1182211322 Q * [BiG^BrotheR] Ping timeout: 480 seconds
1182211681 Q * coderanger Ping timeout: 480 seconds
1182212011 P * stefani I'm Parting (the water)
1182213209 M * Solaris so can i use 64bit guests on a 32bit host?
1182213286 M * Bertl nope, not without a system/userspace emulator
1182213364 Q * nebuchadnezzar Read error: Connection reset by peer
1182213367 J * nebuchadnezzar ~nebu@zion.asgardr.info
1182213391 Q * ensc Remote host closed the connection
1182213466 M * Solaris ok thanks..
1182213763 J * DoberMann_ ~james@AToulouse-156-1-15-162.w86-196.abo.wanadoo.fr
1182213873 Q * DoberMann[ZZZzzz] Ping timeout: 480 seconds
1182214078 M * Bertl Solaris: but in theory, you could launch a guest via qemu (userspace emulation) and thus run 64bit guests on 32bit hardware ...
1182216428 Q * chand Quit: chand
1182217101 J * coderanger ~coderange@c-65-96-210-168.hsd1.ma.comcast.net
1182217267 J * fullsick ~bob@203.176.96.250
1182218601 Q * fullsick 
1182218821 M * Bertl okay, off to bed now ... have a good one everyone!
1182218827 N * Bertl Bertl_zZ
1182218827 J * arachnist arachnist@088156185052.who.vectranet.pl
1182218872 J * Greek0 ~greek0@85.255.145.201
1182218877 Q * samueltc Ping timeout: 480 seconds
1182220085 J * Greek0_ ~greek0@85.255.145.201
1182220135 Q * Greek0 Read error: No route to host
1182221776 J * gerrit ~gerrit@c-67-160-146-170.hsd1.or.comcast.net
1182226249 P * cjb ERC Version 5.2 stable pre-release (IRC client for Emacs)
1182226262 J * _jthm_ ~infowolfe@c-24-10-147-179.hsd1.ut.comcast.net
1182226644 Q * infowolfe_ Ping timeout: 480 seconds
1182229257 Q * meandtheshel1 Quit: Leaving.
1182229601 M * mugwump hmmm, etch system (debian 2.6.18-4-vserver-amd64 kernel, 0.30.212) can't use ping in guest...
1182229660 A * mugwump slaps self
1182229663 M * mugwump iptables!
1182229678 M * mugwump that "operation not permitted" error threw me
1182230657 N * DoberMann_ DoberMann
1182232565 J * sharkjaw ~gab@158.36.45.236
1182233167 J * duckx ~Duck@tox.dyndns.org
1182233548 Q * sladen Ping timeout: 480 seconds
1182234146 J * HeinMueck ~Miranda@dslb-088-064-024-002.pools.arcor-ip.net
1182235258 Q * HeinMueck Quit: Aah!
1182235283 J * HeinMueck ~Miranda@dslb-088-064-024-002.pools.arcor-ip.net
1182235493 J * marcfi1 ~mef@c-68-39-177-97.hsd1.nj.comcast.net
1182235623 J * chand ~chand@m167.net81-64-156.noos.fr
1182235835 P * marcfi1 
1182236135 J * dlezcano ~dlezcano@AToulouse-252-1-44-145.w82-125.abo.wanadoo.fr
1182236309 M * matti Morning.
1182236453 Q * dlezcano1 Ping timeout: 480 seconds
1182236773 J * dna ~naucki@249-245-dsl.kielnet.net
1182236776 Q * chand Quit: chand
1182237074 J * infowolfe_ ~infowolfe@c-24-10-147-179.hsd1.ut.comcast.net
1182237481 Q * _jthm_ Ping timeout: 480 seconds
1182238323 J * ruskie ruskie@ruskie.user.oftc.net
1182239187 Q * dlezcano Ping timeout: 480 seconds
1182239540 N * Bertl_zZ Bertl
1182239549 M * Bertl morning folks!
1182239713 J * chand ~chand@212.99.51.254
1182240044 J * chand_ ~chand@212.99.51.254
1182240044 Q * chand Read error: Connection reset by peer
1182240433 Q * chand_ Quit: chand_
1182241202 Q * ruskie Read error: Connection reset by peer
1182241382 J * ruskie ruskie@ruskie.user.oftc.net
1182242642 J * ktwilight_ ~ktwilight@170.69-66-87.adsl-dyn.isp.belgacom.be
1182242646 Q * ktwilight Read error: Connection reset by peer
1182242652 J * pmenier ~pmenier@LNeuilly-152-22-72-5.w193-251.abo.wanadoo.fr
1182243254 J * sladen paul@starsky.19inch.net
1182243337 J * DavidS cool@helios.uni-ak.ac.at
1182243723 Q * DavidS Quit: Leaving.
1182243759 J * DavidS cool@helios.uni-ak.ac.at
1182243863 J * dlezcano ~dlezcano@blueice3n1.uk.ibm.com
1182245319 J * jmt ~jmt@port-212-202-23-50.dynamic.qsc.de
1182245345 J * bzed ~bzed@wireless-4-93.internal.dc7.debconf.org
1182245527 Q * bzed Remote host closed the connection
1182245580 M * jmt Is anybody running vserver on a xfs file system with quota support?
1182245589 M * harry not me
1182245703 M * Bertl jmt: quota as in user/group quota or as in 'disk limits'?
1182245734 M * jmt I have the problem, that the quota accounting does not work when the guest is running. tagxid works fine.
1182245747 M * jmt I'm speaking about the disk limits...
1182245797 M * Bertl okay, you mean, the kernel does not update the reservations?
1182245805 M * jmt Yes!
1182245828 M * Bertl okay, what kernel/patch version?
1182245900 M * Bertl ah, just checked, not implemented yet for xfs
1182245903 M * jmt I tested 2.6.20.11-vs2.3.0.12 from the vserver sources and 2.6.20-vs2.2.0 from the gentoo sources
1182245922 M * Bertl i.e. ext2/3/4/jfs and reiserfs has it
1182245950 M * jmt okay, so i'll give them a try. Are there any plans to implement xfs?
1182245974 M * Bertl if there are enough folks interested in it, and willing to test the stuff, I don't see why not
1182246022 M * jmt So there is one new person interested in it and willing to test ;-)
1182246071 M * jmt And thanks for the fast response..!
1182246097 M * Bertl okay, you have a machine where you could build/install recent patches/kernels and test XFS stuff, potentially destroying your filesystem in the process?
1182246146 M * jmt Yes I have! Hoping that it will only destroy the vserver filesystem. Otherwise I need some time for installation...
1182246189 J * bzed ~bzed@wireless-4-202.internal.dc7.debconf.org
1182246204 M * Bertl welcome bzed!
1182246217 M * bzed thanks Bertl  :)
1182246233 M * Bertl jmt: okay, give me a few minutes to dive into the code ... I can then tell you the odds :)
1182246389 M * Bertl jmt: could you do a 'df -i' and 'df' of one of your xfs filesystems and upload that to paste.linux-vserver.org please?
1182246436 M * jmt One moment. I have to reboot first...
1182246906 M * jmt Done
1182246933 M * Bertl tx
1182247087 M * Bertl hmm, looking at the code makes me wonder if xfs shouldn't already implement something which _could_ be used for this kind of accounting ...
1182247127 M * Bertl jmt: are you familiar with xfs (somewhat) or just using it because it is so <insert quality here>?
1182247193 M * jmt Bertl: I'm mostly using it. 
1182247194 Q * harry Quit: leaving
1182247207 J * harry ~harry@d54C2508C.access.telenet.be
1182247249 M * Bertl jmt: okay, it seems like xfs has support for project quotas, and they can be assigned/inherited from directories (building directory trees)
1182247282 M * Bertl jmt: it also seems like those directory trees 'look' like separate filesystems to df and friends
1182247331 J * lilalinux ~plasma@dslb-084-058-207-105.pools.arcor-ip.net
1182248118 M * jmt Bertl: I'm just trying to test it, but i've problems with my gentoo guest. Do you know a place where i can download a test guest?
1182248142 M * Bertl should be trivial to build a debian guest via network
1182248170 M * jmt Bertl: with debootstrap?
1182248218 M * Bertl vserver etch01 build -m debootstrap --context 10104 --hostname etch01.debian.org --interface eth1:10.1.4.1/24 -- -d etch -m http://ftp.debian.org/debian -- --arch i386
1182248241 M * Bertl adjust the arguments to your liking ...
1182248436 M * jmt Bertl: I get: E: No such script: /usr/lib/debootstrap/scripts/etch
1182248575 M * jmt Bertl: okay, sarge is working...
1182248614 M * Bertl you seem to have older tools then, might want to consider upgrading
1182249011 Q * bzed Remote host closed the connection
1182249043 M * jmt Bertl: okay, the debian guest is running. But the xfs progs does not support project quotas...
1182249152 M * Bertl you have to set that on the host ...
1182249185 M * Bertl and you need to mount the xfs with the prjquota option (or whatever it is called)
1182249276 M * jmt Bertl: I know. I just follow the example in the manpage from xfs_quota. Quota support for xfs is also in the kernel.
1182249310 M * Bertl maybe you need more recent xfs quota tools then?
1182249444 M * jmt I'll try it.
1182250242 J * bzed ~bzed@wireless-5-22.internal.dc7.debconf.org
1182251643 Q * Aiken Quit: Leaving
1182252082 J * [BiG^BrotheR] ~Dr-DreaM@84.23.96.253
1182252220 M * Hollow argh. this clock drives me nuts...
1182252285 M * Bertl need some rest ... will be back later ...
1182252291 N * Bertl Bertl_zZ
1182252577 J * DreaM[BrB] ~Dr-DreaM@84.23.96.253
1182252597 M * awk *sobs* :(
1182252613 M * awk i used up all my b/w and i didn't even know it!
1182252624 M * awk i blame it on all my vservers i built :(
1182252636 M * awk should have setup a local repo and pulled it from there, instead of updating each box! :D
1182252746 Q * pmenier Read error: Connection reset by peer
1182252749 M * Guy- awk: apt-proxy does just that :)
1182252752 M * DavidS awk: take a look at one of the apt-proxy implementations ... even squid with appropriate config helps
1182252759 M * DavidS Guy-: :)
1182252765 M * DavidS c ya all! 
1182252767 Q * DavidS Quit: Leaving.
1182252976 Q * [BiG^BrotheR] Ping timeout: 480 seconds
1182253010 Q * DreaM[BrB] Quit: ][DreaM-ScripT][
1182253145 M * trippeh_ awk: There's still capped connections out there? :-o
1182253175 M * trippeh_ The market here rejected it early, years ago :)
1182253208 M * trippeh_ The few attempts led to bad press and mad customers
1182253283 Q * tanjix 
1182254299 J * rob-84x^ rob@submarine.ath.cx
1182254352 J * pmenier ~pmenier@LNeuilly-152-22-72-5.w193-251.abo.wanadoo.fr
1182256605 J * cruser ~chatzilla@72.242.194.162
1182257000 J * ema ~ema@rtfm.galliera.it
1182257146 Q * bzed Ping timeout: 480 seconds
1182257240 J * bzed ~bzed@wired-241-248.internal.dc7.debconf.org
1182257482 Q * bzed 
1182257487 J * [BiG^BrotheR] ~Dr-DreaM@84.23.96.253
1182257550 J * _jthm_ ~infowolfe@c-24-10-147-179.hsd1.ut.comcast.net
1182257583 J * fatgoose ~samuel@76-10-154-117.dsl.teksavvy.com
1182257875 M * cruser Hi I mounted a samba share /vserver/vguest/foo in a vhost but the vguest is unable to see it. Which is what I want. Is this the right way?
1182257954 Q * infowolfe_ Ping timeout: 480 seconds
1182258459 M * Loki|muh it is only visible if you mounted it before the start of the guest
1182258478 M * Loki|muh because mounts are only visible in a specific namespace
1182258665 J * bzed ~bzed@wired-251-239.internal.dc7.debconf.org
1182259100 M * fatgoose hi
1182259121 M * cruser Loki|muh: Thanks, I will have another look.
1182259249 M * cruser Yes. That did it.
1182260295 Q * bzed Quit: Leaving
1182260347 P * cruser 
1182261078 J * bzed ~bzed@wired-251-239.internal.dc7.debconf.org
1182261532 Q * ema Remote host closed the connection
1182261811 J * ema ~ema@rtfm.galliera.it
1182262004 Q * sharkjaw Ping timeout: 480 seconds
1182263712 J * DreaM[BrB] ~Dr-DreaM@84.23.96.253
1182264006 Q * [BiG^BrotheR] Ping timeout: 480 seconds
1182264306 J * DavidS cool@helios.uni-ak.ac.at
1182264708 M * daniel_hozac coderanger: why would you need a 0.30.214?
1182265550 Q * DavidS Quit: Leaving.
1182265588 J * [BiG^BrotheR] ~Dr-DreaM@84.23.96.253
1182265856 J * sharkjaw ~gab@158.36.45.236
1182265971 Q * DreaM[BrB] Ping timeout: 480 seconds
1182267219 J * DreaM[BrB] ~Dr-DreaM@84.23.96.253
1182267456 Q * [BiG^BrotheR] Ping timeout: 480 seconds
1182268060 J * mountie ~mountie@CPE000f66950c89-CM000a739acaa4.cpe.net.cable.rogers.com
1182268093 J * stefani ~stefani@tsipoor.banerian.org
1182268530 M * stefani I see on wiki a 32-bit fedora and ubuntu on 64-bit debian.   can anyone point me to a howto for 32-bit debian on 64 bit host?    do i only need the --arch=i386 in vserver build ?
1182268559 M * daniel_hozac yep.
1182268701 Q * bzed Ping timeout: 480 seconds
1182268706 M * stefani ok.
1182268710 N * Bertl_zZ Bertl
1182268713 M * Bertl back now ...
1182268754 M * Bertl stefani: as last option section, so something like: -- --arch=i386
1182268829 M * Bertl here is what I tested last time, before I discovered that rpm based installs are broken ...
1182268833 M * Bertl http://paste.linux-vserver.org/2650
1182268865 M * Bertl daniel_hozac: btw, what about rpm based guests with statically linked rpm tool, as in madriva?
1182269064 M * daniel_hozac i'm almost positive that it should work fine. i just haven't had time to test it yet...
1182269088 M * Bertl ah? so should be in TRUNK or what?
1182269111 M * Bertl (means: can I test it for you?)
1182269152 M * daniel_hozac well, no. i'll try to commit it later today.
1182269169 M * Bertl okay, how is princeton?
1182269183 M * daniel_hozac quite a bit warmer than home ;)
1182269186 M * daniel_hozac or at least more humid.
1182269195 M * daniel_hozac but other than that, i like it.
1182269209 M * Bertl hehe, kinda old english in some places ...
1182269261 M * Bertl make sure to get a campus tour ...
1182269283 M * daniel_hozac yeah, i'm planning to.
1182269340 M * Bertl ah, and you probably want to get a version of the latest K&R C book (or similar) for signing :)
1182269409 M * daniel_hozac hehe, yeah.
1182269568 M * Bertl okay, I figured that the pid=1 virtualization is unfixable without changing the way how proc now works
1182269580 M * daniel_hozac really?
1182269587 M * daniel_hozac can't we just skip it like in the past?
1182269612 M * Bertl yes, we can either get wrong readdir entries, or the disappearing pid=1 issue
1182269632 J * bonbons ~bonbons@2001:5c0:85e2:0:20b:5dff:fec7:6b33
1182269643 M * Bertl the reason for that is that proc has no state and assumes that always inode == task
1182269665 J * cedric ~cedric@rny93-2-82-66-66-30.fbx.proxad.net
1182269666 M * Bertl what we need to do to solve that is the following:
1182269689 M * Bertl we can either make pid=1 a 'special' proc entry, with additional logic/functions
1182269720 M * Bertl or alternatively, add state information to the proc dentries about the context (espeically for pid=1)
1182269754 M * Bertl I'm currently investigating if it is possible to reuse the xid tagging for this, which IMHO should work
1182269785 M * Bertl but I have to write a test suite for that first, because repeating 20+ tests every time is not really fun
1182269843 M * daniel_hozac hehe, i bet.
1182269950 Q * lilalinux Remote host closed the connection
1182270600 M * coderanger daniel_hozac: Because some of the function names in libvserver have changed since 213, and we prefer to keep packages in fedora (instead of dealing with then ourselves) 
1182270617 M * daniel_hozac coderanger: really? i don't remember doing that.
1182270631 M * daniel_hozac i know that i restructured quite a bit for 0.30.213.
1182270651 M * daniel_hozac but since 0.30.213... i don't think lib/vserver.h should've been changed at all.
1182270730 Q * dlezcano Read error: Connection reset by peer
1182270892 J * [BiG^BrotheR] ~Dr-DreaM@84.23.96.253
1182270940 M * jkl are there instructions somewhere on how to cleanly bring up another IP address in a vserver without restarting it?
1182271055 M * Bertl naddress --help
1182271149 Q * mstrobert Ping timeout: 480 seconds
1182271186 M * jkl thanks.
1182271190 M * Bertl np
1182271260 M * jkl do you run that from within the guest or on the host
1182271281 Q * DreaM[BrB] Ping timeout: 480 seconds
1182271295 M * Bertl jkl: on the host
1182271348 M * daniel_hozac (you have to add the IP address to the host too, with e.g. ip addr add)
1182271380 M * jkl oh, and then run naddress to add it to the vserver
1182271389 M * daniel_hozac yep.
1182271396 M * daniel_hozac (though the order doesn't matter)
1182271883 Q * bulkBoy Quit: Leaving
1182272416 Q * pmenier Quit: pmenier
1182273082 J * infowolfe_ ~infowolfe@c-24-10-147-179.hsd1.ut.comcast.net
1182273156 Q * bonbons Quit: Leaving
1182273178 Q * sladen Ping timeout: 480 seconds
1182273501 Q * _jthm_ Ping timeout: 480 seconds
1182273527 M * jkl daniel_hozac, bertl: naddress works great, but it does not add that address information to /etc/vservers/NAME/interfaces/
1182273534 M * jkl is there a way to accomplish this?
1182273549 M * jkl (using naddress)
1182273555 M * Bertl nope
1182273560 M * Bertl use 'echo' for that :)
1182273565 M * jkl :) 
1182273767 J * sladen paul@starsky.19inch.net
1182274265 M * Bertl welcome sladen!
1182274281 M * Bertl sladen: how are you? still kicking?
1182274551 J * DreaM[BrB] ~Dr-DreaM@84.23.96.253
1182274926 Q * [BiG^BrotheR] Ping timeout: 480 seconds
1182275340 J * meandtheshell ~markus@85-124-39-185.dynamic.xdsl-line.inode.at
1182275372 Q * ema Quit: leaving
1182276083 J * [BiG^BrotheR] ~Dr-DreaM@84.23.96.253
1182276441 Q * DreaM[BrB] Ping timeout: 480 seconds
1182276651 J * bzed ~bzed@wireless-1-86.internal.dc7.debconf.org
1182276942 J * bulkBoy ~eSa|@ip-87-238-2-45.adsl.cheapnet.it
1182276958 M * sladen Bertl: @ Debconf, coping with a randomly rebooting personal box
1182277023 M * Bertl sladen: good to hear that you are alive (and well, I hope?)
1182277045 M * sladen Bertl: reasonably, just started a new job in Helsinki
1182277052 M * sladen a change from being freelance
1182277064 M * Bertl ah, nice, and? do you like it?
1182277133 M * sladen Bertl: first week in Helsinki, second in Beijing, third in Edinburgh
1182277145 M * sladen we'll see, maybe things will settle down
1182277839 J * ensc ~irc-ensc@p54B4CEA7.dip.t-dialin.net
1182278819 J * coderanger_ ~laptop@wireless-19-112.media.mit.edu
1182278976 M * dilinger sladen: heh, no more constant traveling?
1182279314 J * hardwire` ~bip@rdbck-1961.wasilla.mtaonline.net
1182279334 Q * hardwire Ping timeout: 480 seconds
1182279942 M * coderanger_ vc_ctx_create says it makes a ctx without starting it, how does one actually start it?
1182279963 M * daniel_hozac hmm
1182279964 M * daniel_hozac ?
1182279999 M * daniel_hozac vc_ctx_create only creates the context, putting the current process in it.
1182280021 M * coderanger_ I thought you needed to do a vc_ctx_migrate to move the current process
1182280031 M * coderanger_ or is that just for entering a previously created ct
1182280043 M * daniel_hozac well, you'll want to do both.
1182280082 M * coderanger_ When we do the migrate its comes back with EPERM (yes this is running as root)
1182280110 M * daniel_hozac from where?
1182280115 M * Bertl coderanger_: STATE_ADMIN gone?
1182280139 M * coderanger_ Bertl: from the default (host?) context, running with all defaults
1182280141 M * daniel_hozac echo 255 > /proc/sys/vserver/debug_switch should help...
1182280165 M * daniel_hozac coderanger_: i meant more like, from where are you calling it. (e.g. what stage of the process)
1182280218 M * coderanger_ created a context, set an single rlimit on it, and then tried to migrate the current python process
1182280254 M * daniel_hozac that should work fine.
1182280268 M * Bertl daniel_hozac: hmm? shouldn't he be already inside the context?
1182280296 M * coderanger_  /proc/sys doesn't exist
1182280336 M * Bertl coderanger_: try sysctl -a
1182280341 M * Bertl or /sys
1182280354 M * daniel_hozac Bertl: yeah, but the way util-vserver gets rid of all the caps is to migrate to the context (on older kernels)
1182280379 M * Bertl daniel_hozac: there is no older kernel :)
1182280389 M * daniel_hozac right.
1182280391 M * daniel_hozac or, hmm.
1182280409 M * daniel_hozac --migrate-self doesn't actually migrate.
1182280410 M * coderanger_ sysctl doesn't see anything related to vserver
1182280421 M * daniel_hozac coderanger_: then you don't have vserver debugging enabled.
1182280421 M * matti Hi Bertl :)
1182280433 M * Bertl coderanger_: who is doing the kernels?
1182280436 M * coderanger_ daniel_hozac: How would I enable that?
1182280442 M * coderanger_ Bertl: On this box, me
1182280447 M * coderanger_ Bertl: I can try this on the XO
1182280463 M * daniel_hozac coderanger_: CONFIG_VSERVER_DEBUG, IIRC.
1182280478 M * Bertl the kernel builds should enable VSERVER_DEBUG for the devel kernels, and disable it for 'PRODUCTION'
1182280492 M * coderanger_ Ahh, okay
1182280504 M * coderanger_ Lemme check the XO kernel t osee if it is on there
1182280558 M * coderanger_ yeah, it is enabled in our XO kernel
1182280565 M * coderanger_ lemme try this whole shebang on there again
1182280618 M * Bertl neuralis: btw, cjb is responsible for the kernel configs or dilinger?
1182280627 Q * Johnnie Read error: Operation timed out
1182280643 M * coderanger_ I think dilinger has been doing all of this
1182280650 M * coderanger_ cjb has been busy with other things
1182280672 M * Bertl neuralis: thing is, we should decide what config we want on what kernels .. not that urgent, but should be done at some point
1182281246 J * Johnnie ~jdlewis@c-67-163-247-109.hsd1.pa.comcast.net
1182281977 J * DreaM[BrB] ~Dr-DreaM@84.23.96.253
1182282221 Q * [BiG^BrotheR] Ping timeout: 480 seconds
1182282247 M * coderanger_ Okay, we now have pylibvserver on an XO
1182282253 M * coderanger_ vc_ctx_create(2)
1182282267 M * coderanger_ vc_ctx_migrate(2, 0) --> EPERM
1182282280 M * fatgoose hey where is that  pylibvserver?
1182282296 M * Bertl coderanger_: turn on, debug_switch with 255
1182282297 M * daniel_hozac on coderanger_'s laptop :)
1182282300 M * coderanger_ fatgoose: Still under active devel, I wouldn't recommend trying to use it yet
1182282326 M * coderanger_ fatgoose: But if you want to look, dev.laptop.org/git?p=security
1182282375 M * coderanger_ Bertl: I activated it using sysctl, and I see some stuff in /var/log/messages
1182282377 A * coderanger_ pastebins
1182282385 M * Bertl okay, tx
1182282420 M * coderanger_ http://paste.lisp.org/display/43065
1182282436 M * coderanger_ Hopefully that means more to you than it does to me ;-)
1182282459 M * Bertl better than that, I'm going to explain it to you :)
1182282521 M * Bertl coderanger_: for example, you can let vcmd interpret the calls, like this:
1182282526 M * Bertl vcmd -ndC VCMD_09_1[0]
1182282573 M * Bertl but, to get you an idea how this works, the calls consist of a category (09) a command (1) and a version (0)
1182282610 M * Bertl you actually turned on more than just the switch, I'd say
1182282627 M * coderanger_ I think I turned on xid too
1182282629 M * daniel_hozac http://linux-vserver.org/VCMD_HowTo
1182282633 Q * rob-84x^ Remote host closed the connection
1182282643 M * daniel_hozac (has a snippet for turning the dmesg output into names)
1182282643 J * bonbons ~bonbons@2001:5c0:85e2:0:20b:5dff:fec7:6b33
1182282653 M * Bertl ah, nice :)
1182282679 M * coderanger_ Is there a fedora package for vcmd somewhere?
1182282680 M * Bertl coderanger_: the values in [] at the end, give some idea where the command 'ended' in the syscall switch
1182282691 M * daniel_hozac coderanger_: no, it's a hack/developer's tool.
1182282696 M * coderanger_ Hmm
1182282704 M * coderanger_ Makes it hard to use on here ;-)
1182282709 Q * Johnnie Ping timeout: 480 seconds
1182282713 M * coderanger_ (we don't put gcc et al on the laptop images)
1182282720 M * daniel_hozac how come? you just compile it statically and copy the binary...
1182282744 M * Bertl coderanger_: shouldn't be a problem for the OLPC distro maintainers to add it to the devel images
1182282756 M * coderanger_ Yeah, just less convenient than yum install vcmd
1182282770 M * Bertl well, I guess they can do an rpm for that too :)
1182282774 M * coderanger_ Bertl: We don't add things that aren't in the fedora repos without a very good reason
1182282815 M * Bertl well, then you have to work around it :)
1182282833 M * coderanger_ I can also just make a spec for it and poke ensc to maintain it :)
1182282853 Q * duckx Quit: Client exiting
1182282855 M * Bertl as I already said last time, I wouldn't consider that a good idea :)
1182282889 M * Bertl the purpose of this tool is to have something _more_than_recent_ to poke at the _new_and_untested_ interfaces :)
1182282911 M * Bertl there is no point in having a version a day older than the latest :)
1182282927 M * coderanger_ Where is the vcmd download again?
1182282933 Q * cedric Quit: cedric
1182282949 M * Bertl http://vserver.13thfloor.at/Experimental/TOOLS/
1182282962 J * derjohn2 ~aj@e180208007.adsl.alicedsl.de
1182283125 J * Johnnie ~jdlewis@c-67-163-247-109.hsd1.pa.comcast.net
1182283330 J * cedric ~cedric@rny93-2-82-66-66-30.fbx.proxad.net
1182284203 Q * HeinMueck Quit: Aah!
1182285232 Q * Vudumen Remote host closed the connection
1182285242 J * Vudumen ~vudumen@perverz.hu
1182286228 Q * sladen Ping timeout: 480 seconds
1182286634 Q * bonbons Quit: Leaving
1182286783 J * yarihm ~yarihm@84-75-109-39.dclient.hispeed.ch
1182286833 M * coderanger_ Should anyone ask in the future, I have a spec/RPM for vcmd (and it can be easily changed for future versions when needed)
1182286859 M * Bertl good!
1182286889 P * stefani I'm Parting (the water)
1182287712 J * Aiken ~james@ppp121-45-250-169.lns2.bne4.internode.on.net
1182287891 M * cedric hi all! how could I limit cpu for a given vserver to 10% of the master capacity?   the doc about the rlimits says cpu(.soft|hard) is a valid resource but I have no idea what would be a legal value in it?
1182288108 Q * fatgoose Quit: fatgoose
1182288795 M * Bertl cedric: really? it does?
1182288819 Q * bzed Quit: Leaving
1182288822 Q * Greek0_ Ping timeout: 480 seconds
1182288822 M * cedric Bertl: ?
1182288858 M * Bertl I wonder about the fact that 'the doc' says something like that ,,,#
1182288881 M * Bertl cedric: what doc are you referring to?
1182288909 M * cedric this one: http://www.nongnu.org/util-vserver/doc/conf/configuration.html    and I think I've seen something similar on the wiki
1182289005 M * Bertl funny thing .. daniel_hozac: maybe that should be changed, or 'how' does it do that for cpu?
1182289036 M * Bertl cedric: thing is, even if that would work (which I doubt), it wouldn't be what you want
1182289068 M * Bertl cedric: it would limit the total amount of cpu for that context in seconds, once the cpu is consumed, the context would be terminated
1182289088 M * Bertl cedric: of course, this is not implemented, because it is basically useless nowadays
1182289099 M * cedric oh ok :)   so, for what I'd like to do I guess I should look at sched then?
1182289109 M * Bertl cedric: you want to look a paragraph further down, yes
1182289134 M * Bertl cedric: and there is a wiki page describing how to set up the scheduler
1182289170 M * Bertl http://linux-vserver.org/CPU_Scheduler
1182289243 M * cedric the "cpu" resource is quite misleading, I would have loved to set something like "10% guaranteed" in the file and then sched/tokenbucket values get selected automagically :p    well it's time to look again at formulas :)
1182289297 M * Bertl the thing is, 10% guaranteed is a very vague description :)
1182289329 M * Bertl could mean you want the guest to consume at least 10% :)
1182289357 M * Bertl or could mean, you want to make sure that everything else consumes 90% at most
1182289833 M * cedric Bertl: hum yeah of course :)     in my case I want a hard limit for vserver consuming 10% at most, so do I just have to set sched/fill-rate to 1 and sched-interval to 10 and restart the vserver to make it work?
1182289855 M * Bertl yes, but that covers only one aspect
1182289876 M * cedric oh? what else do I have to care of?
1182289884 M * Bertl for example, 1/10 or 10/100 or even 100/1000 will give the same percentage (10%)
1182289895 M * Bertl nevertheless, they will give different behaviour
1182289915 M * Bertl you have an idea what a token bucket is? how it works?
1182290093 M * cedric yeah a little, setting 1/10 or 10/100 would have an effect in latency like when changing kernel timer tick right?
1182290112 M * Bertl yep, with large values, the guest will start to stutter
1182290118 M * cedric 1/10 would be high-latency best-throughtput and 10/100 the contrary
1182290127 M * Bertl with too small values, the scheduling overhead will increase
1182290150 M * cedric understood :)
1182290154 M * Bertl besides this aspect, you can also set max and min values for a token bucket
1182290167 M * Bertl which will control the burst behaviour
1182290221 M * Bertl and finally, with recent releases (kernel) you can also set rate/interval for idle time (i.e. when the cpu would go idle)
1182290226 Q * dna Quit: Verlassend
1182290251 M * cedric dynticks you mean?  this work ok with vserver patches?
1182290273 M * Bertl that's not what I'm referring to, but yes, dynticks work :)
1182290393 M * cedric nice!    well thanks a lot for your explanations :)
1182290400 M * Bertl you're welcome!
1182290801 J * [BiG^BrotheR] ~Dr-DreaM@84.23.96.253
1182291181 Q * DreaM[BrB] Ping timeout: 480 seconds
1182291295 N * DoberMann DoberMann[ZZZzzz]
1182291560 Q * meandtheshell Quit: Leaving.
1182291630 M * coderanger_ Bertl: So I am still not sure why I can't enter a new context
1182291647 M * Bertl you first create the context, right?
1182291658 M * coderanger_ libvserver is calling VCMD_get_space_mask [v0] and VCMD_enter_space [v1] 
1182291660 M * coderanger_ yes
1182291671 M * coderanger_ The latter of those is erroring out with EPERM
1182291715 M * Bertl let me check that, maybe an actual bug .. sec
1182291759 M * Bertl where did you paste the debug log?
1182291791 M * coderanger_ Lemme make a new one with the vcmd'd messages
1182291802 M * Bertl okay, please put it on paste.linux-vserver.org
1182291807 J * meandtheshell ~markus@85-124-37-10.dynamic.xdsl-line.inode.at
1182291855 M * coderanger_ http://paste.linux-vserver.org/2659
1182291886 M * coderanger_ That is just vc_ctx_create(2), which works, and then vc_ctx_migrate(2, 0)
1182291899 M * Bertl okay, the enter_space doesn't make much sense here, unless you specify a mask of 0
1182291920 M * coderanger_ Yes, that is the 2nd arg to vc_ctx_migrate
1182291925 M * coderanger_ should I have something else?
1182291927 M * Bertl let me explain the mechanisms you are invoking here:
1182291962 M * Bertl the context create, will make a new context, with (in this case, v0) a default set of flags
1182291978 M * Bertl (you probably want to use the v1 call there)
1182292001 M * coderanger_ libvserver doesn't seem to expose the v1 call, but I can just do it manually I suppose
1182292018 M * Bertl then the get_version and vci will return the version/feature information
1182292032 M * Bertl (doesn't need to be done more than once actually)
1182292045 M * Bertl i.e. the data could and probably should be buffered somewhere
1182292064 M * Bertl the same for the space_mask, which is strictly read only
1182292082 M * coderanger_ again, thats all internal, so poke daniel_hozac ;-)
1182292091 M * Bertl do you want me to explain the details of those calls?
1182292106 M * coderanger_ I think I understand them
1182292121 M * Bertl okay, the context has no spaces (ns/uts/ipc) assigned so far
1182292127 M * coderanger_ I just don't understand why I don't have permissions to enter the space
1182292138 M * Bertl so enter_space is simply stupid to do at that point
1182292150 M * coderanger_ Yes, it should be a totally useless container
1182292158 M * coderanger_ i.e., it doesn't actually change anything
1182292176 M * Bertl okay, let's look at the code (kernel) what actually fails and why
1182292214 M * Bertl         __VCMD(enter_space,      5, VCA_VXI,    VCF_ADMIN);
1182292232 M * Bertl so that says that you have to be host admin to call it
1182292267 M * Bertl and, because you already created a context, and (probably) got automigrated into it (you can check that with get_vx_info())
1182292287 M * Bertl you do not possess host admin rights, and get an EPERM
1182292293 M * Bertl (now let's verify that in the code)
1182292351 M * coderanger_ vc_get_task_xid(os.getpid()) == VC_NOCTX ====> True
1182292437 M * Bertl [0,5,1,2] -> [compat, perm, args, flags]
1182292440 J * markus_ ~chatzilla@chello213047089232.17.14.vie.surfer.at
1182292445 M * markus_ hi
1182292472 M * Bertl coderanger_: you should also get a line like this:
1182292479 M * Bertl vc: VCMD_%02d_%d[%d] = %08lx(%ld) [%d,%d]
1182292496 M * markus_ How can I restrict a specific vserver to have a specific amount of memory? I'm looking at http://linux-vserver.org/util-vserver:Documentation but can't find anything .. ?
1182292528 M * Bertl you can set limits for VM/AS as well as RSS and locked memory
1182292580 M * coderanger_ Bertl: You mean in the kernel log?
1182292582 M * Bertl markus_: 
1182292589 M * Bertl http://linux-vserver.org/Memory_Limits
1182292613 M * Bertl coderanger_: yep, if not, please set debug_switch to 255
1182292724 M * coderanger_ Okay,  I see them
1182292746 M * Bertl we are interested in the matching line for this call
1182292749 Q * arachnist Remote host closed the connection
1182292770 M * coderanger_ [88007.126220] VCMD_enter_space [v1]          = ffffffff(-1) [5,0]
1182292803 M * markus_ Bertl: thanks! Not exactly a no-brainer but the project doesn't advertise it to be so ;)
1182292804 M * Bertl so -1 is the return value (means error) and [5,0] -> [state, permit]
1182292831 M * Bertl markus_: you're welcome! 
1182292842 M * coderanger_ Bertl: What does state=5 mean?
1182292855 M * Bertl check kernel/vserver/switch.c (kernel code)
1182292879 M * Bertl you will see that (line ~393) the state is incremented like a post code
1182292936 M * Bertl in this case, the one which failed is the admin check (as I suspected)
1182292981 M * Bertl coderanger_: you are already in context 2, as administrating process
1182293006 M * Bertl coderanger_: thus, you cannot invoke the enter_space()
1182293145 M * markus_ Bertl: are the as/rss limits reflected inside the vserver, e.g. by calling free? I can't see a difference there although I see processed being killed or vserver not starting if the values are too low
1182293158 M * coderanger_ Bertl: Okay, I'll dig some more, thanky
1182293193 M * Bertl markus_: yes, the rss hard limit is shown as RAM+Swap, while the rss soft limit is shown as RAM
1182293213 M * Bertl markus_: once you hit the hard limit, the OOM killer will strike
1182293248 M * Bertl coderanger_: np, just ask if something is unclear ...
1182293259 M * markus_ hmm .. my output of "free" inside vserver and outside matches exactly although I set pretty low rss/as limits for this vserver (10000, 10000)
1182293277 M * Bertl http://linux-vserver.org/Capabilities_and_Flags
1182293288 M * Bertl check out VIRT_* especially VIRT_MEM :)
1182293465 M * markus_ wow
1182293484 M * markus_ although I don't understand all of that stuff I think I got it working ;) I know see only the 4kb pages. kewl :)
1182293799 M * markus_ thanks a lot and good night!
1182293801 Q * markus_ Quit: ChatZilla 0.9.78.1 [Firefox 2.0.0.4/2007051502]
1182293858 J * DreaM[BrB] ~Dr-DreaM@84.23.96.253
1182293877 Q * DreaM[BrB] 
1182294011 Q * [BiG^BrotheR] Ping timeout: 480 seconds
1182294965 J * derjohn3 ~aj@e180206227.adsl.alicedsl.de
1182295408 Q * derjohn2 Ping timeout: 480 seconds
1182295595 Q * jmt Quit: Ciao!
1182295672 J * bzed ~bzed@wireless-1-249.internal.dc7.debconf.org
1182295688 J * Daniel15 ~daniel@c211-28-82-207.sunsh7.vic.optusnet.com.au
1182295779 M * Daniel15 Hi guys :)
1182295791 M * Daniel15 I've just got a quick question...
1182295797 Q * lylix Read error: Connection reset by peer
1182295810 M * Daniel15 If I give a vserver the server's main IP address, will anything bad happen?
1182295832 M * Daniel15 I'm assuming that setting "net.ipv4.conf.all.promote_secondaries=1" in sysctl.conf is all that's needed?
1182295863 M * Bertl actually nothing is needed
1182295881 M * Bertl you should just make sure that the tools do not take that ip up and down for you
1182295895 M * Bertl otherwise a guest shutdown will 'kill' your network connection
1182295930 M * Daniel15 All I'm using is the "newvserver" script in Debian to create the vserver, and vserver <name> start/stop to stop it
1182295943 M * Daniel15 What should I do to make sure it doesn't take the ip down?
1182295958 Q * cedric Quit: cedric
1182295982 M * Bertl Daniel15: first, throw away the newvserver script, it's deprecated and useless
1182296012 M * Bertl Daniel15: use 'vserver <name> build ...' instead
1182296032 M * Daniel15 The only reason I was using it was because it does things like set the timezone, set the root password, and set the locale... But I suppose I could do all that manually
1182296035 M * Bertl and you want to avoid specifying any ethX or alias name in the config
1182296051 M * Daniel15 OK, so just the IP address, no interface name?
1182296063 M * Bertl Daniel15: that should be handled by the post installation scripts of util-vserver
1182296090 M * Bertl Daniel15: you might want to check that debian uses them and/or report that to the maintainers
1182296118 M * Bertl yep, you want to have the ip and netmask/prefix, but not an interface (you want nodev instead)
1182296217 M * Bertl Daniel15: btw, is there a reason you want to give your host ip to the guest?
1182296273 M * Daniel15 Long story there... I'm going to be installing Plesk inside the vserver, but its license is IP-limited. I don't really want to change the server's main IP address to another one as I've already screwed up the /etc/network/interfaces file once (and had to get the data centre to fix it)
1182296360 M * Bertl i.c. and the host already has the proper ip, yes?
1182296360 J * sladen paul@starsky.19inch.net
1182296481 M * Daniel15 Yeah, the host has that particular IP assigned to eth0 (I have other IPs assigned to aliases eth0:0, eth0:1, etc.)
1182296496 M * Daniel15 vserver web1 build -m debootstrap --rootdir /home/vservers/ --hostname web1.daniel15.com --interface nodev:66.79.185.80/255.255.255.248  -- -d sarge -m http://204.152.191.7/debian/
1182296498 M * Daniel15 Would that be fine?
1182296511 M * Bertl ah, i.c and all of them (ips) are public ones?
1182296516 M * Daniel15 Yeah, all public
1182296553 M * Bertl the line would be fine, except for the fact that the debian version probably won't know about nodev
1182296571 M * Bertl so better leave that out, i.e. --interface 66.79.185.80/255.255.255.248
1182296587 M * Daniel15 OK
1182296600 M * Daniel15 (yeah, i'm using an old Debian version... Plesk doesn't work on Debian Etch yet :P)
1182296615 M * Daniel15 "No device specified for interface '0'; do not forget to set the 'nodev' option"
1182296620 M * Bertl you probably have separate network scripts for eth0/eth0:0 ...
1182296620 M * Daniel15 I guess I can ignore that message?
1182296645 M * Daniel15 My host is Debian Etch, but the vserver guest will be Sarge.
1182296650 M * Bertl you can set that with a simple 'touch /etc/vservers/<name>/ip/0/nodev
1182296886 M * Daniel15 Hmmm... I get some messages when shutting down the vserver.
1182296887 M * Daniel15 http://paste.linux-vserver.org/2662
1182296904 M * Daniel15 I think it's still trying to shutdown the interface: "Rebooting... ifdown: shutdown eth0: Permission denied "
1182296926 M * Bertl yep, the guest is not cleaned up properly yet
1182296940 M * Bertl means the debian version does not support a proper post init script
1182296972 M * Daniel15 What can I do to fix that?
1182296973 M * Bertl you might want to remove all hardware/filesystem related runlevel scripts
1182297001 M * Bertl i.e. stuff used in /etc/rc?.d
1182297039 M * Bertl and keep only those for services you are actually running inside the guest
1182297202 M * coderanger_ Bertl: So I really don't understand this, my ccaps are all set
1182297212 M * coderanger_ Where do I check for the admin bit?
1182297233 M * Daniel15 OK, thanks for all your help Bertl :D
1182297240 M * Daniel15 I have to go now; see you later :)
1182297242 Q * Daniel15 Quit: Leaving
1182297246 M * Bertl coderanger_: you lost that when you entered a context, i.e. you cannot be in a context and have host admin
1182297300 M * coderanger_ Bertl: Does making a new context migrate you automatically?
1182297307 M * Bertl yes
1182297319 M * Bertl you become the admin process of this context
1182297335 M * Bertl only exception are persistant contexts, which do not auto migrate you
1182297336 M * coderanger_ Hmm, then why does vc_get_task_xid($MYPID) not return 2?
1182297368 M * Bertl because you are probably still in setup mode
1182297381 M * Bertl i.e. you didn't clear the ADMIN_SETUP flag yet
1182297387 M * coderanger_ I think there is a step here that I am missing ;-)
1182297401 M * Bertl sorry, INIT_SETUP flag
1182297415 M * coderanger_ Okay, which call do I use for that?
1182297436 M * coderanger_ I would imagine that is in the vc_get/set_cflags
1182297493 M * coderanger_ I see a VC_VXF_STATE_*
1182297515 M * Bertl yep
1182297523 M * Bertl it's late here :)
1182297550 M * Bertl  STATE_SETUP (bit 32)
1182297565 M * Bertl this is the one, which closes the setup when you remove it
1182297568 M * coderanger_ So there is VC_VXF_STATE_SETUP and VC_VXF_STATE_INIT, do I clear both?
1182297573 M * coderanger_ Okay
1182297596 M * Bertl when you clear the STATE_INIT flag, you process will become the context's init process