hozac no, but vshelper will remove IP addresses and such. 1156874196 M * daniel_hozac unless the context has the persistent flag set, it will vanish as soon as the last process exits. 1156874231 M * nayco Init can stop itself ? 1156874330 M * daniel_hozac yes, any process can stop itself. 1156874333 M * daniel_hozac exit(0) :) 1156874362 M * daniel_hozac (and a lot of guests don't have an init) 1156874582 M * nayco Oh, yes ;-) 1156874805 M * nayco So, vshelper is triggered by the stopping guest, to perform all the stopping/restarting operations the guest cannot do itself, right ? 1156875336 M * daniel_hozac and for some other events. 1156875731 M * nayco :D...Is there an up to date doc somewhere, from which I can get matter do make a small summary sentence ? 1156875834 M * daniel_hozac of vshelper? probably not... 1156876009 Q * Viper0482 Remote host closed the connection 1156876617 Q * shedi Quit: Leaving 1156876909 Q * Johnnie Remote host closed the connection 1156877598 Q * lilalinux__ Remote host closed the connection 1156877879 Q * click Server closed connection 1156877885 J * click click@ti511110a080-2980.bb.online.no 1156878971 J * shedi ~siggi@inferno.lhi.is 1156879282 Q * Vudumen Ping timeout: 480 seconds 1156879507 J * Vudumen ~vudumen@perverz.hu 1156880246 J * kaner_ kaner@strace.org 1156880291 Q * glut Server closed connection 1156880314 J * glut glut@no.suid.pl 1156880348 Q * kaner Ping timeout: 480 seconds 1156881248 J * matti_ matti@linux.gentoo.pl 1156881410 Q * matti Ping timeout: 480 seconds 1156881410 N * matti_ matti 1156881625 Q * Nam Server closed connection 1156881639 J * Nam ~nam@S0106001195551ff0.va.shawcable.net 1156882545 Q * cdrx Ping timeout: 480 seconds 1156883216 J * yarihm ~yarihm@84-75-121-93.dclient.hispeed.ch 1156884059 J * shedii ~siggi@inferno.lhi.is 1156884096 P * stefani I'm Parting (the water) 1156884119 J * DavidS ~david@chello062178045213.16.11.tuwien.teleweb.at 1156884337 J * s4edi ~siggi@inferno.lhi.is 1156884397 Q * shedi Ping timeout: 480 seconds 1156884565 J * shedi ~siggi@inferno.lhi.is 1156884754 Q * shedii Ping timeout: 480 seconds 1156884755 J * shedii ~siggi@inferno.lhi.is 1156884875 M * DavidS hoi, is there a way to take the NET_BIND_SERVICE _away_ from a vserver? 1156884888 Q * s4edi Ping timeout: 480 seconds 1156884894 M * DavidS (capability that is) 1156884909 M * daniel_hozac echo "~CAP_NET_BIND_SERVICE" >> bcapabilities? 1156884960 M * mnemoc what the ~ means there? 1156885006 M * daniel_hozac disable :) 1156885010 M * daniel_hozac ! should also work. 1156885095 M * DavidS daniel_hozac: if you ever yome to vienna, I owe you one :) 1156885099 Q * shedi Ping timeout: 480 seconds 1156885100 M * DavidS come even .. 1156885119 M * daniel_hozac hehe. 1156885121 A * DavidS definitely has to finish this job and go to bed 1156885136 M * DavidS daniel_hozac: bertl knows where to find me ;) 1156885151 Q * bonbons Quit: Leaving 1156885726 M * derjohn2 daniel_hozac, : next Usermeeting in Vienna, DavidS pays ;) ! 1156885752 N * Bertl_oO Bertl 1156885761 M * Bertl evening! sounds good :) 1156885788 M * mnemoc how many can stay at Bertl's ? :) 1156885905 M * Bertl at least room for 20+ I guess :) 1156885991 M * mnemoc :D 1156886119 Q * DavidS Ping timeout: 480 seconds 1156886195 M * Bertl ntrs is continuing to report strange behaviour, mainly mysql having problems and dropping socket connections to the rest of the apps, anybody experienced that with recent devel kernels too? 1156886862 Q * s0undt3ch Ping timeout: 480 seconds 1156887166 J * s4edi ~siggi@inferno.lhi.is 1156887543 Q * shedii Ping timeout: 480 seconds 1156887781 J * shedii ~siggi@inferno.lhi.is 1156887868 Q * dna Quit: Verlassend 1156887988 Q * meandtheshell Quit: bye bye ... 1156888158 Q * s4edi Ping timeout: 480 seconds 1156888246 J * s0undt3ch ~s0undt3ch@bl7-240-60.dsl.telepac.pt 1156888278 M * derjohn2 Bertl, recent? I dunno how much connections is "much" 1156888715 M * Bertl neither do I :) 1156888820 M * harry Bertl: any ideas on release of 2.0.2 ? 1156888859 A * harry looking forward to new tools + release of 2.0.2 :) 1156888889 J * s4edi ~siggi@inferno.lhi.is 1156889113 M * Bertl harry: when hollow will get back 1156889155 J * shedi ~siggi@inferno.lhi.is 1156889283 Q * shedii Ping timeout: 480 seconds 1156889433 Q * s4edi Ping timeout: 480 seconds 1156889643 Q * shedi Ping timeout: 480 seconds 1156889656 Q * id23 Ping timeout: 480 seconds 1156889761 Q * matti Read error: Operation timed out 1156890124 J * matti matti@linux.gentoo.pl 1156890701 M * nayco Bertl: Are you available ;-) ? 1156890746 M * nayco I mean, for a couple of questions ? 1156890767 M * Bertl almost, should be fine in 20min or so, have to get seomething to eat first 1156890777 M * nayco :D ! 1156890782 M * nayco K 1156890807 M * nayco Wich hour is it, in vienna ? 1156890819 M * Bertl half past midnight 1156891022 M * nayco ??? Like in France, I wouldn't have thought... Anyway, eating at 00h30 is geeky :P ! 1156891274 J * matti_ matti@linux.gentoo.pl 1156891451 Q * sladen Ping timeout: 480 seconds 1156891729 Q * matti Ping timeout: 480 seconds 1156891729 N * matti_ matti 1156891793 M * Bertl nayco: okay, shoot! 1156891801 M * nayco K 1156891831 M * nayco I start all over, because this is all over the logs, so... 1156891851 M * Bertl ah, good :) 1156891986 M * nayco As I saw that Hollow written an annoucement for 2.0.2 release. I thougth it a good idea to write myself a french article to publish in http://linuxfr.org , which is a high audience linux french site. The french speaking (canada/belgium/swiss...) slashdot. 1156891989 M * nayco Ok, so, 1156892033 M * nayco I translated and summed up the official announcement, and it is ready to publish. 1156892047 M * Bertl excellent idea! 1156892117 M * nayco There is only one of my questions left (others here answer already), but I will tells you a little about the other after. So : 1156892151 M * Bertl okay ... 1156892155 M * nayco Which short sentence can best describe the "kernel helper" ? What is its purpose ? 1156892260 M * nayco [just for fun : http://linuxfr.org/ runs in a vserver ;-) I gonna ask the admins to add a small note on this - patch version, reliability...] 1156892314 M * nayco Back to the subject : I need a small description of vshelper because the annoucement talks about its improvements in the summary/intro 1156892318 M * Bertl the helper is there to 'relay' or 'delegate' certain events to userspace 1156892332 M * Bertl best example for that is the sys_reboot() 1156892338 M * nayco yes... 1156892363 M * Bertl when userspace calls into sys_reboot() normally the system is rebooted or halted 1156892385 M * Bertl we do not really want that in a guest, so we need to intercept that, and handle it differently 1156892424 M * Bertl instead of rebooting, the kernel helper is called, and informed that guest soandso has called sys_reboot() 1156892461 M * Bertl (very much like the hotplug helper) 1156892559 M * Bertl so, in one sentence: 1156892570 M * nayco Ok, so, vshelper, on the hosts, listens for "commands" sent by the kernel, to 1156892595 M * nayco do on the host what cannot be done inside a guest. 1156892615 M * nayco If I understant it sell ? 1156892616 M * Bertl the kernel helper is a mechanism to handle context specific events in userspace, without the need of putting policy into the kernel. 1156892633 M * nayco whaooo 1156892648 M * nayco er... Let me translate into french.... ;-) 1156892654 M * Bertl and you can add (similar to the hotplug helper) 1156892664 M * nayco good idea ;-) 1156892700 M * nayco "in userspace", you mean "on the host" ? 1156892730 M * Bertl yes, but most importantly, outside the kernel :) 1156892732 M * FaUl mh, anyone here who is a clue where the kernel-helper hides in debian and how to start it? 1156892737 M * FaUl s/is/has 1156892761 M * Bertl /proc/sys/kernel/vshelper knows :) 1156892867 M * FaUl ah 1156892943 M * nayco Bertl: Ok, thx, I think that'll be sufficient for a small article. Anyway, I want to know for myself : " but most importantly, outside the kernel" => you mean that for example, when destroying the context of a stopping vserver, the kernel itself cannot remove IP adresses from its tables (I heard that is one of vshelper's purposes) ? It needs to ask vshelper, wich in turn call some shell command to do it ? 1156893039 M * Bertl of course the kernel could do that, but it adds policy to the kernel, i.e. hard coded decisions, what to do when, etc 1156893063 M * Bertl IMHO those (usually very mutable) decisions belong into userspace 1156893185 M * Bertl btw, here is a small overview I did (mainly for myself) what changed between version 2.0.1 and the upcoming 2.0.2 1156893193 M * Bertl http://paste.linux-vserver.org/283 1156893206 M * Bertl not all of it is really interesting for a broad audience 1156893336 M * nayco I'll read this thanks. And I hope the IRC logs don't break until tommorow :D 1156893348 M * nayco Well, for the other items : 1156893369 M * nayco I made a small sentenc to explain BME 1156893379 M * Bertl FaUl: and it is invoked by the kernel, kind of automatically 1156893422 M * Bertl nayco: let's hear! 1156893426 M * nayco something like "BME allows for --bind mounts with different options than the original mountpoint" 1156893449 M * nayco Someone here explained me 1156893450 M * Bertl hmm, yeah, that's pretty much it 1156893457 M * nayco Ok... next... 1156893463 M * Bertl BME stands for Bind Mount Extensions 1156893464 M * FaUl Bertl: looks like hotplug-interface, doesn't it? 1156893498 M * Bertl nayco: and the 'extension' is that you can have 'normal' mount flags, like read-only or noexec on such bind mounts 1156893507 M * Bertl FaUl: precisely! 1156893534 M * FaUl Bertl: i think this isn't the badest idea 1156893681 M * nayco Bertl: ok, this is more clear... Until this morning, I did not even know that one couldn't have options on a --bind :D !!! 1156893694 M * nayco I never use it... 1156893697 M * nayco Ok. 1156893769 M * nayco Ok, last thing, but I already got the answer, now : This announcment is to be made after Hollow's return from Berlin, at the beginning of next week, right ? 1156893805 M * Bertl yes, we are very likely to release 2.02 then 1156893908 M * FaUl Bertl: what about 2.2, anyway? 1156893911 M * nayco Ok, I will wait until then. After that, the article may take a couple of days to be moderated... 1156893922 M * FaUl 2.2 will be the next stable-series, nor? 1156893982 M * nayco Bertl: I think that's all, I will finish the article tomorrow. Oh, and it will include your official announcement ;-) 1156893985 M * nayco Thanks. 1156893999 M * Bertl nayco: thank you! 1156894019 M * Bertl Fuayes, the next stable branch, basically the distillation of 2.1.x 1156894040 M * Bertl *hmm* that was meant to be FaUl! 1156894057 M * FaUl Bertl: yes, - is there an eta? 1156894075 M * Bertl I think it will at least take a month 1156894077 A * FaUl personally uses 2.1 since rc14 or something and never had problems with it ;- 1156894081 M * FaUl what are open issues? 1156894105 M * Bertl mainly cleanups (required for the 2.1.1 release anyway and testing ... 1156895145 M * nayco Bertl: Do you know when your last XFS quota enhancements could make it to the devel patch ? 1156895208 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1156895224 M * nayco I must say I've been only using devel versions in production or nearly 2 years, and I never had a problem :O ! Good job. 1156895236 M * nayco since 1.9.x... 1156895347 M * Bertl I have to rework that hack patch first, but it will get into devel soon, mazbe before 2.1.1 even 1156895487 M * nayco Great ! 1156895594 M * nayco So, I have server to put in prod in a couple of days with quotas, I better use 2.0.2(-rc?), no ? IIRC, quotas work in this release, although one needs to use /dev/hdvx instead of /mountpoint, right ? 1156895654 M * Bertl yep, seems so 1156895670 Q * yarihm Quit: Leaving 1156895856 M * nayco Ok, i'll do this. Oh, I forgot : I've written a french article on XFS quotas with vserver (added a link on the wiki), but it's not useful for now, not before your hack is incorporated :rofl: ! When validated, I will try to translate it in english and put it on the wiki. 1156895897 M * Bertl okay, good work! 1156895920 M * nayco Bertl: Oh, and do quotas work only for XFS, or do your last patch applies to etx2-3, or do ext2-3 were already fully working ?