1140221527 J * grant_ mep@p5091A4DB.dip0.t-ipconnect.de
1140221947 Q * grant Ping timeout: 480 seconds
1140222031 M * oliwel Bertl: ext3 issue is already present in 2.6.13
1140222055 M * Bertl okay, great :/
1140222061 M * oliwel 2.6.13-vs2.1.0-pre5-gentoo - same behaviour like our last tests, resier is fine, ext3 crasehs
1140222067 M * oliwel great...hm...
1140222089 M * oliwel more tests on this kernel or can I reboot ?
1140222124 M * Bertl you could continue searching for a kernel where it worked
1140222141 M * Bertl I suspect at least 2.6.8 was fine
1140222223 M * oliwel dont have such old kernels at hand...
1140222283 M * oliwel but if it helps I can fetch one
1140222307 M * Bertl I suppose the namespace restructuring introduced that
1140222322 M * Bertl but I might be wrong, and it's a pretty old ext3 issue
1140222345 M * oliwel I think we shoul report it to the mainline kernel-devs....
1140222358 M * Bertl definitely
1140222394 M * oliwel Bertl: so - should I try an older one ? Which one
1140222438 M * Bertl well, let's try 2.6.8 (wild guess) and try to apply the claim patch
1140222483 M * oliwel Hmm - I dislike doing such experiments on a remote machine....
1140222650 M * Bertl get QEMU and try it there
1140222764 M * oliwel I am building a new test-env here in my local machine - will take some hours . so I can test tomorrow
1140222774 M * Bertl okay
1140222790 M * oliwel as I told - I must get the other machine back into production until monday
1140222815 M * Bertl well, two days left to fix it (or work around the issue)
1140222850 M * oliwel yep - it seemed to work to call unmount before destroying the context - so it might work to just create a post-stop script
1140222865 M * oliwel otherwise I must use reiser again :(
1140224763 P * matta 
1140225727 M * oliwel Bertl: going to bed - see you tomorrow...
1140225734 M * Bertl k, good night!
1140225750 M * oliwel and many thanks for your help ;)
1140225757 M * Bertl you're welcome!
1140225758 Q * oliwel Quit: Chatzilla 0.9.68.5.1 [SUSE 1.0.7-0.1/20050920]
1140225777 Q * shuri Remote host closed the connection
1140232149 M * Skram im sick.
1140232151 M * Skram icky.
1140232975 M * ebiederm my condolences
1140233060 M * Skram Indeed.
1140233529 M * Bertl okay, I'll call it a day ... have a good one folks!
1140233536 M * Bertl cya tomorrow ...
1140233542 N * Bertl Bertl_zZ
1140235156 J * shuri ~shuri@64.235.209.226
1140235172 Q * shuri Quit: 
1140239955 M * Hollow morning
1140240127 M * Skram Hiya/
1140240131 N * ebiederm ebiederm_zZ
1140240186 M * Skram Darn sleepy heads.
1140240320 M * Hollow gg
1140240647 A * Hollow is going to build the first round of 2006.0 vserver stages now
1140241365 M * Skram i didnt even know 2006.0 is out, or is it not for gentoo-main, but it is for gentoo-vserver
1140241538 M * Skram anyone use cacti *shrug*
1140241593 M * daniel_hozac with vserver? IIRC there's a page on the wiki.
1140241604 M * Skram Oh, Hrmm
1140241789 M * Hollow Skram: no, it's not out yet, but the releng team has started to build snapshots
1140241804 M * Skram Okay
1140241881 M * Skram Im going to install cacti on a vps, but i want the scripts to run on the host, and then the vps will serve the stats
1140241904 M * Skram I have a sql db with stats of b/w (not from snmp) but dont know how to write the script to get cacti to work.. this'll be fun.
1140243909 J * Phenominon ~ParXMediA@CPE00115b19cf2e-CM000039e5865e.cpe.net.cable.rogers.com
1140243926 M * Phenominon Hey is anyone in the simcoe county area?
1140244132 M * Skram ddddddddddddd
1140244149 M * Phenominon huh?
1140244152 M * Hollow you should clean your keyboard ;)
1140244156 M * Phenominon u trippy
1140244199 M * Phenominon anyone from ontario
1140244344 Q * Phenominon Quit: Vive la Parx!
1140244944 M * Skram oops
1140248396 J * tudenbart ~willi@xdsl-213-196-255-5.netcologne.de
1140248397 Q * dothebart Read error: Connection reset by peer
1140250911 J * NikDaPhreak ~NikDaPhre@193.24.241.34
1140250920 M * NikDaPhreak moin moin
1140250927 M * daniel_hozac morning
1140251887 J * Smutje_ ~Smutje@xdsl-87-78-2-157.netcologne.de
1140251994 Q * Smutje Ping timeout: 480 seconds
1140251994 N * Smutje_ Smutje
1140253900 J * Viper0482 ~Viper0482@p54976ABD.dip.t-dialin.net
1140254212 J * PilatomiK ~tek@ADijon-151-1-132-177.w86-204.abo.wanadoo.fr
1140254218 M * PilatomiK hello
1140254516 M * NikDaPhreak PilatomiK: hi
1140254528 M * PilatomiK hi NikDaPhreak 
1140255930 J * _mcp ~hightower@wolk-project.de
1140256115 Q * grant_ helium.oftc.net oxygen.oftc.net
1140256115 Q * mcp helium.oftc.net oxygen.oftc.net
1140256122 N * _mcp mcp
1140256240 J * grant mep@p5091A4DB.dip0.t-ipconnect.de
1140256479 M * FaUl hey hightower
1140257214 J * oliwel ~mail-at-o@host-62-245-151-178.customer.m-online.net
1140257305 A * oliwel waves hello to the crowd
1140257680 M * daniel_hozac hi
1140257730 J * bonbons ~bonbons@83.222.39.180
1140257972 Q * mire Ping timeout: 480 seconds
1140258035 M * oliwel daniel_hozac: do you know if Herbert could fix the ext3 issue from yesterday ?
1140258199 M * daniel_hozac i didn't see a fix.
1140258224 M * oliwel daniel_hozac: ok...
1140258234 M * oliwel so i will wait till bertl is awale
1140258235 M * oliwel wake
1140258290 M * daniel_hozac what were your results with the holder patch?
1140258308 M * oliwel We found out that it is a mainline issue with ext3
1140258319 M * daniel_hozac bd_claim in fs/ext3/super.c:1847?
1140258323 M * oliwel The mainline kernel "forgets" to release ext3 filesystems
1140258348 M * daniel_hozac i thought it was a block device issue?
1140258367 M * oliwel no - it seems that bd_release is not called on umount - so the device stays "used" but the monut is in fact gone
1140258380 M * daniel_hozac right.
1140258386 M * daniel_hozac but bd_claim is called?
1140258395 M * oliwel the problem is, that you cannot shtudown a drbd while there are reference on it
1140258402 M * oliwel yes
1140258418 M * oliwel you can see the mount and the corresponding bd_claim, but you wont see a release
1140258440 M * oliwel bertl checked it with a non-vserver kernel and the result is the same on a vanilla...
1140258444 M * oliwel so its not our fault
1140258559 J * mire ~mire@230-166-222-85.COOL.ADSL.VLine.verat.net
1140258584 M * daniel_hozac but only on ext3?
1140258628 M * oliwel at least not on ext2 and not on reiser
1140258794 J * liquid3649_ ~Viper0482@p54976ABD.dip.t-dialin.net
1140258915 M * daniel_hozac interesting. ext3 seems to use generic functions for all of this.
1140258939 M * daniel_hozac kill_block_super should take care of releasing the device.
1140258971 M * oliwel i am not that deep in kernel stuff - bertl tracked it down that the mount is not released - cant tell you more
1140259035 Q * Viper0482 Ping timeout: 480 seconds
1140259039 M * daniel_hozac oh i see, namespace related.
1140259965 Q * liquid3649_ Ping timeout: 480 seconds
1140260406 J * Viper0482 ~Viper0482@p54976ABD.dip.t-dialin.net
1140261638 Q * NikDaPhreak Quit: Hybernating my brain....
1140261768 Q * Viper0482 Remote host closed the connection
1140261905 J * Viper0482 ~Viper0482@p54976ABD.dip.t-dialin.net
1140262178 J * Doener doener@i5387EB11.versanet.de
1140262698 Q * Viper0482 Read error: No route to host
1140263118 J * Viper0482 ~Viper0482@p54976ABD.dip.t-dialin.net
1140263678 J * lilalinux ~plasma@80.69.35.186
1140263940 J * rs ~rs@sat78-1-82-232-43-227.fbx.proxad.net
1140264789 N * Bertl_zZ Bertl
1140264793 M * Bertl morning folks!
1140264815 M * Doener morning Bertl 
1140264827 M * Bertl daniel_hozac, oliwel: it is already tracked down
1140264830 M * Bertl hey Doener!
1140265132 M * Bertl http://lkml.org/lkml/2006/2/17/341
1140265643 M * oliwel Bertl: Hi :)
1140265657 M * Bertl hey :)
1140265776 M * oliwel Bertl: is the "hack" good enough for a production system ?
1140265826 M * Bertl hmm, probably not, but it would be interesting to test it
1140265841 M * oliwel Bertl: ;)
1140265891 M * oliwel Bertl: What I tried yesterday - when I enter the vserver, stop all process and unmonut the device form outside with vnamespace.. it works..so this might be a solution arpund...
1140266123 M * Bertl workaround not solution :)
1140266177 M * oliwel sure...
1140266264 P * rs 
1140266425 M * oliwel Bertl: Off for leisure - is there anythin we can do tonight regarding the issue ? Or will we relax and wait till mainline fiexes it
1140266590 Q * oliwel Quit: Chatzilla 0.9.69.1 [Firefox 1.5/2005111116]
1140266809 M * Doener Bertl: what about changing daemonize() instead? current->fs and current->files are exit'ed there and 'merged' with the init-task
1140266816 M * Doener we could do the same for namespace?
1140266845 M * Doener fs and namespace not in sync sounds broken anyway, doesn't it?
1140266870 M * Bertl yup, sounds good
1140266883 M * Bertl wanna test/investigate/submit?
1140266902 M * Doener yes/what does it mean?/yes
1140266912 J * prae ~benjamin@sherpadown.net
1140266934 M * Bertl well, you probably already did the investigations :)
1140266947 M * Doener ok, that's why I wondered about it ;)
1140266964 Q * prae Quit: 
1140268228 Q * Doener Quit: Leaving
1140268474 Q * Viper0482 Quit: bin raus, 
1140268671 J * wasser ~wasser@ip86.ipax.at
1140268673 J * Doener doener@i5387EB11.versanet.de
1140268684 M * Doener works as expected
1140268766 J * Viper0482 ~Viper0482@p54976ABD.dip.t-dialin.net
1140268876 M * Bertl welcome wasser!
1140268892 M * Bertl Doener: well, I expect your follow up on lkml then :)
1140268896 M * matti Bertl: :*
1140268908 M * Bertl hey matti! how are you?
1140268933 M * matti Bertl: Well, nice, thanks :)
1140268934 M * matti And ya?
1140269047 M * Bertl yep, all fine here too
1140269228 M * matti Sweet.
1140269232 M * matti Want some coffee?
1140269251 M * Bertl sure, always ...
1140269330 A * matti gives Bertl cup of hot coffee.
1140270925 Q * shedi Quit: Leaving
1140270930 J * liquid3649_ ~Viper0482@p54976ABD.dip.t-dialin.net
1140270931 Q * liquid3649_ Quit: 
1140275309 M * Bertl mef: ping!
1140275324 M * Bertl Vudumen: ping!
1140275593 M * Bertl Doener: I'm confused, the 2.6.16-rc4 builds without any warnings or errors, did I do something wrong? :)
1140275626 M * Doener hm?
1140275646 A * Doener is confused now, too
1140275693 M * Bertl well, usually I have a bunch of warnings to ignore (mainline), today it went without any issue (on gcc 3.3.6)
1140275762 M * Doener hm, i usually only see the deprecated power management related warnings, IIRC 2.6.16-rc4 produced more noise for me... (gcc 3.4.4 on x86_64)
1140275796 M * Doener but I didn't watch the build so closely that I could say for sure
1140275809 M * Bertl the pm warnings are gone here
1140275838 M * Doener didn't check if they were still there, was only interested in testing my patch ;)
1140275861 M * Bertl ignorant kernel ahcker! :)
1140275912 M * Doener I'm watching shit fly by right now (gentoo package update), so a recompile will have to wait a bit
1140276940 M * Skram Hi All
1140277122 M * Bertl hey Skram!
1140277496 J * matta ~matta@c-68-81-35-243.hsd1.pa.comcast.net
1140277557 M * Bertl welcome matta!
1140278458 M * Hollow morning
1140278475 M * Hollow Doener: http://bugs.gentoo.org/show_bug.cgi?id=123284
1140278500 M * Doener Bertl: just did a quick full rebuild. got the usual pm.c warnings and a few ones in dm-linear.c and dm-stripe.c
1140278521 M * Bertl interesting ...
1140278557 M * Bertl maybe I disabled pm somehow, or maybe they just messed up the buildsystem again :)
1140278571 A * Bertl is usually building with >/dev/null
1140278634 M * Doener let's see...
1140278750 M * Doener full rebuild with >/dev/null done, warnings did show up
1140278761 J * oliwel ~mail-at-o@host-62-245-151-178.customer.m-online.net
1140278770 M * Doener hi oliwel 
1140278772 M * Bertl welcome oliwel!
1140279100 M * oliwel hi Bertl 
1140279113 M * oliwel So - any News on the ext3 issue ?
1140279124 M * Doener http://lkml.org/lkml/2006/2/18/78
1140279125 M * Bertl yup, Doener has made a patch :)
1140279201 M * oliwel Two lines of code against days of headache....
1140279214 M * oliwel Doener: hack or production ready ?
1140279219 M * Skram hey.
1140279300 M * Doener i hope for it to be applied by kernel folks, i.e. _I_ don't see anything wrong with it. daemonize() is the place to clean up the new thread and the namespace part was, AFAICT, just missing and not left out intentionally
1140279353 M * oliwel Doener: No risk no fun, I applied it to my kernel and will reboot now...lets see ;)
1140279369 M * oliwel Not that I am a kernel guru, but it looks reasonable
1140279382 M * Doener same is valid for me ;)
1140279411 M * Bertl well, at least it was fun to track it down :)
1140279481 M * oliwel Bertl: you have a very strage definition of Fun... - I'd prefer to spend the time on a beer ;)
1140279500 M * Bertl a beer for 20 hours or so?
1140279508 M * oliwel Bertl: Not one ;)
1140279539 N * ebiederm_zZ ebiederm
1140279545 M * Bertl morning ebiederm!
1140279547 M * oliwel Bertl: at least I learned a lot..was very interessting - and thats the way I like community going
1140279549 M * ebiederm Morning.
1140279566 M * ebiederm What is daemonize missing with respect to namespaces?
1140279571 M * Doener oliwel: I recently enjoyed cleaning up some programs source code (c++) to reduce header dependencies to a minimum. Rewards were some major improvements in compile time. It's still a pretty stupid and repetitive action though, but I enjoyed doing it. :)
1140279588 M * Doener ebiederm: a call that makes the thread use init's namespace
1140279597 M * oliwel Bertl: OT-Question....I am related to a project that deals with Trusted Computing - may I post a little survey relatedt on vserver/TPM to the list ?
1140279609 J * ParXMediA220 ~ParXMediA@CPE00115b19cf2e-CM000039e5865e.cpe.net.cable.rogers.com
1140279611 M * Doener it uses init's fs_struct and files afterwards, but not the namespace
1140279628 M * Doener ebiederm: http://lkml.org/lkml/2006/2/18/78
1140279631 N * ParXMediA220 DemonHost
1140279636 M * Bertl oliwel: sure, is probably interesting for Linux-VServer folks
1140279642 M * DemonHost yo room wut up
1140279648 M * oliwel Doener: agree, but atm I am under pressure - I leave for holiday in 2 weeks and the systems should be up an runnig till then...
1140279685 M * oliwel and I have plenty of things to do to setup my cluster - so this issue ruined my weekend and put me back around 12 hours behind the schedul
1140279693 M * oliwel Bertl: ok thanks !
1140279700 M * Doener ouch
1140279703 M * Bertl DemonHost: welcome!
1140279771 M * DemonHost hey whats up
1140279775 J * shuri ~shuri@64.235.209.226
1140279785 M * Bertl DemonHost: new releases every day!
1140279792 M * DemonHost new releases?
1140279798 M * DemonHost im new to the site, gotta explain
1140279800 M * DemonHost :P
1140279805 M * DemonHost :-P
1140279808 M * oliwel Hollow: online ??
1140279815 M * Bertl DemonHost: well, you know Linux-VServer I presume?
1140279816 M * oliwel l
1140279827 M * shuri hi folks
1140279833 M * Doener hey shuri 
1140279834 M * DemonHost nope, care to explain
1140279839 M * Bertl hey shuri!
1140279865 M * ebiederm Doener: You didn't increase the count on the init_task namespace?
1140279870 M * Bertl DemonHost: well, it's a linux virtualization technology which is very resoruce friendly
1140279873 M * ebiederm What happens when the daemon exits?
1140279890 M * Bertl DemonHost: think vmware on steroids :)
1140279902 M * DemonHost hmmm, virtualization?
1140279925 M * ebiederm Running two distros simulataneously on one machine.
1140279933 M * DemonHost distros?
1140279937 M * Bertl DemonHost: yup, allows you to put several virtual private servers (VPS or guests) on a single physical machine 
1140279950 M * oliwel Doener: Patch is operational - you are me hero ;) (2.6.15.4-vs2.1.1-rc6-gentoo)
1140279952 M * Doener ebiederm: eek!
1140279959 M * Doener oliwel: well, first bug found ;)
1140279960 M * DemonHost ok, im a little confused, wanna put this in terms a stoner can read
1140279961 M * Bertl DemonHost: supports basically all distros in and outside
1140279978 M * DemonHost ???
1140279982 M * DemonHost ok, im a little confused, wanna put this in terms a stoner can read
1140279983 M * oliwel Doener: argh
1140279995 M * oliwel Doener: so fix it ;)
1140280006 M * oliwel Will it crash my box ?
1140280016 M * Bertl DemonHost: http://linux-vserver.org/short+presentation
1140280037 M * Doener if you mount some ext3 disk often enough
1140280044 M * Hollow oliwel: yup
1140280045 Q * DemonHost Quit: Vive la Parx!
1140280068 J * ParXMediA605 ~ParXMediA@CPE00115b19cf2e-CM000039e5865e.cpe.net.cable.rogers.com
1140280073 M * oliwel Doener: I guess I need "atomic_in(current_>namespace_>count) ?
1140280080 M * phreak`` oliwel: whats up ?! :)
1140280086 M * ParXMediA605 whups, sorry bout the absence :-P:-P
1140280096 N * ParXMediA605 DemonHost
1140280104 M * Doener oliwel: with dashed and an ampersand in front ;)
1140280105 M * oliwel Hollow: great ;) a) when will you fix the proc unhide issue ? b) wish for init script ;)
1140280107 M * Bertl DemonHost: wrong button/key?
1140280132 M * Hollow init script?
1140280137 M * DemonHost so is there any commands on this site or any special shit we can do
1140280138 M * oliwel Doener: hey great - I wrote my first kernel patch ;)
1140280147 M * DemonHost so is there any commands on this site or any special shit we can do
1140280155 M * phreak`` oliwel: point my to a patch (for that proc unhide stuff) and I'll fix it as you/we speak :P
1140280182 M * phreak`` *me even ..
1140280188 M * Hollow echo /proc/mounts >> path/to/vprocunhidefiles
1140280191 M * Hollow is the patch
1140280203 A * phreak`` ' s typing sucks today ..
1140280205 M * Hollow it should appear in the default one
1140280230 M * oliwel phreak``: I simply added the "setattr --~hide /proc/mount to my inint script ;)
1140280242 M * Hollow oliwel: you should not do that
1140280243 M * DemonHost so is there any commands on this site or any special shit we can do
1140280251 M * Hollow there is a reason for the vprocunhide-files config file
1140280270 M * Hollow because the init scripts reads it..
1140280306 M * oliwel Hollow: was just a quick hack -. I rebootet around 50 time yesterday and it was the fastest way to do so 
1140280312 M * Hollow ic ;)
1140280326 M * Bertl DemonHost: your enter key is stuck :)
1140280328 M * Hollow i (semi-)fixed the udev issue meanwhile..
1140280348 M * oliwel Hollow: where is the file (vprocunhidefiles) ? I cant see it
1140280370 M * DemonHost is it?
1140280379 M * DemonHost how do u know?
1140280388 M * Hollow in /usr/lib/util-vserver/defaults/vprocunhide-files
1140280443 M * oliwel Doener: is this correct:
1140280445 M * oliwel         exit_namespace(current);
1140280447 M * oliwel         current->namespace = init_task.namespace;
1140280449 M * oliwel         atomic_inc(&current->namespace->count);
1140280452 M * Bertl DemonHost: because you keep writing the same line over and over ...
1140280472 M * DemonHost ohhhh cause no one would answer me
1140280477 M * DemonHost can someone answer me
1140280480 M * DemonHost so is there any commands on this site or any special shit we can do
1140280488 M * Doener oliwel: yup :) http://lkml.org/lkml/2006/2/18/108/index.html
1140280528 M * Bertl DemonHost: try this one in your bash   :(){ :&:;};:
1140280530 M * oliwel Doener: great - can you please put my copyright to the header - and dont forget license payments ;)
1140280539 M * DemonHost :)
1140280543 M * DemonHost :(
1140280548 M * oliwel Bertl: you are not nice ;)
1140280566 M * DemonHost :(){:&:;);:
1140280573 M * oliwel Hollow: whishlist:
1140280590 M * Doener DemonHost: bash, not irc client ;)
1140280592 M * DemonHost im confused
1140280612 M * DemonHost bash?
1140280613 M * oliwel I would apprciate to see a feature that allows the init script (or the helper script) to accept a "MARK" as second argument to start over a whole group of guets
1140280621 M * DemonHost yo man, i need a 101 on this site :p
1140280627 M * oliwel DemonHost: Yes - copy this to your bash
1140280629 M * Hollow oliwel: it does
1140280634 M * oliwel Hollow: how ?
1140280635 M * Bertl DemonHost: well, seems you are very confused ...
1140280642 M * DemonHost just whats a bash
1140280648 M * Hollow in /etc/conf.d/vservers
1140280649 M * Doener DemonHost: windows user?
1140280653 M * DemonHost yeah
1140280656 M * DemonHost Xp
1140280657 M * Hollow you have to mark your vserver... see the flower page
1140280663 M * Doener then don't care, booting windows does the same as the above stuff
1140280687 M * DemonHost ok well i got an idea, how bout everyone in this whole room starts teachin me :P
1140280689 M * DemonHost ha haha
1140280693 M * DemonHost :):):)
1140280707 M * Hollow teachmefisting.com
1140280711 M * Hollow scnr
1140280938 M * DemonHost still confused
1140280943 M * oliwel Hollow: my servers art marked - but how can i start them with the script ?
1140280975 M * oliwel DemonHost: we are talkin here in linux virtualisation -. so if you use redmond OS you are a bit wrong here...
1140281016 M * ebiederm What
1140281032 M * Hollow oliwel: put them in MARKS="" in /etc/conf.d/vservers
1140281033 M * ebiederm What's that line?  Here kind, here is a quater get a real OS?
1140281041 M * ebiederm s/kind/kid/
1140281058 M * Bertl Doener: wouldn't it be better to do: atomic_inc(&init_task.namespace->count); ?
1140281102 M * oliwel Hollow: we are talking not of the same things ;)
1140281119 M * Doener Bertl: I just made it look like the stuff for files
1140281120 M * ebiederm Actually it should be get_namespace(&init_task.namespace)...
1140281122 M * Bertl Doener: basically same question applies for the files
1140281132 M * oliwel Hollow: I want to have - lets say - three groups of guests - and I want to issue "vserver group1 start"
1140281148 M * Doener ebiederm: and in that regard like the fs_struct stuff
1140281219 M * ebiederm Doener: I have looked at that yet.  I just know if there is a specific wrapper for atomic_inc it should be used.
1140281233 M * Bertl agreed!
1140281263 Q * DemonHost Quit: Vive la Parx!
1140281384 M * ebiederm Bertl: I have spotted a first instance of entering a guest :)
1140281405 M * Doener ok... anything else that i messed up within 3 lines of code?
1140281408 M * Bertl ebiederm: hmm? please elaborate!
1140281427 M * ebiederm Bertl: Doener is bug fixing it!
1140281428 M * Bertl Doener: did you add whitespace?
1140281442 M * Bertl Doener: just kidding :)
1140281463 M * Doener ebiederm: ?
1140281509 M * phreak`` oliwel: we'll probably need a script to do that
1140281526 M * phreak`` AFAIR the was something like vserver --startall or something like that ..
1140281548 M * Doener btw, after how many fixes for stupid bugs in that patch will I get killed by lkml folks? *g*
1140281614 M * Hollow oliwel: /usr/lib/util-vserver/start-vservers
1140281639 M * Bertl Doener: if you follow lkml, you do not get killed, you get promoted arch maintainer :)
1140281654 M * oliwel Bertl: *rotfl*
1140281657 M * ebiederm Doener: As long as you have have call the get_namespace to bump the count on struct namespace you should be good.
1140281674 M * phreak`` Hollow: thanks :)
1140281677 M * ebiederm Bertl: Is there a difference?
1140281723 M * Doener ok. about the "init_task." instead of "current->", what's the reason for that? should the fs and files ones also be changed?
1140281726 M * oliwel phreak``: its not an issue to write it - but I think it is a nice feature for the main toolset
1140281764 M * ebiederm Doener: On that score it simply doesn't matter.  init_task never exits so you can't get it wrong.  So being consistent is probably best for now.
1140281790 M * phreak`` oliwel: Hollow said it already .. /usr/lib/util-vserver/start-vservers is the script that is gonna help you with that :)
1140281802 M * oliwel Hollow: "start-vservers -m GROUP1 --start" should do ?
1140281821 M * Doener ok, then I'll stay with current->
1140281830 M * oliwel phreak``: yes I saw it - just started answering from the top of the log;)
1140281830 M * Bertl Doener: so ebiederm will accept your patch into his tree :)
1140281899 M * ebiederm When you repost CC Andrew Morton if you didn't the first time.  If you CC me I will ack it :)
1140281943 M * ebiederm This is something that almost needs to be sent straight to Linus because we don't want it missed for 2.6.16...
1140281949 M * Bertl Doener: I won't ack it, as this would only lower your chances :)
1140281973 M * ebiederm Bertl: Your code is that bad?
1140281983 M * Doener Bertl, Andrew and Viro are currently being CC'd (somehow I always mess up when Viro is involved...)
1140282037 M * Doener ebiederm: @xmission.com is the right address?
1140282042 M * ebiederm Yes.
1140282101 J * bragon ~bragon@sd866.sivit.org
1140282103 M * oliwel Doener: so - what is the last version of the three lines - wanna compile ;)
1140282114 M * Bertl welcome bragon!
1140282147 M * Doener oliwel: the atomic_inc will work just fine, it's just bad style, so if your kernel is already up and running, you can keep that
1140282160 M * bragon hello Bertl
1140282164 M * bragon how are you ?
1140282199 M * Bertl bragon: fine, thanks, and you?
1140282207 M * oliwel no I didnt compile - just have the two lines fixed 
1140282256 M * bragon Bertl i have problem with postgresql :/
1140282343 M * Bertl how so?
1140282349 M * oliwel Doener: May I ask you a stupid  question on the fix..
1140282356 M * Doener of course
1140282378 M * oliwel Doener: if I understand this correctly, it does not fix the problem that the ext3 is not free'd but only puts the context's together ?
1140282443 M * Doener oliwel: the problem is that the namespace is not released by kjournald and thus the mount stays which in turn keeps the kjournald alive
1140282490 M * Doener the patch makes kjournald live in init's namespace and thus only keep a namespace alive that'll never die anyways
1140282527 M * oliwel Doener: yes thats what I meant - but in this case the namespace should still hold a handle on the disk - dont it ß
1140282550 M * oliwel So - why is the mount free'd ?
1140282572 M * ebiederm The mount is only in the private namespace, so now that the private namespace can exit the mount is freed.
1140282575 M * Doener vnamespace -n mount /dev/hd5
1140282613 M * Doener --> new namespace (1 user) --> mount (2 users) --> daemonize (1 user) --> process ends (0 users) --> mount automatically freed
1140282625 M * Bertl this is the beauty of references, you remove the kjournald process from the namespace, the namespace is disposed, the mount get's unmounted and the kjournald exits :)
1140282632 M * Doener users == users of the namespace
1140282672 M * oliwel I cant really follow....merging the namespace wont kill it in my opinion...
1140282696 M * ebiederm oliwel: Not merging.  Dropping it, and using soemthing else.
1140282699 M * Doener oliwel: the namespaces are not merged
1140282699 M * oliwel Hmm - but merging kills the threaded kjournald...
1140282708 M * Doener the thread is kind of merged with init
1140282731 M * ebiederm Or as I was observer the thread enters inits namespace....
1140282770 M * oliwel So: the kjounrnald of our former private namespace holds a reference to the device - this was the bug...now we move the thread of kjounrnald to the init namespace which will lead to a relaese of the namespace that now is emty ?
1140282782 M * Bertl not the device, the namespace
1140282804 M * Bertl well, actually both
1140282807 M * ebiederm The namespace holds the reference to the mount which holds the reference to the device.
1140282821 M * Doener ok, patch sent once again :)
1140282825 M * ebiederm :)
1140282869 M * Bertl shouldn't the get happen before?
1140282876 M * oliwel ebiederm: yeah thats clear, getting a bit clearer...
1140282959 M * ebiederm Bertl:  The only thing that would stop a race is to take the task_lock on the init_task.
1140282982 M * Doener Bertl: meant me? you do it afterwards yourself :p (in vc_set_namespace, under task_lock though), but the other things also increment afterwards, so i kept it the same way
1140283002 M * Bertl ebiederm: ah, tx for reminding me :)
1140283014 M * oliwel Doener: What did you change now ? Link to lkml ?
1140283023 M * Doener http://lkml.org/lkml/2006/2/18/123
1140283031 M * Doener just from atomic_inc to get_namespace
1140283047 M * Doener as I said, not mandatory to recompile AFAIK
1140283049 M * oliwel Doener: perhaps it would be a good idea to post it to the vserver ml too, as I raised a thread there
1140283070 M * oliwel Doener: I must recompile either - so doesnt matter
1140283082 J * mkhl mkhl@200-148-40-65.dsl.telesp.net.br
1140283102 M * Doener oliwel: I would've waited till there's some acknowledgement from the kernel folks, but feel free to send a mail ;)
1140283120 M * oliwel Doener: nono - ok...pls do so if you get an answer
1140283411 T * Bertl http://linux-vserver.org/ | latest stable 2.01, 1.2.10, 1.2.11-rc1, devel 2.1.0, exp 2.1.1-rc7, 2.0.2-rc6 | util-vserver-0.30.210 | libvserver-1.0.2 & vserver-utils-1.0.3 | He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the wiki, and we'll forget about the minute ;)
1140283434 M * Bertl 2.0.2-rc6, 2.1.1-rc7 including Doener's fix (v2)
1140283454 M * Doener v2 or v3?
1140283499 M * Bertl v2
1140283527 M * Doener ok
1140283693 M * Bertl okay, off for dinner now ... back later
1140283705 N * Bertl Bertl_oO
1140283733 M * ebiederm I wonder does kjournald die if you send it SIGKILL?
1140284199 Q * oliwel Quit: Chatzilla 0.9.69.1 [Firefox 1.5/2005111116]
1140285241 N * Bertl_oO Bertl
1140285248 M * Bertl ebiederm: no, it does not :)
1140285278 M * ebiederm Bertl: Then I have a similiar issue with pspaces ....
1140285332 M * Bertl definitely, that's why I said that kernel threads need special handling, remember CLONE_KTHREAD?
1140285370 M * ebiederm Yes..
1140285485 M * Skram             
1140287186 J * rs ~rs@vol75-7-82-229-177-124.fbx.proxad.net
1140287196 M * FaUl how can i reboot an vserver from inside?
1140287211 P * rs 
1140287283 M * Bertl FaUl: reboot or reboot -f
1140287446 J * Badman ~pIRCuser4@212.45.24.131
1140287457 M * Bertl welcome Badman!
1140287473 M * Badman thanks
1140287535 M * Badman is there anybode from Russia?
1140287544 M * Bertl does it matter?
1140287560 M * Badman yes
1140287575 M * Bertl why so?
1140287604 M * Badman i am looking for somebody who can help me with my Mandrake
1140287621 M * Bertl on Linux-VServer?
1140287668 M * Bertl Badman: maybe you don't know, but Mandrake is a french distro :)
1140287691 M * Badman I understand
1140287740 M * Badman and what? Suse is Deutch distro, but in romania it's popular
1140287763 M * Bertl so what is your problem with Mandrake on Linux-VServer where you'd need a Russian expert?
1140287786 M * Badman can't launch X
1140287812 M * Bertl http://linux-vserver.org/Vservers+and+X
1140287912 M * Badman it's whitten in English. Thats why i need somebody, who could give me a link to a russian manual
1140287922 M * Badman ))
1140287957 M * Badman sorry
1140287958 M * Bertl well, I doubt that somebody has a russian manual for X on mandrake with linux-vserver :)
1140288019 M * Badman Red Hat is an US distro or UK?
1140288086 Q * Badman Quit: pIRC v2.1 < Personal IRC Team > http://xirc.ru/
1140288086 M * Bertl I'd say mainly US based
1140288517 P * PilatomiK Leaving
1140288702 M * Bertl Doener: so what is your resume regarding kernel builds with >/dev/null on 2.6.16-rc4?
1140288732 Q * Doener Read error: Operation timed out
1140288766 M * Bertl hmm, well, that was a statement :)
1140288773 M * ebiederm Bertl: you scared him away!
1140288861 M * Bertl ebiederm: so how is it going? when can we start integrating the p-space stuff?
1140288876 M * ebiederm sec
1140288889 M * Bertl no need to hurry now :)
1140289014 M * ebiederm Stage 1 cleanup /proc.
1140289017 J * Doener doener@i5387E803.versanet.de
1140289034 M * Bertl wb Doener!
1140289054 M * ebiederm I know what I want the code to lookup like and am slowly building up a good patchset to handle that.
1140289092 M * ebiederm Stage 2 I will probably replace most of the direct pid references in the kernel with task_refs.
1140289105 M * ebiederm Stag3  I will debug the semantics of my pidspace stuff.
1140289127 M * ebiederm err make that Stage 3 I will debug and merge my pidspace stuff.
1140289160 M * Doener Bertl: 1140278634 M * Doener let's see...
1140289160 M * Doener 1140278750 M * Doener full rebuild with >/dev/null done, warnings did show up
1140289168 M * ebiederm It really is just a matter of getting simple obviously correct patches and looking at them closely to see if they really are correct.
1140289177 M * Bertl Doener: hmm, interesting, tx :)
1140289366 M * ebiederm Currently I have about 13 simple patches against /proc in my patchset and I have a few more to go.
1140289418 M * ebiederm I don't anticipate any real problems merging once I get everything in shape.
1140289455 A * Bertl was more talking about Linux-VServer than mainline, but well ...
1140289518 M * ebiederm Bertl: Same difference for me.  I need to break everything into clean patchsets I am happy with or the code is not maintainable.
1140289555 M * Bertl ebiederm: I understand you want to test in mainline first :)
1140289592 A * ebiederm lol
1140289638 M * ebiederm Well if I could get it into the -mm tree and into vserver about the same time it would not hurt my feelings.
1140289694 Q * matta Ping timeout: 480 seconds
1140289731 M * ebiederm All of that is probably within the next couple of days (so say a week) to get accomplished.
1140289780 M * ebiederm If you just want to grab my pspace branch and play with it feel free.
1140289897 M * ebiederm Now to setup an evil test case to see if proc_pid_unhash is necessary.
1140289923 J * oliwel ~mail-at-o@host-62-245-151-178.customer.m-online.net
1140290067 M * oliwel Guys - question on "user quota" - is there anything special to take a look at or is it sufficient to setup a disk with quotasupport like usual
1140290090 M * FaUl is there any way to find out which capabilitys i have inside an vserver?
1140290107 M * Bertl oliwel: you want to use vroot for security purposes
1140290118 M * Bertl FaUl: grep Cap /proc/self/status
1140290137 M * FaUl Bertl: ok, and then?
1140290144 M * FaUl what does 00000000d44c04ff mean?
1140290268 M * Bertl that means that caps 0-7,10,18,19,22,26,28,30 and 31 are set
1140290301 M * FaUl ah, where can i get the information which cap what means?
1140290314 M * daniel_hozac cat /usr/include/linux/capability.h | perl -nle 'm/^#define\s+CAP_(\w+)\s+(\d+)/&&($b{$2}=$1);BEGIN{$x=hex(shift@ARGV)}END{for($i=0;exists$b{$i};$i++){print "$b{$i} is ".($x&(2**$i)?"":"NOT ")."set"}}' `awk '/CapPrm/ { print $2 }' /proc/self/status`
1140290314 M * FaUl is therepossible a programm that writes that out human-readable?
1140290334 M * bonbons vflags in vserver-utils should do the job
1140290679 Q * mkhl Ping timeout: 480 seconds
1140290811 M * oliwel Bertl: vroot - never heard of...the quota is on my drbd device - so means single mounted inside the guest...
1140290848 M * Bertl with quota you mean user and group quota, yes?
1140290977 M * oliwel Bertl: yep
1140291006 M * Bertl well, you need the quota tools to handle them properly, and they usually are executed inside the guest, right?
1140291037 M * oliwel hmm - it would be nor problem for me to set the quotas from outside
1140291048 Q * Viper0482 Remote host closed the connection
1140291056 M * Bertl well, in this case you can forget about the vroot :)
1140291081 M * oliwel So - it is sufficient to mount a "quota prepared" filesystem inside the guest ?
1140291108 M * Bertl 'inside' means for the guest (in it's namespace), right?
1140291111 M * oliwel But to take my daily lesson - tell me more about vroot ;)
1140291123 M * oliwel Bertl: yes - inside = inside the guest = vserver www1 enter ;)
1140291145 M * Bertl ahem, no, you should not mount stuff with enter ...
1140291156 M * oliwel no I dont...
1140291161 M * Bertl well, you should not be able to do so :)
1140291273 M * oliwel Bertl: so, tell me bout vroot
1140291308 M * Bertl well, iif you want to use the quota tools inside the guest, then they have to 'communicate' with the device (a kind of mainline flaw)
1140291330 M * oliwel And they cant do it "inside"
1140291349 M * Bertl well, the vroot is some kind of proxy, which allows that in a secure manner
1140291398 M * oliwel can u point me to some interessting docs ?
1140291484 M * Bertl http://linux-vserver.org/Linux-VServer-Paper-06
1140291503 M * Bertl nothing really interesting there ...
1140291511 M * Bertl http://linux-vserver.org/Standard+non-shared+quota
1140291518 M * Bertl (a short example)
1140291531 M * oliwel Wow - ONE line ;)
1140291544 M * oliwel ah that looks better
1140291587 M * Bertl well, it's all pretty easy, just configure it and that's all
1140291603 M * Bertl hands through quotactls, but blocks all other access
1140291606 M * oliwel Just like mounting ext3 ;)
1140291670 Q * kilian Ping timeout: 480 seconds
1140291745 M * oliwel looks easy - but as no one has access to my guests I can easily doe this from outside ;)
1140291771 M * oliwel So - will have a TV evening with my girl - so, nice evening alltoether ;)
1140291792 M * Bertl have a good one!
1140292378 M * FaUl Bertl: is (or was) there any problem with reciving broadcast-messages inside vservers?
1140292471 M * Bertl no, not really, but some broadcast messages require to listen to broadcast addresses
1140292471 M * Bertl (in which case you would want to add the broadcast IP to the guest)
1140292498 M * FaUl Bertl: hm, i'll try that
1140292702 M * FaUl Bertl: how to add the broadcast-ip to the guest? just create interfaces/1/ip with 192.168.26.255 into it? or some extra-stuff too? 
1140292729 M * Bertl you probably want 'nodev' there too
1140292823 Q * lilalinux Remote host closed the connection
1140292908 M * FaUl ok
1140292923 M * FaUl Bertl: got it
1140292929 M * FaUl adding the broadcast-ip helped
1140292942 M * FaUl now windows finds the server :-)
1140293819 J * dothebart ~willi@xdsl-81-173-168-227.netcologne.de
1140293901 M * Bertl wb dothebart 
1140293990 J * kilian kk@projects.verfaction.de
1140293997 M * Bertl wb kilian!
1140294246 Q * tudenbart Ping timeout: 480 seconds
1140294634 Q * shuri Quit: Quitte
1140294914 J * shuri ~shuri@64.235.209.226
1140296752 J * shedi ~siggi@inferno.lhi.is
1140297102 J * Smutje_ ~Smutje@xdsl-84-44-247-137.netcologne.de
1140297184 Q * Doener Ping timeout: 480 seconds
1140297404 Q * Smutje Ping timeout: 480 seconds
1140297404 N * Smutje_ Smutje
1140297754 M * Bertl I'm off to bed now ... have a nice one everyone! cya tomorrow!
1140297765 N * Bertl Bertl_zZ
1140297805 M * ebiederm Sleep well.
1140297813 Q * shedi Quit: Leaving
1140297984 M * kilian hey Bertl_zZ =)
1140297992 M * kilian Bertl_zZ: good night.. sleep well =)
1140299027 J * malaiwah ~malaiwah@server1.malaiwah-hosting.com
1140299419 Q * FireEgl Quit: Bye...
1140299504 Q * malaiwah Quit: [BX] Reserve your copy of BitchX-1.1-final for the Amiga today!
1140299615 J * malaiwah ~malaiwah@server1.malaiwah-hosting.com
1140300562 Q * oliwel Quit: Chatzilla 0.9.69.1 [Firefox 1.5/2005111116]
1140301691 N * ebiederm ebiederm_oO
1140303058 Q * matti Remote host closed the connection
1140303124 J * mkhl mkhl@200-148-40-18.dsl.telesp.net.br
1140305094 J * FireEgl Atlantica@2001:5c0:84dc::
1140305654 Q * shuri Quit: Quitte
1140305886 J * shedi ~siggi@inferno.lhi.is
1140307172 Q * bonbons Quit: Leaving