1140912301 Q * Loki|muh Quit: leaving
1140912383 M * tek_ ++
1140912388 P * tek_ Leaving
1140912833 J * Loki|muh loki@satanix.de
1140913594 J * matta ~matta@c-68-81-35-243.hsd1.pa.comcast.net
1140914164 J * matt1 ~matta@c-68-81-35-243.hsd1.pa.comcast.net
1140914499 Q * matta Ping timeout: 480 seconds
1140915129 Q * matt1 Ping timeout: 480 seconds
1140922474 J * dothebart ~willi@xdsl-213-196-253-213.netcologne.de
1140922901 Q * tudenbart Ping timeout: 480 seconds
1140923561 J * Smutje_ ~Smutje@xdsl-87-78-6-67.netcologne.de
1140923669 Q * Smutje Ping timeout: 480 seconds
1140923669 N * Smutje_ Smutje
1140924045 M * SuperLag Hollow: you still around?
1140924579 M * SuperLag Hollow: baselayout-vserver is broken.  No init scripts will start.
1140924584 M * SuperLag Hollow: http://rafb.net/paste/results/yPybKc42.html
1140926043 M * Hollow SuperLag: make.conf.example is only an example file..
1140926159 M * Hollow but /lib is no symlink..
1140926174 M * Hollow that's why the init scripts fail
1140926200 M * Hollow i'll look at this tomorrow... just woke up in front of the tv..
1140926487 M * SuperLag Hollow: sorry, meant to say /etc/make.conf :)
1140926528 J * matta ~matta@71.224.125.126
1140926615 M * Hollow dionysos test # grep CHOST etc/make.conf
1140926615 M * Hollow CHOST="x86_64-pc-linux-gnu"
1140926652 M * SuperLag Okay.
1140926655 M * Hollow regarding the pts issue... http://list.linux-vserver.org/archive/vserver/msg12117.html and http://dev.croup.de/proj/vserver-utils/browser/trunk/src/tools/vlogin.c?rev=98
1140927075 M * Hollow off to bed then.. cu tomorrow..
1140927656 Q * teukka Ping timeout: 480 seconds
1140932604 J * fwl ~fwl@83.215.237.1
1140932711 Q * fwl Quit: 
1140932815 J * fwl ~fwl@83.215.237.1
1140934257 Q * fwl Quit: This computer has gone to sleep
1140934348 J * fwl ~fwl@83.215.237.1
1140934760 Q * fwl Quit: This computer has gone to sleep
1140937447 J * oliwel ~mail-at-o@host-62-245-151-178.customer.m-online.net
1140937488 A * oliwel waves hello to the crowd
1140937497 M * oliwel anybody in ?
1140939337 Q * oliwel Quit: Chatzilla 0.9.69.1 [Firefox 1.5/2005111116]
1140944091 J * fwl ~fwl@83.215.237.1
1140944201 J * bonbons ~bonbons@83.222.39.180
1140948159 J * PilatomiK ~tek@ADijon-151-1-132-133.w86-204.abo.wanadoo.fr
1140948170 P * PilatomiK 
1140948697 M * bonbons Bertl_zZ: I tried latest patch (2.6.16-rc4-vs2.1.1-rc9) to see if I get the same kind of kernel oops as Oliver Welter or phreak`` but no error
1140948733 J * Dan99991 ~chatzilla@c-eb83e055.124-1-64736c12.cust.bredbandsbolaget.se
1140948745 J * teukka ~tmatilai@backport.ri.fi
1140948759 M * daniel_hozac bonbons: and you're using fakeinit?
1140948785 M * daniel_hozac i mean, init inside the guest.
1140948796 M * bonbons I'm running mostly reiserfs partitions, except ext2 for boot and /usr/portage.
1140948812 M * bonbons no, guests are all with a real init
1140948830 M * daniel_hozac that would be yes then :)
1140948838 P * Dan99991 
1140948939 M * bonbons hm, understood fake init the other way round: just some ghost entry displayed for pid=1, but not real process behind :)
1140948983 M * daniel_hozac well, the fake shouldn't have been there :)
1140949007 M * bonbons might it be that the bug is more related to ext2/3 changes in rc8/rc9 than the reaper thing?
1140949068 M * phreak`` bonbons: I'm only running ext2 and thats on /boot .. all other partitions are xfs
1140949099 M * daniel_hozac bonbons: the BUG_ON is reaper related.
1140949172 M * bonbons though nothing complains on my side, other difference I have, on my system I'm using vserver-utils, not util-vserver
1140949200 M * daniel_hozac that's probably i.
1140949214 M * daniel_hozac +t
1140949291 M * bonbons phreak``: is your guest vunified, or behind a chroot-barrier?
1140949328 M * phreak`` bonbons: chroot-barrier .. unification doesn't work *that* well on Gentoo ;)
1140949340 M * daniel_hozac bonbons: could you try with util-vserver?
1140949357 M * phreak`` (if your have different useflags within the vservers)
1140949365 M * phreak`` *you even
1140949400 M * bonbons I don't have the barrier nor do I have unification... Will try out with util-vserver
1140949439 Q * fwl Quit: This computer has gone to sleep
1140949467 M * bonbons phreak``: well there's a good deal of static files /usr/share/** that could be unified, even if some binaries differ (like by compiled-in compile-time)
1140949663 J * Dan99991 ~chatzilla@c-eb83e055.124-1-64736c12.cust.bredbandsbolaget.se
1140949668 P * Dan99991 
1140950173 M * bonbons daniel_hozac, phreak``: is there a specific version of util-vserver I should use? (0.30.209, .210, .210-r1, .210-r2)
1140950217 M * daniel_hozac either one should show it, IMHO.
1140950247 M * phreak`` bonbons: use .210 :)
1140950755 M * bonbons with util-vserver I get the oops as well...
1140950848 J * fwl ~fwl@83.215.237.1
1140951141 M * daniel_hozac what does your trace look like?
1140951326 M * phreak`` bonbons: same as http://phreak.xnull.de/oops.txt ?
1140951340 M * bonbons Mine is here: http://pastebin.com/573030
1140951777 M * bonbons I was using rc9, will also try with rc8. After that I will look at the sources for util-vserver and vserver-utils to eventually get out what they may so differently
1140951799 M * daniel_hozac doesn't vserver-utils set up the context from outside?
1140951816 M * phreak`` bonbons: could you please have a look at kernel/exit.c and see whats in line 551 ? :)
1140951840 M * daniel_hozac phreak``: the choose_new_parent BUG_ON.
1140951885 M * daniel_hozac i _think_ the bug is because vcontext number 1 exits, but it is its child that starts the init.
1140951924 M * phreak`` daniel_hozac: yeah, same here
1140951929 M * bonbons exactly
1140951984 M * phreak`` its freaking (or oops'ing :P) on this one: BUG_ON(p == reaper);
1140952029 M * phreak`` (Bertl_zZ to split this BUG_ON into two, to see which one was the cause)
1140952045 M * daniel_hozac yeah.
1140952074 M * phreak`` bah, *told me to even .. (dammit, wake up brain)
1140952123 M * bonbons I'm recompiling with the BUG_ON split and rebooting to test again
1140952475 M * bonbons yep, bails out on (p == reaper)... retrying with the zombie state first
1140952570 M * daniel_hozac execve doesn't cause a do_exit, does it?
1140952762 M * bonbons it's only the (p == reaper) part that bails out
1140952793 J * Viper0482 ~Viper0482@p54976FAA.dip.t-dialin.net
1140953263 M * daniel_hozac bonbons: what does addr2line say about c0121bc5?
1140953378 M * bonbons daniel_hozac: on what binary?
1140953413 M * daniel_hozac vmlinux of your pastebin.
1140953424 M * daniel_hozac or whatever EIP is on your latest oops/kernel.
1140953571 M * bonbons hmm, I just have bzImage and vmlinux.bin, but addr2line does not recognize either once
1140953639 M * bonbons ok, found, was looking at wrong place -> exit.c:0
1140956634 J * Wenix ~jrc@cpe.atm0-0-0-1101164.0x50a6e7aa.bynxx5.customer.tele.dk
1140960579 M * Wenix I've just upgrade to 2.6.15.4-vs2.1.1-rc9-gentoo (utils: 1.0.3) (lib: 1.0.2), and it seems that alot of things have changed since 2.1.0-rc4
1140960600 M * Wenix I've create the context.conf file and now when I try to start a vserver I get this error:
1140960616 M * Wenix *** glibc detected *** free(): invalid pointer: 0x00007fffffbfebe0 ***
1140960616 M * Wenix /usr/share/vserver-utils/lib/vps.sh: line 346: 29718 Aborted                 ${_VNAMESPACE} -E -x ${VX_XID} -- ${_VMOUNT} -M -f ${fstab} -m etc/mtab
1140960619 M * Wenix An error occured while trying to mount filesystems for 'jrc.wenix.dk'
1140960630 M * Wenix .. can anyone point me in the right directions?
1140960684 M * daniel_hozac bonbons and Hollow are probably the only ones capable of that :)
1140960716 M * Wenix ok, its not a common part of an upgrade? ;)
1140960727 M * Wenix I thought I forgot to do something obvious
1140960737 M * daniel_hozac vserver-utils and libvserver are not really mainstream yet.
1140960754 M * daniel_hozac most of us still use util-vserver.
1140960776 M * daniel_hozac (regardless of how insane the code makes you :))
1140960785 M * Wenix ah, no worries then - I'll downgrade a bit then
1140960809 Q * mnemoc Ping timeout: 480 seconds
1140960821 M * daniel_hozac i'm sure bonbons and Hollow will want to fix that though.
1140961065 M * Wenix oh, I have no doubt in that - but I think I'll prefer to be a bit more mainstream on this server
1140961169 J * mnemoc ~amery@200.73.88.3
1140961438 M * bonbons Wenix: can you provide your context.conf (pastebin.com or upload somewhere)?
1140961472 M * bonbons Second question: are libvserver and vserver-utils compiled against glibc or dietlibc?
1140961615 M * yang Is it possible to have ipv6 on guest?
1140961630 M * bonbons yang: not yet
1140961786 M * daniel_hozac bonbons: how far have you gotten?
1140961869 M * bonbons currently in progress for adding/adjusting code in IPv6 network stack (with small pause for initng inside)
1140961912 M * daniel_hozac dynamicized the addresses?
1140962022 M * bonbons userspace-kernel part is unchanged since patch I uploaded (and works, just no dynamic memory management for the address lists)
1140962055 M * daniel_hozac hmm, you didn't change the interface, did you?
1140962080 M * daniel_hozac (hint, i've lost the URL :))
1140962164 M * bonbons http://homepage.internet.lu/brunop/net_ipv6.diff (well, just replaced the uint32[4] by a union with in6_addr, theoretically same size)
1140962193 M * bonbons though that patch looks old
1140962240 M * bonbons wait, will refresh patch with Bertls comments fixed
1140962385 M * bonbons http://homepage.internet.lu/brunop/vserver/net_ipv6.diff
1140962624 M * bonbons Wenix: this patch might be of interest for you: http://homepage.internet.lu/brunop/vserver/vserver-utils-1.0.3.1.diff
1140962683 M * bonbons daniel_hozac: did you find out something in relation to the util-vserver+vreaper issue?
1140962724 M * bubulak  /t
1140962728 M * daniel_hozac bonbons: no, i haven't tried reproducing it... my test machine is off atm.
1140962754 M * daniel_hozac bonbons: you know IPv4 has a struct in_addr too, right?
1140962786 M * bonbons yep, but just left the ipv4 part nearly untouched
1140962790 M * daniel_hozac but personally, i'd prefer to keep the vcmd_net_addr_v0 the same.
1140962873 M * daniel_hozac and if userspace supplies masks, isn't it easier to just keep that rather than convert it to a prefix?
1140962965 M * bonbons for IPv6 mask as such is never used as far as I know, always done by prefix 48, 64, 128 is all I've seen in use yet
1140962991 M * daniel_hozac but prefix and mask refer to the same thing.
1140963017 M * daniel_hozac and the vcmd_net_addr_v0 has mask.
1140963087 M * bonbons not really, mask defines the bits as list, prefix tells how many bits are set starting from MSB. In vcmd_net_addr_v0 I'm using mask either way depending on value of type
1140963287 M * daniel_hozac but for any given mask, there's a prefix. for any given prefix, there's a mask.
1140963366 M * bonbons no, that works only in one direction: for any prefix there's a mask, not the other way round! /4 ~11110000, 10110000 has no equivalent prefix
1140963403 M * daniel_hozac but netmasks must be consecutive 1s, followed by consecutive 0s.
1140963468 M * Hollow *bounce*
1140963474 M * daniel_hozac so 10110000 is invalid.
1140963487 M * daniel_hozac morning Hollow.
1140963507 M * Hollow hi daniel_hozac 
1140963522 M * Hollow this amd64 stage drives me crazy
1140963526 M * bonbons Hi Hollow
1140963530 N * Bertl_zZ Bertl
1140963534 M * Bertl morning folks!
1140963537 M * Hollow morning Bertl 
1140963544 M * bonbons morning Bertl
1140963551 M * daniel_hozac morning Bertl
1140963623 M * Hollow hm, is there ipv6 support already?
1140963631 M * daniel_hozac bonbons is working on it ;)
1140963641 M * Hollow ah
1140963691 M * Hollow bonbons: the chdir("/"); you added to vnamespace... is it for a particular reason?
1140963738 M * Hollow ah.. and a console->stdin redirect.. nice things i see
1140963740 M * Hollow *g*
1140963816 M * bonbons Hollow: hm, that's already some time ago, need to check :)
1140964068 M * bonbons not sure by now, think it's for rbind, to make sure getting at right/valid place
1140964131 M * bonbons the console->stdin is to get rid of many anoyances with initng ;) just still missing something to get the associated tty correct :(
1140964228 M * Hollow ic..
1140964952 M * bonbons Hollow: what does vserver-utils do so differently when starting guest compared to util-vserver? Should be something related to fork/exit/exec* (and SIGCHLD)
1140965396 M * Hollow hm, donÄt know the differences..
1140965409 M * Hollow the code is too different anyway
1140965535 M * bonbons something in util-vserver triggers a reparenting issue which does not happen in vserver-utils (aka vreaper)...
1140965978 M * Wenix bonbons: I'm back - I'll just take a look at what you posted
1140966020 M * Wenix bonbons: Its compiled against glibc
1140966024 M * Wenix (afaik)
1140966118 M * bonbons that can cause trouble for communication with kernel -- better use dietlibc (check is simple, ldd /usr/sbin/vexec, static binary ~ dietlibc)
1140966212 M * bonbons Wenix: but the error you get is probably unrelated to this fact, would be useful to see your config (possibly unset/incorrectly set variable)
1140966771 M * Wenix bonbons: just a moment
1140966804 M * Wenix bonbons: I were just about to downgrade, so I am just upgrading again
1140966877 M * Bertl okay, I finnaly found some time for the vreaper issue, will investigate now ...
1140966980 M * Bertl do we know who reported that first?
1140967058 M * Bertl was phreak`` I think ...
1140967142 M * bonbons Bertl: I think too, seems anybody on util-vserver + rc8/9 can reproduce it, just vserver-utils does not trigger the issue (at least for me)
1140967170 M * Bertl yup, I know what happens, just not why yet ...
1140967334 M * Bertl http://vserver.13thfloor.at/Stuff/BUGHUNT/phreak-0001/
1140967338 M * Bertl (will add details here)
1140967450 M * Bertl have to find a way to reproduce it without complete guest start
1140967480 M * Wenix Is there any documentation anywhere on the context.conf?
1140967572 M * bonbons Wenix: none I know of (ther's just the example, and reading the code)
1140967595 M * Wenix This example, I suppose: http://dev.croup.de/proj/vserver-utils/browser/trunk/etc/examples/context.conf?rev=51
1140967617 M * bonbons exactly
1140967650 M * Wenix bonbons: I haven't set much more than VX_XID=1000, VX_INIT="gentoo", VX_SHELL="/bin/bash"... just tried with a minimal configuration
1140967683 M * Wenix bonbons: I think you are right that the problem may be unset variables
1140967706 M * bonbons I'm currently assuming it's the missing VDIR that causes the error...
1140967721 M * Wenix Aye, I'll take a look
1140967802 M * bonbons and I see why the scripts don't bail out, missing quotes
1140967891 M * bonbons In vps.sh, find line "if [ ! -d ${VDIR} ]; then", and add quotes arount ${VDIR}, that should fix it on the script side
1140967901 M * Wenix ok
1140967905 M * Wenix I'll do that
1140968007 J * tudenbart ~willi@xdsl-213-196-240-243.netcologne.de
1140968036 M * Wenix VDIR is set
1140968053 M * Wenix It doesn't seem like the problem
1140968055 M * Wenix VDIR: /vservers/jrc.wenix.dk
1140968055 M * Wenix *** glibc detected *** free(): invalid pointer: 0x00007fffffaaf6a0 ***
1140968055 M * Wenix /usr/share/vserver-utils/lib/vps.sh: line 348: 14043 Aborted                 ${_VNAMESPACE} -E -x ${VX_XID} -- ${_VMOUNT} -M -f ${fstab} -m etc/mtab
1140968058 M * Wenix An error occured while trying to mount filesystems for 'jrc.wenix.dk'
1140968113 M * Bertl hmm, seems my devel machine is gone again *sigh*
1140968408 M * Wenix This is the line that fails:
1140968412 M * Wenix  /usr/sbin/vnamespace -E -x 1000 -- /usr/sbin/vmount -M -f /usr/share/vserver-utils/defaults/fstab -m etc/mtab
1140968421 M * bonbons Wenix: can you get more output? Run vserver with env VUDEBUG=1 and UVDEBUG=1 (don't know which of both variables is the one to enable debugging in 1.0.3) and save all the output to a file (e.g. vserver start guest 2&>1 | tee log)
1140968435 Q * dothebart Ping timeout: 480 seconds
1140968475 M * Wenix hold on, removing my old debug info first ;) - didn't know this option
1140968708 M * Wenix http://pastebin.com/573308
1140968804 M * bonbons Wenix: is the context still up, or did it already get released? ([ -d /proc/virtual/1000 ] && echo "Still active")
1140968833 M * Wenix bonbons: its down
1140968920 M * Wenix bonbons: I think its important to mention that I haven't changed anything in the configuration since 2.1.0-rc4 except for adding the context.conf file - is there anything else I need to do?
1140969136 M * bonbons stracing vnamespace or vmount would be good, currently looking into vmount. For one more test, could you replace "-m etc/mtab" by "-n"?
1140969155 M * bonbons Looks like vmount has the bug in mtab code
1140969155 M * Wenix sure, hold on 
1140969438 M * Wenix bonbons: with "-n etc/mtab", no UVDEBUG, no VUDEBUG, and strace  -- http://pastebin.com/573325
1140969462 M * Wenix bonbons: I only pasted the last page of output - let me know if you need more
1140969483 J * Smutje_ ~Smutje@xdsl-87-78-17-192.netcologne.de
1140969589 Q * Smutje Ping timeout: 480 seconds
1140969589 N * Smutje_ Smutje
1140969642 M * bonbons better add the strace inside lib/vps.sh on the execution of vnamespace, stracing bash is overkill (and makes reading strace output hard -- what belongs to which process)
1140969657 M * Wenix ah, I see
1140969665 M * Wenix I
1140969825 M * Wenix more sane strace: http://pastebin.com/573340
1140970370 M * bonbons Yep, bug seen "free(cwd);" on line 253 in vmount.c, cwd is declared as char cwd[PATH_MAX];
1140970650 M * bonbons Hollow: Until vserver-utils gets sufficiently complete for next release, it would probably be useful to open a 1.0.* branch and fix the various 1.0.3 bugs
1140970709 M * Hollow yup, we can do that
1140971278 M * Wenix bonbons: Will you post a patch? or shall I get it from the repository?
1140971399 M * bonbons for this bug the fix is to remove the free(cwd) on line 253 in vmount.c
1140971435 M * Wenix ok, I can manage that ;)
1140971792 Q * shuri Remote host closed the connection
1140971990 M * SuperLag Hollow: howdy.
1140972035 M * Hollow howdy SuperLag 
1140972040 M * SuperLag Hollow: so... you said the baselayout thing was a symlink issue.  I tried manually creating the link from /lib to /lib64.
1140972043 M * SuperLag no dice
1140972073 M * daniel_hozac surely you don't want that.
1140972081 M * Hollow SuperLag: i found out that, during stage compilation, when glibc is installed somehow the symlink gets lost, and is replaced by a normal dir... but dunno why this is so yet
1140972096 M * SuperLag daniel_hozac: don't want what?
1140972106 M * daniel_hozac /lib being a symlink to /lib64
1140972115 M * Hollow daniel_hozac: it's default for gentoo on amd64
1140972116 M * SuperLag absolutely
1140972129 M * daniel_hozac so 32-bit libc goes in /lib64?
1140972133 M * daniel_hozac that seems... odd.
1140972144 M * Hollow no, it goes to /lib32
1140972146 M * SuperLag no
1140972155 M * SuperLag beat me to to :)
1140972158 M * SuperLag to it
1140972177 M * daniel_hozac ah, right, no packages.
1140972180 M * daniel_hozac i forget :)
1140972195 M * Hollow :)
1140972356 Q * shedi Quit: Leaving
1140972955 J * jg ~jg@85.102.207.223
1140972982 P * jg 
1140973257 J * shedi ~siggi@inferno.lhi.is
1140973266 M * Hollow bonbons: vcc started a guest for the first time :)
1140973304 M * bonbons cool :)
1140973455 M * Hollow but.. clone is evil
1140973466 M * daniel_hozac how so?
1140973499 M * Hollow well, not quite evil... but i still somehow fail to setup a new stack for the child in the right way
1140973628 M * daniel_hozac hmm, do you really need a stack? doesn't the child just execve something?
1140973769 M * Hollow not necessarily..
1140973805 M * Hollow i want to use CLONE_VM to be able to share global vars, but it seems to need a new stack, but if i create one, the child immediately gets SIGSEGV
1140973818 M * daniel_hozac how do you create it?
1140973847 M * Hollow void **newstack = (void **) malloc(16384); newstack = newstack = (void **) (STACKSIZE + (char *) newstack);
1140973858 M * Hollow STACKSIZE = 16384
1140973869 M * daniel_hozac why void **?
1140973875 M * Hollow why not?
1140973890 M * daniel_hozac it looks really odd.
1140973900 M * cehteh how about void***************** ?
1140973903 M * Hollow i took it from this example: http://www.tldp.org/FAQ/Threads-FAQ/clone.c
1140973905 J * `DoM`` ~DoM@151.56.240.80
1140973907 M * `DoM`` lo all
1140973915 M * daniel_hozac have you tried with just void *?
1140973925 M * cehteh and not casting malloc :)
1140973925 M * Hollow hm, no..
1140973940 M * cehteh void *newstack = malloc(16384);
1140973941 M * Hollow but i'm creating a test app atm.. so i'll try
1140973970 M * cehteh casting malloc is considered bad practice by a lot of people
1140973985 M * cehteh besides it actually returns void* ... 
1140973991 M * daniel_hozac yes, indeed.
1140974003 M * daniel_hozac you should never have to cast malloc.
1140974008 M * cehteh but even if not it doesnt need the cast
1140974019 M * Hollow i just did c&p...
1140974025 M * Hollow normally i don't cast malloc
1140974036 M * cehteh no excuses! :P
1140974229 M * daniel_hozac Hollow: you noticed the last revised date on that FAQ, right?
1140974236 M * daniel_hozac "(Last revised 21 Jan 1997)"
1140974411 M * Hollow so, how do i create the stack?
1140974439 M * daniel_hozac void *newstack = malloc(16384); newstack += 16384;?
1140974466 M * Hollow doesn't work
1140974519 M * Hollow http://phpfi.com/104234
1140974524 M * cehteh  newstack = ((char*)newstack) + 16384;
1140974527 M * daniel_hozac unrelated: why are you using clone anyway?
1140974546 M * daniel_hozac void * is like char *.
1140974554 M * cehteh well i didnt followed, what do you try to do anyways?
1140974556 M * daniel_hozac in addtion, etc.
1140974564 M * cehteh daniel_hozac: ah ok
1140974574 M * Hollow because of CLONE_NEWNS mainly
1140974581 M * cehteh forgot that .. usually i try to avoid obscure constructs :)
1140974769 M * daniel_hozac i have to admit i've never touched clone...
1140974813 M * Hollow because it's evil! ;)
1140974845 M * daniel_hozac fork has always done what i wanted :)
1140974905 M * Bertl including creating new namespaces?
1140974943 M * daniel_hozac i've never had to do that ;)
1140974983 M * Hollow Bertl: any advice on clone?
1140975202 M * Bertl Hollow: use the syscall directly, forget about the *libc implementations, they are not really tested :)
1140975224 M * Hollow i'm currently using your shiny8
1140975235 M * Bertl good :)
1140975247 M * Hollow no, doesnÄt work either :)
1140975271 M * Hollow see http://phpfi.com/104234
1140975356 M * Bertl newstack += 16384; seems wrong
1140975372 M * Bertl see C void pointer arithmetic
1140975376 M * Hollow i tried so many different ways for the newstack..
1140975403 M * Hollow 1) void **newstack = (void **) malloc(16384); newstack = newstack = (void **) (STACKSIZE + (char *) newstack);
1140975409 M * Hollow 2) newstack = ((char*)newstack) + 16384;
1140975424 M * Hollow 3) void *newstack = malloc(16384); clone(..., newstack + 16384, ...);
1140975426 M * Hollow nothing works
1140975443 M * Bertl use printf("%p, %p, %ld\n", mem, newstack, newstack-mem);
1140975453 M * Hollow and mem is?
1140975463 M * Bertl and assign void *mem = malloc(16384);
1140975475 M * Bertl then do void *newstack = mem;
1140975500 M * Bertl (just to get an idea what you do there)
1140975564 M * Hollow ok, and then pass newstack to clone?
1140975592 M * Bertl IIRC, the child stack might be NULL for sys_clone()
1140975606 M * Hollow yeah, if you don't use CLONE_VM NULL is ok..
1140975623 M * Bertl okay
1140975627 M * Hollow but, gtg... back in about half an hour
1140975640 M * Bertl k
1140976277 M * `DoM`` howto to limit ram usage? for example 192 mb for each user? and hwo to limit cpu usage ? for example cpu is avaiable for all vservers without limit but when cpu reach 100% only ppl with a particular priority can access to cpu ? 
1140976286 M * `DoM`` hi Bertl :)
1140976787 M * `DoM`` found :d
1140977114 M * `DoM`` mmm it's normal that with top command inside a virtualserver i can read real amount of ram ?i mean i set a limit about 192 and on proc/limit i can see rss limit
1140977194 M * Bertl http://linux-vserver.org/Caps+and+Flags
1140977209 M * Bertl check out the virt_* flags
1140978036 M * Wenix $ vserver start jrc.wenix.dk 
1140978036 M * Wenix Failed to migrate to network context: No such process
1140978037 M * Wenix An error occured while trying to load init for 'jrc.wenix.dk'
1140978146 M * Wenix .. does that refer to the NX_ADDR? - or what may be the problem here..
1140978338 M * bonbons Wenix: something in relation to network context, that is, somehow the context disapeared
1140978383 M * bonbons Don't think it's because of missing NX_ADDR, but would need to be checked
1140978440 Q * mire Quit: Leaving
1140978443 Q * waldi Remote host closed the connection
1140978563 M * bonbons Wenix: what contexts do exist? (ls -ld /proc/virt*/*/)
1140978574 M * Wenix none currently
1140978620 M * Wenix do I need to insert a debug line somewhere, to show the contexts while the script is running?
1140978853 M * bonbons In start.sh, insert a long sleep into the exit handler part, so you have time to check what has been done by hand (or even put an exit there)
1140978993 J * oliwel ~mail-at-o@host-62-245-151-178.customer.m-online.net
1140979041 M * oliwel (hi
1140979043 A * oliwel waves hello to the crowd
1140979120 M * Wenix bonbons: no contexts
1140979184 M * bonbons hm, and how does it come to try to start init if there is not context? Probably better to do the whole process by foot once
1140979194 J * Doener doener@i5387E627.versanet.de
1140979283 M * Wenix *nod*
1140979348 M * bonbons So let's start by hand: vcontext -c -x 1000 -f PERSISTANT
1140979398 M * bonbons vncontext -c -n 1000 -f PERSISTANT
1140979400 M * Wenix bonbons: -C?
1140979423 M * bonbons yep, capital C, for both
1140979430 M * Wenix *nod*
1140979456 M * Wenix done
1140979467 M * Wenix they show up in /proc/virtual
1140979578 M * bonbons now cd ${GUEST_ROOT}, then do vexec -x 1000 -n 1000 -c -- ls /
1140979639 M * bonbons from ${GUEST_ROOT}, also execute vexec -x 1000 -n 1000 -- ps aux (you should just see ps itself)
1140979663 M * Wenix master jrc.wenix.dk # vexec -x 1000 -n 1000 -c -- ls /
1140979664 M * Wenix Failed to migrate to network context: No such process
1140979667 M * Wenix ... on the first one
1140979708 M * Wenix uhm, and now the entry disappeared from /proc/vitual 
1140979709 M * bonbons the contexts still show up in /proc/virt*/?
1140979781 M * Wenix as soon as vexec is run, the contexts disappears
1140979890 M * bonbons retry, and strace vexec, the syscall communication may provide some info
1140979899 M * Wenix ok, hold on
1140979953 M * bonbons before executing vexec, you can "vflags -G -x 1000" and "vnflags -G -n 1000", to make sure flags are set correctly
1140979961 M * Wenix http://pastebin.com/573621
1140979973 M * Wenix oh, didn't see your message about vflags
1140979985 M * Wenix I'll do it once more ;)
1140980042 Q * oliwel Quit: Chatzilla 0.9.69.1 [Firefox 1.5/2005111116]
1140980052 M * Wenix vflags here: http://pastebin.com/573623
1140980132 M * bonbons from the flag's point of view your contexts are fine, they should not be disapearing...
1140980199 M * Wenix .. somehow I am glad to hear that - I've been awfully afraid of wasting your time with something I could have fixed by reading the documentation ;)
1140980377 M * bonbons I'm wondering why your syscalls ( vserver(...) )look so different compared to mine: you have always 0 in the last argument. As I have no 64bit box available I can't check on that arch...
1140980473 M * Wenix Would it help you to have an account on the computer? (not sure its possible, but I can find out if it will help any)
1140980504 M * bonbons Did you recopile libvserver and vserver-utils with dietlibc?
1140980510 M * Wenix no, glibc
1140980518 M * Wenix  - should I recompile with dietlibc?
1140980539 M * bonbons yep, that's better, glibc does not always to the right things!
1140980567 M * bonbons there are issues with syscalls, somehow depending on CFLAGS
1140980721 M * Wenix .. emerge fails with the diet USE flag.. (investigating)
1140980762 M * bonbons I will be back in about half an hour, I will read backlog, so don't hesitate to post!
1140980792 M * Wenix right ;)
1140980798 M * Wenix thanks so far
1140981148 M * Wenix bonbons: http://pastebin.com/573650
1140981163 M * Wenix bonbons: same error, different strace
1140981236 M * Wenix bonbons: above is with dietlibc
1140981553 J * lettuce ~Suh@c-67-168-199-7.hsd1.or.comcast.net
1140981614 M * Bertl welcome lettuce!
1140981625 M * Bertl okay, off for dinner now .. back later ...
1140981630 N * Bertl Bertl_oO
1140981733 Q * tso Quit: BitchX-1.1-final -- just do it.
1140982365 M * bonbons Wenix: syscalls have changed a lot: last two args were "swapped", and the non-zero one "truncated"
1140982411 J * lilalinux ~plasma@80.69.35.186
1140982936 M * Wenix bonbons: yup, I can see it changed alot
1140982960 M * Wenix bonbons: do you have any idea what breaks?
1140983066 M * bonbons no, I don't maybe Hollow, Bertl or daniel_hozac know more about AMD64... at least to check that the syscalls issued are correct
1140983079 M * bonbons for a recall, which kernel patch are you using?
1140983080 J * tso ~tso@249-158.adsl.pool.ew.hu
1140983082 M * tso hi
1140983091 M * cthompson anyone do some sort of NFS server inside a vserver?
1140983099 Q * rs Quit: rs
1140983114 M * daniel_hozac cthompson: the userspace NFS daemon is supposed to work.
1140983128 M * daniel_hozac bonbons: i've yet to touch an x86_64...
1140983154 M * cthompson daniel_hozac: I'll have to try that
1140983154 P * matta 
1140983176 M * Wenix bonbons: 2.6.15.4-vs2.1.1-rc9-gentoo
1140983198 M * Wenix bonbons: I can check exactly which patches are applied if needed?
1140983220 M * bonbons that info is sufficient
1140983768 Q * lettuce Quit: NO CARRIER
1140983819 M * bonbons Wenix: Ok, probably no need to try more, vexec in 1.0.3 makes the contexts non-persistant before use. Take the 1.0.4 branch from SVN
1140983835 M * bonbons eventually wait a bit, so I can test my pending changes before commit
1140984104 M * Wenix bonbons: oki, how long should I wait before taking 1.0.4 from SVN? (minutes?, hours?, days? weeks?)
1140984138 M * bonbons half an hour, time to compile and test a bit
1140984156 M * Wenix alright, I'll wait a bit
1140984212 M * Wenix if you can drop a line here when you committed your changes, I'd be happy to test it
1140984310 M * bonbons will do so
1140984893 M * Hollow bonbons: pong.. which syscall?
1140984975 M * bonbons Hollow: on AMD64: http://pastebin.com/573650 (dietc), http://pastebin.com/573623 (glibc), but I guess it's more a problem with 1.0.3 being quite broken (on dietc)
1140985021 M * Hollow well, the list implementation in 1.0.* is horrible
1140985059 M * Hollow hm.. maybe i should read till the end
1140985060 M * Hollow :P
1140985128 M * Hollow hm..
1140985245 M * bonbons in vexec, the PERSISTANT flag was being removed before migrating -> context went away just before vexec should enter it!
1140985285 M * Hollow heh
1140985289 M * Hollow broken logic
1140986879 M * bonbons Wenix: Ok, works sufficiently now, will commit very soon (start/stop/proc(un)hide)
1140987136 M * bonbons Wenix: commited to branches/1.0.4
1140987476 M * daniel_hozac shouldn't that be branches/1.0 and tags/1.0.4? :)
1140987490 M * Wenix bonbons: Ok, I'll give it a try.. (it may take a little while, since I will try to make an ebuild file for it)
1140987557 M * bonbons reusing the one for 1.0.3 + some small changes should be sufficient: depends: +libowfat, +dietlibc, need to save the init scrpit from a previous version
1140987582 M * bonbons and remove the diet from use flags
1140987888 M * Wenix ok, I'll try.. it would be a second ebuild file ever.. (first one was trying to make the 1.0.3 use the patch you mentioned earlier)
1140987907 M * Wenix but so far, I am still looking for the svn address ;)
1140987953 M * bonbons Wenix: That's the easiest part :) http://dev.croup.de/repos/vserver-utils/branches/1.0.4
1140987964 M * Wenix thanks ;)
1140988325 J * Aiken ~james@tooax6-178.dialup.optusnet.com.au
1140988945 Q * fwl Quit: This computer has gone to sleep
1140989136 M * Wenix bonbons: vexec is now a success
1140989237 M * bonbons Fine :) maybe it starts a bit to verbose, but better to verbose than not at all ;)
1140989274 M * bonbons Wenix: it you have time, play around and tell any issues you encounter
1140989317 M * Wenix bonbons: I'll do that - first one is that it already think the vserver is running - but I think I can manage to figure that out myself
1140989383 Q * Viper0482 Remote host closed the connection
1140990673 J * fwl ~fwl@83.215.237.1
1140991360 M * Wenix master etc # /etc/init.d/vservers start
1140991361 M * Wenix  * Cannot find util-vserver installation
1140991361 M * Wenix  * (the file '/usr/lib/util-vserver/util-vserver-vars' would be expected)
1140991368 M * Wenix .. what should this file contain?
1140991448 M * bonbons  /etc/init.d/vserver should mainly contain a call to "vserver procfs unhide" in start, and "vserver procfs hide" in stop
1140991535 M * bonbons Wenix: then you probably want to add an initscript to start/stop each individual guest (or group of guests)
1140991710 M * Wenix thanks, I'll try
1140991838 M * Wenix hehe, its takes time to get used to write "start <server>" instead of "<server> start"
1140991976 Q * fwl Quit: This computer has gone to sleep
1140992034 M * bonbons Wenix: hehe, that's not easy :)
1140992123 M * Wenix is it possible for iptables to mark packets based on its context?
1140992149 M * bonbons I don't know
1140992288 M * Wenix well, its not important yet.. but in the future I need to do some traffic accounting on each vserver, and some may share the same IP
1140992316 M * Wenix currently all vservers has each own IP, so accounting by IP is how I do it now
1140992361 M * Wenix do you know how large the userbase of the vserver project is?
1140992365 N * Bertl_oO Bertl
1140992367 M * Bertl back now
1140992404 M * bonbons Wenix: That would be a question for Bertl I think... and if it's possible (e.g. info theoretically available), iptables must still be able to handle it
1140992478 M * Bertl hmm?
1140992496 M * derjohn Bertl, <Wenix> is it possible for iptables to mark packets based on its context?
1140992503 M * derjohn good question ;)
1140992515 M * Bertl ah, well, we have code for that
1140992524 M * Bertl but it's currently not used ...
1140992538 M * Bertl usually it's sufficient to tag by IP
1140992543 M * derjohn .. and will be obsoleted with ngnet ?
1140992573 M * Bertl hmm, not sure .. might be interesting as future addition for non-ngnet netwroking
1140992576 M * Bertl *networking
1140992590 M * derjohn context -> fwmark ?
1140992595 M * Bertl yup
1140992611 M * Bertl or as we already had, a separate marking
1140992619 M * Bertl you just need a patched iptables for that
1140992646 M * Bertl would even allow to make rules for incoming traffic
1140992659 M * derjohn well, I only makes sense with a single IP and multipe RFC1918 guests .... or do I miss the point?
1140992664 M * Bertl basically disallowing socket delivery when the network id doesn't match
1140992691 M * Wenix Bertl: Another workaround now it to assign dummy interfaces to the vservers, and then running them all through a local NAT gateway
1140992715 M * Bertl hum, dummy interface? nat gateway?
1140992736 M * Bertl derjohn: no, there IMHO it's not even required
1140992748 M * Bertl derjohn: more for guests which 'share' an IP
1140992780 M * derjohn can two guests have the same IP ? 
1140992789 M * Bertl yup
1140992797 M * daniel_hozac theoretically, guests can even shared the network context :)
1140992810 M * daniel_hozac s/shared/share/
1140992811 M * derjohn ok ok ... I meant: and the work ?
1140992815 M * derjohn *they
1140992834 M * Wenix they work now
1140992844 M * derjohn hm, so it depends on the port they bind?
1140992845 M * Wenix I have a few vservers sharing IP's now
1140992850 M * Bertl yes, as long as they do not try to bind to the same ports
1140992857 M * Wenix *nod*
1140992859 M * Bertl think 'apache guest', 'postgres gruest' ...
1140992866 A * derjohn thinks: damn stuff
1140992893 M * Bertl but I think we will extend the non legacy networking in the future
1140992913 M * Bertl i.e. the non ngnet networking which is based on sharing the stack
1140992922 M * derjohn Bertl, so: no hope for ngnet in the near1^W middle future ?
1140992956 M * Bertl oh, yes, ebiederm is working on mainline virtualization, and the network folks are working on something similar
1140993015 M * Bertl but I do not intend to implement ngnet right now unless:
1140993036 M * Bertl  a) somebody sponsors the development
1140993053 M * Bertl  b) ebiederm/mainline decides not to go in that direction
1140993080 M * Bertl  c) a really excellent reason for doing that pops up 
1140993117 M * derjohn When it comes to mainline: How much is Sam involved there? Does he differne4t stuff than eric ? The IBM folks?
1140993135 M * Bertl well, it's complicated, as usual ...
1140993154 M * Bertl - the IBM folks want virtualization in mainline
1140993170 M * Bertl - the SWsoft folks want _their_ virtualization in mainline
1140993190 M * Bertl - Sam is preparing Linux-VServer for mainline inclusion
1140993207 M * Wonka who's swsoft?
1140993225 M * derjohn and UML will make the race? :/
1140993227 M * Bertl - Eric is looking for a generic virtualization approach which should get into mainline to make is company happy (which is interested in migration)
1140993257 M * Wonka derjohn: UML is a separate kernel - eats too much memory, imo
1140993260 M * Bertl derjohn: luckily UML is already in mainline :)
1140993275 M * Bertl derjohn: and Xen (as arch) will hopefully soon be
1140993282 M * tso Wonka they develop virtuozzo and plesk
1140993289 M * derjohn Bertl, ok .. luckily .. Wonka: I should have set an </irony> 
1140993296 M * Wonka :)
1140993320 M * Wonka virtuozzo... that's the openvz stuff, isn't it?
1140993332 M * Bertl ah, and I forgot: Bertl is trying to ensure that mainline virtualization will not become too bloated :)
1140993363 M * Bertl Wonka: yup, Virtuozzo(tm) is the commercial product they care about
1140993377 M * Wonka i liked the chcontext stuff, which enables one to just run eg chcontext apache, instead of starting a whole vserver... which would be needed with openvz.
1140993424 M * Bertl that's why I like Eric's *space concept as it is now .. it is very similar to what Linux-VServer does (regarding breaking down functionality)
1140993863 M * Wenix aiieee.. the new vserver took down my eth0 interface :-/
1140993899 M * Wenix oh well, I better plan a trip to the hosting center soon then - and fix the error that made it possible
1140993933 M * derjohn  Wenix , yes. I am sorry but I know that feature too well ...
1140993947 M * derjohn if you are they enable the other behaviour 
1140993968 M * derjohn "alias promotion"
1140993978 M * derjohn Wenix, ...by adding "net.ipv4.conf.all.promote_secondaries = 0" in /etc/sysctl.conf or via sysctl comm...
1140993987 M * Wenix derjohn: I think I may have a script that checks for the gateway once in a while, and if unavailable reboots the computer
1140994008 M * derjohn Wenix, that cool if you ISP reboot its router !
1140994019 M * Wenix derjohn: hopefully this scripts still runs
1140994028 M * Wenix derjohn: well, it pings it for an hour
1140994033 M * derjohn ah .... i press thumbs  for you
1140994044 M * Wenix derjohn: and if a single ping works, then no reboot
1140994069 M * Wenix derjohn: so, hopefully it will be up in like 5-6 hours ;)
1140994070 M * derjohn Wenix, Red October :) ?
1140994079 M * Wenix derjohn: sleep ;)
1140994081 M * derjohn 5-6h ? fsck ?
1140994096 M * Wenix no, the scripts only runs a few times a day
1140994104 M * Wenix (if at all)
1140994114 M * derjohn ok :)
1140994122 M * derjohn no remote hands ?
1140994147 M * Wenix well, perhaps.. I don't want to pay the price ;) - and its not that important
1140994164 M * Wenix its not running any critical services..
1140994231 M * Wenix my own svn repository and mail is probably what I will miss the most :)
1140994278 M * Wenix well, I'd better catch some sleep, and check again tommorrow
1140994286 M * derjohn Wenix, if your code was opensource you could get it from some mirroe now :)
1140994313 M * Wenix derjohn: I am not allowed to make it opensource... (yet)
1140994321 M * derjohn k#
1140994352 M * Wenix derjohn: I work for the danish national police creating software to help their investigations... they prefer not to tell about their methods
1140994383 M * derjohn Wenix, thats sad, so I cannot create any counter-measures ;)
1140994413 M * Wenix derjohn: hehe
1140994442 M * Wenix derjohn: I hope you won't need counter measures for what i am making.. (childpornography filters and scanners)
1140994467 M * derjohn Wenix, I touched systems of the German authorities, but not necessarily codeing, more implementation.
1140994496 M * derjohn Wenix, classifiers?
1140994568 M * Wenix derjohn: no, I wouldn't call it classifiers
1140994576 J * stuart ~stuart@203-206-78-108.dyn.iinet.net.au
1140994581 M * derjohn Wenix, but? Watermarks???????
1140994613 M * Wenix derjohn: no, watermarks is only interresting if we distribute the files.. and I hope we dont ;)
1140994631 M * Wenix derjohn:  ah
1140994645 M * derjohn Wenix, the spanish cops did IMO. The was an article about that soem days ago.
1140994665 M * Wenix derjohn: watermarks made by the manufactores 
1140994673 M * derjohn k
1140994722 M * cthompson it is possible to run a fedora vserver on a gentoo host, right?
1140994735 M * derjohn cthompson, it should, yes.
1140994762 M * cthompson so, how the heck do I create a fedora guest image without having a fedora box?
1140994787 M * Wenix derjohn: sorry, I think I got you confused..  - In my department we do not work much with watermarks (not our own software anyway), but more about locating illegal content on different networks and a way to handle the content to make it easy to find connections
1140994824 M * daniel_hozac cthompson: tarballs, or install yum etc.
1140994828 M * derjohn cthompson, I dunno gentoo. But in Debian you would use "vserver build foo ...." (ckeck out the wiki)
1140994849 M * Wenix derjohn: but the area is still a bit new to me ;)
1140994868 M * derjohn ok, too me, too ! :)
1140994881 M * cthompson there should be a repository of minimal images of dists
1140994907 M * daniel_hozac cthompson: make one ;)
1140994908 M * Wenix derjohn: and Linux is very new to them (which is why I run svn on my own personal server an dnot theirs)
1140994908 M * Bertl welcome stuart!
1140994934 M * cthompson daniel_hozac: so, "patches welcome" :)
1140995042 M * stuart ello Bertl 
1140995045 M * stuart do I know you? :)
1140995048 N * stuart PerlLaptop
1140995065 M * Bertl PerlLaptop: don't know, maybe :)
1140995090 M * derjohn cthompson, there are fedora images around ... maybe not the newest FC ... -> wiki ?
1140995103 M * cthompson yeah, I'm on the wiki now
1140995132 Q * Wenix Quit: Keep up the good work :
1140995143 M * cthompson http://www.virtualinfrastructure.nl/blog/index.php?op=ViewArticle&articleId=6&blogId=1
1140995147 M * cthompson cool
1140995528 Q * tso Quit: [BX] Who ate my nuggets?!
1140995611 M * `DoM`` Bertl, there is possibility to download all manual about linux-vserver in a single html webpage ?
1140995648 M * derjohn `DoM`, the complete wiki? wget ? 
1140995696 M * `DoM`` i would like to have all in one webpage
1140995712 M * `DoM`` just to know if exists or not
1140995728 M * derjohn `DoM`, puuuh, there is no such thing as "the documentation". The wiki is the best source and there are man pages.
1140995733 M * `DoM`` better httrack then wget for this case :P
1140995799 M * derjohn `DoM`, you are looking for a manual like a pinted multi-page foo?
1140995842 M * derjohn `DoM`, vserers develops so much that we couldn't stay even close to the reality ... (at least with manpower we have here)
1140995870 M * derjohn `DoM`, the wiki is you source.
1140995876 M * derjohn *our
1140996142 Q * Doener Ping timeout: 480 seconds
1140996501 M * Bertl `DoM``: if you like to start a documentation project (i.e building ma/info pages or so) be my guest ...
1140996689 J * rs ~rs@vol75-7-82-229-177-124.fbx.proxad.net
1140996706 Q * derjohn Quit: by(t)e
1140996727 M * Bertl welcome rs!
1140996734 M * rs hi
1140996930 J * ntrs ~ntrs@68-188-37-15.dhcp.stls.mo.charter.com
1140996976 J * Doener doener@i5387D510.versanet.de
1140998253 N * ebiederm ebiederm_oO