1113955270 J * erwan_taf ~erwan@AToulouse-105-2-2-69.w217-128.abo.wanadoo.fr 1113955467 M * Bertl welcome erwan_taf! 1113955536 Q * maharaja Ping timeout: 480 seconds 1113955783 J * mugwump ~samv@210-54-92-184.ipnets.xtra.co.nz 1113955796 M * Bertl ah, evening sam! 1113955802 M * Bertl (morning I mean ;) 1113955983 M * Loki|muh hehe 1113955986 M * Loki|muh depends ;) 1113956719 Q * ensc|w Quit: leaving 1113956820 M * Bertl Doener: to answer your question, I would check the whole IPC stuff, the tagxid flags and the new xid= mount option as well as the scheduler ... 1113956835 M * Doener ok 1113957474 Q * matta Quit: Sphinx may be image of a failed attempt to cross humans with lions. 1113957835 M * Bertl kalou: hmm, looks like I hot fixed that ;) 1113957858 M * kalou hey ... 1113957869 M * kalou nice 1113957938 M * kalou which file should I patch ? 1113957955 M * Bertl sec, just trying to cool it down a little bit ;) 1113958189 J * lilo ~lilo@lilo.usercloak.oftc.net 1113958201 M * Bertl wb lilo! 1113958584 Q * lilo Quit: bbiab 1113958840 Q * mugwump Read error: Connection reset by peer 1113958862 J * lilo ~lilo@lilo.usercloak.oftc.net 1113958916 J * mugwump ~samv@210-54-92-184.ipnets.xtra.co.nz 1113959170 M * mugwump dammit, when I start this freshly created vserver, it downs my dummy0 interface 1113959316 M * Bertl hey Sam, we have some issues with the 'sched' functionality in 2.4, wanna take a look? 1113959317 M * romke Bertl: romke 13016 0.0 0.0 1492 372 pts/2 D+ 03:01 0:00 rm kernel.spec~ 1113959331 M * romke Bertl: ps in vserver 1113959363 M * mugwump sure, why not. I'll put the compiler on. 1113959374 M * Bertl romke: hardware issues? 1113959402 M * romke Bertl: should i suspect my hdd is failing? 1113959426 M * Bertl well, could be, any warnings in dmesg? 1113959458 M * Bertl mugwump: the scenario is the following ... 1113959465 M * Bertl chcontext --flag sched --xid 100 /tmp/cpuhog & 1113959476 M * Bertl chcontext --flag sched --xid 200 bash -c "/tmp/cpuhog & /tmp/cpuhog &" 1113959517 M * mugwump ok, unless you set the proportions that will end up with xid 200 using 66% cpu 1113959564 M * Bertl hmm, proportions? 1113959572 M * kalou yep, that's exactly what happens .. but where should we put the proportions ? 1113959585 M * mugwump proportions of cpu usage - probably a bad term for "vsched settings" 1113959600 M * Bertl mugwump: we are talking 2.4 with the ticks counter ;) 1113959610 M * mugwump people still use 2.4? 1113959620 M * Bertl yes, curious, isn't it? 1113959634 M * mugwump indeed. Did you backport the vsched stuff to 2.4? 1113959663 M * Bertl nope, but I can refresh your memory regarding 2.4 1113959666 A * mugwump checks http://linux-vserver.org/Release+FAQ 1113959687 M * Bertl in goodness() 1113959704 A * mugwump runs screaming from the 2.4 scheduler 1113959709 M * Bertl if (p->s_info && (p->s_info->flags & VX_INFO_SCHED)) { 1113959709 M * Bertl weight = atomic_read(&p->s_info->ticks) / 1113959709 M * Bertl atomic_read(&p->s_info->refcount); 1113959709 M * Bertl weight = (weight+p->counter) >> 1; 1113959709 M * Bertl } else 1113959711 M * Bertl weight = p->counter; 1113959756 M * Bertl and in schedule: 1113959758 M * romke dmesg is full of SCB warnings 1113959768 M * Bertl what's SCB? 1113959778 M * kalou scsi ... 1113959788 M * Bertl ah, so that's probably the reason then ... 1113959833 M * ciphernaut is cq-tools still complete? 1113959834 A * mugwump hands Bertl the sacrificial scapegoat 1113959851 M * Bertl lol 1113959933 A * mugwump plunges a large dagger into the goat's chest, which starts spurting SCSI bus errors 1113959944 M * mugwump "begone, unstable platforms!" 1113959978 M * romke firstly rotfl 1113960018 M * Bertl ciphernaut: parse error! 1113960035 M * romke second(ly?) it isn't funny, my scsi hdd (or adapter) is dying :(... 1113960141 M * mugwump ok, there's a minute's silence, happy? :) 1113960216 M * romke no, wont be, bad luck :| 1113960222 M * romke but thx anyway :P 1113960357 M * mugwump nopaste some of the messages if you want a second opinion... 1113960378 M * mugwump nopaste is at http://www.rafb.net/paste/ 1113960627 M * romke mugwump: http://www.rafb.net/paste/results/IsthuK60.html 1113960732 M * mugwump I can see why it was hard to tell from that which part has a problem. 1113960751 M * mugwump esp, with lines like: APIC error on CPU0: 40(40) 1113960769 M * mugwump Where's SunSolve when you need it 1113960780 M * mugwump .oO( was that out loud? ) 1113960965 M * ciphernaut I havent been able to find a complete cq-tools package, which provied the cqhadd and cqhdlim programs 1113960967 N * kalou kalou_zZ 1113960988 M * Bertl night kalou_zZ! 1113960989 M * romke mugwump: :P 1113961018 M * mugwump ok, my opinion on that is that your driver doesn't give good enough error messages :) 1113961181 M * ciphernaut what is currently the best way to limit disk space per vserver? using 2.6 kernel, debian testing 1113961241 M * ciphernaut CONFIG_INOXID_UGID24=y 1113961251 M * mugwump Two options: give each vserver its own LVM partition, or use disk limits 1113961283 M * Bertl well, has not to be LVM ;) 1113961302 M * ciphernaut with disk limits, where can I locate the tools to set/veiw disk limits? 1113961309 M * ciphernaut doh 1113961321 M * Bertl vdlimit 1113961455 M * mugwump practically, setting up seperate filesystems without LVM (or loopback filesystems, perhaps) is an admin nightmare :) 1113961544 M * Bertl hmm, what about EVMS? 1113961603 M * ciphernaut which archive is vdlimit part of? 1113961665 M * mugwump true, evms is an alternative. good point. ever used it? 1113961711 M * ciphernaut I found http://www.opengenesis.org/vdlimit-0.01.tar.bz2 but have no way of telling wether this is recent or not... 1113961741 M * Bertl ciphernaut: should be part of the tools by now? 1113961910 M * ciphernaut hmmm am running 0.30.204 .... 1113962040 N * romke romke_zZ 1113962070 M * ciphernaut ok sweet. that fixed it. 1113962104 M * ciphernaut Will I need to assign a permanent xid to each vserver that is to be limited? 1113962542 M * Bertl yup, definitely 1113962679 M * ciphernaut Thanks Bertl 1113964578 J * Jdogg Jdogg@c-24-1-174-133.hsd1.tx.comcast.net 1113964738 M * Bertl wb Jdogg! 1113964761 M * Bertl Doener: it seems that vdlimit is able to add limits to non xid tagged filesystems ... 1113964775 M * Bertl (maybe we should add a check in the kernel?) 1113964842 M * Doener yep, sounds like a reasonable solution 1113965022 M * Bertl btw, I recently spend some time on investigating the various syscall entries of the kernel architectures ... 1113965209 M * Bertl and this is what I came up so far ... 1113965210 M * Bertl http://vserver.13thfloor.at/Stuff/SYSCALL/ 1113965217 M * Bertl +with 1113965244 M * Bertl I have code for the other archs too, just not cleaned up yet 1113965314 M * mugwump man, all that code could be summarised by syscall.info 1113965317 M * Doener hm.. 34k... will (try to) have a look at this tomorrow and get an hour of sleep now 1113965328 M * Doener good night! 1113965333 M * Bertl night! 1113965335 N * Doener Doener_zZz 1113965361 M * ciphernaut night 1113965495 M * Bertl mugwump: yeah, well, if all archs would work the same ;) 1113965521 M * Bertl besides, if you just want the vserver syscall, you can roughly reduce it to 1/6th ... 1113965732 M * mugwump right. it's really a shame you can't template all that up though, and have the source included 1113966034 Q * mugwump Read error: Connection reset by peer 1113966646 M * Bertl okay, I'm off to bed now too ... 1113966658 M * Bertl have a nice whatever, everyone ... cya tomorrow ... 1113966695 N * Bertl Bertl_zZ 1113968167 M * ciphernaut night Bertl 1113972280 Q * Loki|muh Ping timeout: 480 seconds 1113972727 J * mugwump ~samv@210-54-92-184.ipnets.xtra.co.nz 1113972772 M * mugwump Hello. Did I mention that dropping the dummy0 problem? :) 1113973430 J * Loki|muh loki@satanix.de 1113980473 M * DaCa hmm, I was wondering, could two vservers communicate thru a unix socket? either by hardlinking or bind mounts? (vs1.2.10, so no namespaces) 1113981859 J * wfamy bill@lns-vlq-48-mar-82-251-20-28.adsl.proxad.net 1113981869 M * wfamy hi 1113982054 M * wfamy i have to execute a set-uid script under v vserver (2.6.9 patch 1.9.3) util-vserver 0.30.204-4, i chmod 0755 my script but when i execute it i have this error 1113982061 M * wfamy YOU HAVEN'T DISABLED SET-ID SCRIPTS IN THE KERNEL YET! 1113982061 M * wfamy FIX YOUR KERNEL, PUT A C WRAPPER AROUND THIS SCRIPT, OR USE -u AND UNDUMP! 1113982072 M * wfamy any idea? 1113982516 M * eyck well, 1113982531 M * eyck what about this jump-to-conclusions-mat idea? 1113982558 M * eyck you get a mat, with various conclusions written on it, you put it on the floor, 1113982573 M * eyck and then you can -JUMP- to -CONCLUSIONS-. 1113982579 M * eyck isn't it brilliant? 1113982637 M * DaCa eyck: quick! patent it before someone else does it! 1113985339 M * SiD3WiNDR rofl eyck 1113985482 J * prae ~prae@84.14.106.134 1113985639 M * wfamy have to go see you. 1113985642 Q * wfamy Quit: Leaving 1113986811 Q * knoppix_ Ping timeout: 480 seconds 1113987015 Q * ndim Read error: Operation timed out 1113987330 J * ndim hun@helena.bawue.de 1113987341 J * knoppix_ ~knoppix@dsl-213-023-140-078.arcor-ip.net 1113988806 N * romke_zZ romke 1113991170 M * SiD3WiNDR sorry for asking this again - is ipv6 possible on 1.9.5/2.0-pre ? 1113991174 M * SiD3WiNDR inside the vserver 1113991185 M * SiD3WiNDR do I need ngnet for it? - how stable is ngnet? 1113991537 M * eyck as stable as 2.6.x, 1113991741 M * SiD3WiNDR 1.9/2.0 is as stable as 2.6, but is ngnet too? 1113991752 M * SiD3WiNDR when I heard of it it was pretty much in a very early stage 1113996344 J * DuckMaster ~Duck@dyn-83-157-150-121.ppp.tiscali.fr 1113996697 Q * ripman Read error: Connection reset by peer 1113996766 Q * DuckKing Ping timeout: 480 seconds 1113997799 M * Hollow SiD3WiNDR: it's usable, but not far from stable i'd say 1113997852 M * SiD3WiNDR what does stable mean in this context - unstable = kernel panic / stable = no panics ? :) 1113997861 M * SiD3WiNDR or more like 1113997870 M * SiD3WiNDR sometimes networking goes broken or does not support feature x 1113998745 J * yarihm ~yarihm@80-218-3-32.dclient.hispeed.ch 1114000385 M * wurd hi, I've built a vserver with the following command: vserver server7 build -m apt-rpm --hostname test --interface eth0:10.99.1.16/16 -- -dfc3 1114000412 M * wurd but i get the following error message : Building Dependency Tree... Done 1114000414 M * wurd E: Couldn't find package termcap-5.4-3 1114000444 Q * Vudumen Ping timeout: 480 seconds 1114000474 M * wurd while 'termcap-5.4-3' is the first package listed in the file named /usr/lib/util-vserver/distributions/fc3/pkgs/01 1114000556 M * wurd i've tried uncommenting all the lines in /etc/vservers/.distributions/fc3/apt/sources.list (once at a time) 1114000597 M * wurd and i'm using the latest vserver-utils ( 206 ) 1114000677 J * Vudumen vudumen@perverz.hu 1114000801 M * wurd i added a long list of packages in /usr/lib/util-vserver/distributions/fc3/pkgs/01 because when i first built my vserver, i was missing a great amount of packages. (for example i didnt have the rpm command, nor the passwd command) 1114000812 M * wurd but they wont install 1114002437 N * kevinp|gone kevinp 1114002459 M * kevinp wurd: have you tried removing the version information at the end of the package? 1114002472 M * kevinp so that it just reads termcap? 1114002831 M * wurd i'll try 1114002930 M * wurd it's working..! 1114002950 M * wurd god, if only i knew before the problem was THAT simple 1114003035 M * wurd would there be a command that would delete all version numbers from all lines? 1114003152 N * Doener_zZz Doener 1114003162 M * Doener hey folks! 1114003171 M * aba Hey Doener 1114003233 M * Doener wurd: if the version informations only consists of numbers dots and -'s (is that called hyphen?) try :%s/[0-9\.-]*$// (in vim) 1114003235 M * wurd anyway, thanks kevinp 1114003393 M * wurd that was great Doener ;) 1114003403 M * wurd you just saved me an hour 1114003417 M * Doener you're welcome 1114003425 M * Doener and yes, vim _is_ great 1114003445 M * wurd yeah, i didnt know you could do that WITH vim 1114003465 M * wurd i thought you had to do this by the shell 1114003993 M * wurd Doener i have a question 1114004009 M * wurd does the line you provided me deletes every number ? 1114004031 M * wurd because some packages have a number in its name, for example bzip2 1114004049 M * wurd i need not to have these numbers deleted 1114004071 M * Doener numbers at the end may be deleted... probably :%s/-[0-9\.-]*$// is a better one 1114004119 M * Doener -[0-9\.-]*$ means: first a -, then any number of digits, dots or -'s and then the line has to end 1114004149 M * Doener i.e. this would only make bzip2-1.2.3 become bzip2 1114004160 M * wurd good 1114004162 M * Doener while the first one would make bzip2-1.2.3 become bzip 1114004168 M * wurd seems to have worked perfectly 1114004177 M * Doener because the first one didn't say: it has to start with a - 1114004339 M * wurd damn, its hard 1114004351 M * wurd everytime a packages isnt found, the build process aborts 1114004375 M * wurd i have 800 packages 1114004380 M * wurd this will never work... 1114004391 M * Doener hm, you're using apt-rpm, right? 1114004395 M * wurd yes :( 1114004403 M * Doener ok, gimme a minute... 1114004454 M * kalou_zZ rpm --nodeps ? 1114004466 N * kalou_zZ kalou 1114004477 M * Doener kalou: apt-rpm should resolve dependencies 1114004494 M * Doener wurd's package names probably aren't correct 1114004508 M * rs hello 1114004517 M * Doener wurd: do you have apt-cache available? 1114004518 M * kalou yep, apt-rpm probably need that the directory has been "aptized" before use 1114004537 M * kalou hi rs 1114004544 M * Doener i never used apt-rpm, only debian's apt stuff, so i don't know which tools are available there... 1114004546 M * Doener hi rs 1114004632 M * kalou I think that a procedure similar to these one has to be used: 1114004634 M * kalou http://www.webmo.net/support/apt_repository.html 1114004661 M * kalou and then, wurd, you'll be able to use apt-get (apt-rpm) on your newly created apt repository 1114004687 M * kalou IMHO :) 1114004691 M * wurd Doener i do have apt-cache 1114004701 M * Doener kalou: he's using apt-rpm to build the vserver, and the package names in his list of to-be-installed packages are partially wrong 1114004750 M * wurd Doener the thing is, i've done this list "by hand" in order to fix a problem 1114004756 M * wurd it's not an "official" package list 1114004773 M * wurd if i've done that, its only because it wouldnt work another way (the right way) 1114004781 M * Doener wurd: i thought so 1114004794 M * Doener just wanted to explain to kalou what's going on ;) 1114004799 M * rs Doener: do you know how to fix the vcontext: setsid(): Operation not permitted issue ? 1114004813 M * wurd well shouldnt it work the right way ?? 1114004818 M * wurd why do i have to do all this ? 1114004820 M * Doener rs: probably CONFIG_SECURITY, but not CONFIG_SECURITY_CAPABILITIES 1114004854 M * Doener wurd: hm, i just re-read your above sentence... 1114004854 M * rs CONFIG_SECURITY is off 1114004869 M * Doener ah, setsid, sorry... 1114004875 M * rs yah :) 1114004887 M * rs I'm reading some log between you and bertl about this 1114004888 M * Doener you got the clrsid patch applied, right? 1114004894 M * rs yeah 1114004910 M * rs delta-clrsid-feat03.diff 1114004938 M * Doener rs: in src/context-sync.hc, replace Esetsid() with setsid() 1114004942 M * kalou wurd: did you try with rpm instead of apt-get ? 1114004960 M * rs ok easy:) 1114004961 M * rs thx 1114004985 M * Doener wurd: what's 'the right way'? and which problems arise with it? 1114005007 M * wurd the right way is typing the build command, and then the vserver building correctly :) 1114005016 M * wurd without having to add 800 packages in the list 1114005025 M * wurd i mean... 1114005035 M * wurd normaly all these 800 packages should already be there, no ? 1114005043 M * wurd why do i have to manually add them? 1114005052 M * wurd are the files corrupted or what 1114005120 M * Doener do you get any errors when building the vserver 'the right way'= 1114005123 M * Doener s/=/?/ 1114005124 M * wurd as for "which problem arise with it", well, when i build my vserver "the right way", it DOES work, BUT: my vserver is way too minimal, for example i dont even have the PASSWD command, nor the RPM command 1114005139 M * wurd so it missed tons of packages 1114005139 M * Doener to check your package list: 1114005142 M * Doener for NAME in `cat test`; do if [ `apt-cache search ^$NAME$ | wc -l ` -eq 0 ]; then echo "Missing/Wrong: $NAME"; fi; done 1114005155 M * Doener replace `cat test` by the name of your package list file 1114005173 M * Doener (only "test" of course, the "cat" has to stay) 1114005175 M * wurd which package list is that? 1114005191 M * Doener your file with the 800 packages 1114005198 M * wurd the one i manually createdd? 1114005202 M * Doener yep 1114005268 M * Doener the line should tell you about all the packages that can't be found in your host's apt-rpm repository... it doesn't necessarily match the repository you use for building the vserver, but for all those common packages it should work just fine 1114005294 M * wurd ok 1114005302 M * wurd the 'command' is processing .. 1114005325 M * Doener but we should check what's going on with your failing build anyways... 1114005380 M * Doener yeah, will probably take some time to complete.. 1114005561 M * kevinp wurd, when I created my own package list, I had the version numbers in there but I was using -m yum instead of -m apt-rpm, very similar, just different tools. Apparently yum can deal with the version numbers 1114005644 M * wurd maybe i should try with yum 1114005664 M * wurd but i know that it isnt rare that yum doesnt find some packages (ive used yum a couple of times before) 1114005687 M * wurd and since the (very long) build process aborts each time it doesnt find a package........... 1114005696 Q * eyck Ping timeout: 480 seconds 1114005837 M * kevinp It found all of my packages so I don't know if yum handles it any better (not finding a package) 1114005838 M * wurd Doener your command isnt finished yet and i have already something like 30 missing packages 1114005860 M * kevinp wurd: do you really have 800 packages in there? 1114005860 Q * kalou Read error: Operation timed out 1114005871 M * Doener 30 out of 800 isn't that much... 1114005892 M * Doener but as i said, we'll check the 'right way'... 1114005897 M * wurd kevinp yes, i didnt bother going through all that to delete all packages that i thought were unuseful 1114005908 M * rs Doener: I guess this change in util-vserver make it un compatible with earlier version of the vserver patch, is that right ? 1114005931 M * Doener rs: no, it just removes an error condition check 1114005953 M * kevinp But still, a minimal install I would think would have less then 300 packages 1114005959 M * rs so it's safe to push this version of util-vserver on an older kernel ? 1114006005 M * Doener rs: i guess so 1114006097 M * kevinp wurd: what is your native language? 1114006099 M * wurd kevinp what do you mean 1114006104 M * wurd kevinp french 1114006175 M * kevinp Take a look at this link, it shows which rpms are for each category, etc. This is from the comps.xml file that I told you about yesterday: http://penguin.triumf.ca/comps/fc3/comps-v4-lang-fr.html 1114006242 M * wurd kevinp ok, but what should i do then? 1114006244 M * kevinp or in english: http://penguin.triumf.ca/comps/fc3/comps-v4-lang-en_US.html 1114006280 M * kevinp well, what are you going to use this vserver for? Install those packages that you will actually use, just like on a normal install 1114006336 M * wurd by copying the package-names i find in the website to the 01 file 6 1114006337 M * wurd ? 1114006345 M * wurd 6 -> ? 1114006354 M * kevinp exactly 1114006363 M * Doener wurd: you only added the file with the package list to the distributions configuration, right? 1114006626 M * wurd im not sure i understand your question 1114006648 M * Doener what changes did you make to /etc/vserver/.distributions/fc3 ? 1114006771 M * wurd i edited the file named 01 1114006778 M * wurd which only contained 2 package names 1114006794 M * wurd and inside this file i added 800 packages 1114006795 M * Doener ok, you got the original file at hand? 1114006808 M * wurd im not sure i kept it :( 1114006827 M * wurd i thought that when adding the 800 packages i have done cat mypkgs >> 01 1114006832 M * wurd but it seems it didnt work 1114006835 M * Doener hm, you said there was a file 01 in /etc/vservers/.distributions/fc3 ? 1114006837 Q * erwan_taf Read error: Operation timed out 1114006843 M * wurd yes 1114006852 M * wurd also theres 02 and 03 1114006906 M * Doener that's below /usr/lib/util-vserver/distributions/fc3 isn't it? 1114006917 M * Doener s/below/in/ 1114006943 M * wurd yes. 1114006962 M * Doener ok, you shouldn't modify those, i'd say... 1114007006 M * Doener anyway, could you move your modified files somewhere else and restore the originals ? 1114007020 M * Doener (in the source tree, they're in distrib/fc3/pkgs 1114007079 M * Doener then try to build a vserver and check if any errors occur. making the output available via pastebin may also be helpful 1114007123 M * kevinp Doener: the way I understand it, that all worked fine previously, wurd just wanted more packages installed then what was installed by default 1114007150 M * Doener kevinp: he didn't have, for example, passwd... i wouldn't call that fine ;) 1114007158 M * wurd kevinp from what i heard, my vserver was so minimal that it didnt seem "fine"! 1114007176 M * kevinp ok 1114007216 M * wurd where is the source tree Doener ? 1114007269 M * Doener wurd: how should i know? i guess you downloaded and untarred the util-vserver tarball ;) 1114007317 M * wurd actually i built an rpm with the util-vserver tarball 1114007324 M * kevinp Doener, can you confirm that my thinking is correct? http://linux-vserver.org/CentOS_HowTo 1114007349 M * kevinp Specifically step C 1114007402 M * Doener kevinp: no idea, i've not looked at those parts of the tools yet... too much bash stuff ;) 1114007426 M * Doener wurd: well, then untar the tarball now ;) 1114007476 M * romke how to set global alias for all shells in system? 1114007503 M * kevinp romke: what distro? 1114007557 M * romke kevinp: pld :P 1114007607 M * kevinp never heard of it, but usually there is a file like /etc/profile or something that executes first for all users 1114007639 M * romke i need this because of xargs, it on large groups of files generates too many arguments to commands, way around it is to pass to xargs -n parameter, but i'd like set it system-wide 1114007647 M * wurd Doener ok so you want me to build the vserver "the right way" with the original 01,02,03 files and to note any error that occurs during build? 1114007672 M * Doener wurd: exactly... 1114007685 M * romke kevinp: will subshels keep that? 1114007687 M * wurd can i do like, " vserver server1 bla bla rest of build command > buildlog " ? 1114007696 M * kevinp Doener: the reason I ask, is I had wurd follow similar steps but for FC3 1114007703 M * Doener wurd: will probably also need 2>&1 1114007714 M * wurd ok thanks i'll do that 1114007740 M * kevinp Last time I had to do it was on an old UNIX OS, but that OS kept it 1114007756 M * kevinp haven't tried it recently 1114007768 M * kevinp one way to find out :) 1114007903 M * wurd ok im ready to build 1114007916 M * wurd should i stay with apt-rpm or should i try something else like yum 1114007925 M * kevinp Doener: anyway, it worked for me, I created a "minimal" install (based off of an actual install of the OS) and then used it as a base for making additional vservers that included even more packages after that 1114007942 M * Doener wurd: stay with apt-rpm, we want to find the bug there, right? 1114007952 M * wurd yes. 1114008023 J * erwan_taf ~erwan@AToulouse-105-2-2-69.w217-128.abo.wanadoo.fr 1114008129 Q * wurd Read error: Operation timed out 1114009431 N * Bertl_zZ Bertl_oO 1114009564 Q * TheSeer Remote host closed the connection 1114009853 J * _jeru ~jeru@p54BFDDAE.dip.t-dialin.net 1114009932 M * _jeru Hi, can anybody tell me where I can find the recent vserver-grsec patch ? 1114010123 M * DaCa _jeru: http://vserver.limehouse.org 1114010159 M * _jeru DaCa: thanks a lot :) 1114010206 M * _jeru DaCa: do you know if this will be a more "permanent" location for the patches ? 1114010309 M * DaCa _jeru: well, its me who made them, I'll try to keep up with new releases :) 1114010335 M * _jeru DaCa: nice to hear :) 1114010785 J * wurd ~kvlt@modemcable143.235-201-24.mc.videotron.ca 1114010790 M * wurd i got disconnected >:( 1114010797 M * wurd Doener are you still there 1114010804 M * Doener sure 1114010809 M * wurd good 1114010822 M * wurd well something happened during the build and i cant explain it 1114010838 M * wurd 7:glibc ########################################### [100%] 1114010838 M * wurd Done. 1114010838 M * wurd Reading Package Lists... Done 1114010838 M * wurd Building Dependency Tree... Done 1114010838 M * wurd E: Couldn't find package pam_krb 1114010850 M * wurd i dont know where pam_krb comes from... 1114010915 M * wurd originally pam_krb was in my "manually-done 01 file" 1114010927 M * Doener probably a dependency of another package... 1114010933 M * wurd the real package name is pam_krb5 but the 5 was deleted when i did your vim command line 1114010959 M * wurd the thing is, i have replaced the files and now the files in /usr/lib/util-vserver/distribution/fc3/pkgs are the original ones !!! 1114010961 M * Doener hm... and you moved your files out of the way? (i.e. an other directory) 1114010972 M * wurd but still the build is looking for pam_krb (package that doesnt exist) 1114010979 J * TheSeer ~theseer@border.office.salesemotion.net 1114010980 M * wurd i didnt move them but i renamed them 1114010986 M * wurd so that is it ? 1114010986 M * TheSeer re.. 1114010991 M * wurd i have to move them ? 1114010993 M * Doener wurd: move them ;) 1114010995 M * wurd :) 1114011010 M * Doener the naming is just the order in which they're evaluated... 1114011019 M * wurd oh :) 1114011057 M * kevinp Bertl_oO: I have a strace of my vserver start where klogd fails at http://roundsphere.com/stuff/strace.out, that's with the command: 'strace -fF -o /root/strace.out vserver min-centos4 start'. It also output the following not to the file: "upeek: ptrace(PTRACE_PEEKUSER,8740,44,0): No such process" and I had to kill the strace process. Let me know if you need more. I can install strace inside a vserver if that is needed. 1114012503 M * kevinp I want to start digging into restricting my vservers now, cpu, memory, etc. Where is the preferred place to change the limits? 1114012742 M * kevinp /etc/vservers/vserver-name/rlimits ? 1114012793 M * Doener yep, except cpu 1114012810 M * Doener the cpu rlimit would limit the cpu time in total, 1114012855 M * kevinp is it possible to make it look from within the vserver that they only have a 500MHz processor or something like that? 1114012969 M * kevinp Or do they always see what the cpu really is, but just can't use it all? 1114013463 M * Doener hm, i guess /proc/cpuinfo is not virtualized... 1114013647 M * DaCa wouldnt be very ethical imho 1114013759 M * kevinp I was just thinking of having different packages, a 500 MHz, a 1 GHz package, etc and they could easily check what they had 1114013783 M * kevinp instead of wondering how much of this 3 GHz processor am I really allowed to use... 1114013810 M * Doener the limits don't work Mhz-wise anyway 1114013860 M * Doener if you only have short bursts, you may get the full 3Ghz all the time... 1114013864 M * Doener s/may/might/ 1114013874 M * kevinp so people that advertise 166Mhz/$20, 333Mhz, $25, etc, must either be using different physical servers or something other than vserver? 1114013918 Q * prae Quit: Client exiting 1114013945 M * kevinp hmm, interesting, so you really can't throttle the cpu then 1114013955 M * kevinp it's more just scheduling 1114013988 M * kevinp can you throttle other things like memory? Say, you only get to use 32MB, for example 1114014095 M * Doener you can limit vsz and rss 1114014478 M * DaPhreak|screen_detached hmm .. weird 1114014547 M * Doener kevinp: you should have a look at the token bucket scheduler stuff in the linux-vserver paper to get a feeling for it, and what is possible... 1114014569 M * DaPhreak|screen_detached very weird .. same as kevinp ... upeek: ptrace(PTRACE_PEEKUSER,3856,44,0): No such process 1114014594 M * Doener you can limit a vserver to ,for example, approx. 33% cpu (or whatever), but it won't be exactly 1Ghz of 3Ghz 1114014613 M * Doener DaPhreak|screen_detached: probably because the process is in an other context 1114014629 M * Doener time for dinner... back later! 1114014633 N * Doener Doener|gone 1114014635 M * DaPhreak|screen_detached well it has actually context 18 :) 1114014756 M * DaPhreak|screen_detached kevinp: what kernel are you using ?! 1114014785 M * kevinp 2.6.11.7-vs2.0-pre1 1114014807 M * DaPhreak|screen_detached nothing like 2.6.11.7-grsec2.1.5-vs2.0pre1 ? 1114014822 M * kevinp Thanks for the info Doener|gone! 1114014837 M * kevinp no grsec patches on mine 1114014856 M * DaPhreak|screen_detached hmm so it might be either something with gentoo or something with vserver .. 1114014860 M * DaPhreak|screen_detached which tools ? 1114014864 M * DaPhreak|screen_detached 205-r1 ? 1114014877 M * kevinp I've had 205, but I'm compiling 206 now 1114014896 M * kevinp I'm running CentOS, so it's not gentoo 1114014904 M * DaPhreak|screen_detached ah ;) sorry for that :) 1114014927 M * DaPhreak|screen_detached so i guess you're linking against dietlibc, right ? 1114014934 M * kevinp yep 1114014965 M * kevinp I just installed beecrypt for 206, didn't have it until now 1114015002 M * DaPhreak|screen_detached i'm expieriencing this problem since i switched to 2.0_pre1 .. my vserver don't even stop :) they hang at the last process inside the vserver exiting 1114015056 M * DaPhreak|screen_detached <-- gone again .. coming back later 1114015098 M * kevinp mine stop fine 1114015801 M * kevinp I had to reboot to clear the context out when I killed the strace 1114016053 M * kevinp Here's the strace with the 206 tools: http://roundsphere.com/stuff/strace.out2, also got "upeek: ptrace(PTRACE_PEEKUSER,12849,44,0): No such process" on stdout 1114016101 Q * pusling Ping timeout: 480 seconds 1114016217 J * stegbth ~stegbth@stegbth.sim.tronicplanet.de 1114016221 M * stegbth hi all 1114016286 M * stegbth i am playing currently with a running vserver on a debian 3.0 with kernel 2.4.30. i moved the directory over to a debian sarge with kernel 2.6.11 newer vserver patches. 1114016292 Q * erwan_taf Remote host closed the connection 1114016297 M * stegbth when i try to start the vserver it looks like it is running 1114016307 M * stegbth but when i try to enter i get: 1114016309 M * wurd Doener|gone / Doener tell me when you'll be back 1114016339 M * stegbth error: vc_new_s_context 1114016378 M * stegbth any idea? 1114016746 M * stegbth the complete error is: 1114016800 M * stegbth chchontext: vc_new_s_context(): Invalid argument 1114016879 M * kevinp I ended up installing strace on a test vserver and ran 'strace -fF klogd -2', here's the results: http://deadbeefbabe.org/paste/376 1114017869 M * kevinp stegbth: you may want to look here: http://vds.pas-mal.com/irclogs/vserver-log.20050122.html 1114018043 M * kevinp Sorry I'm not more help! :) 1114018152 M * stegbth thanx 1114018224 J * pusling ~pusling@195.215.29.124 1114019135 J * virtuoso_ ~s0t0na@80.253.205.251 1114019135 Q * virtuoso Read error: Connection reset by peer 1114020274 P * stegbth Client exiting 1114020559 Q * DaCa Ping timeout: 480 seconds 1114021689 J * DaCa ~danny@mail.limehouse.org 1114022804 J * nayco ~nayco@82.249.51.190 1114022828 M * nayco 'llo !!! 1114023056 J * DuckKing ~Duck@dyn-83-157-157-42.ppp.tiscali.fr 1114023267 Q * DuckMaster Read error: Operation timed out 1114023404 M * wurd is there a vserver expert in here? :) 1114023609 M * kevinp Does anyone know of if there any documents that give sample rlimit configurations for 2.6? 1114023876 M * nayco has anyone seen enrico (ensc) ? 1114026583 J * eyck eyck@81.219.64.71 1114026846 M * DaPhreak|screen_detached kevinp: this issue where you have to kill the vshelper, does it only happen with 206 ?! or which tools version do you have ? 1114027567 N * Doener|gone Doener 1114027575 M * Doener wurd: i'm back 1114027694 M * nayco ''llo Doener ! 1114027700 M * Doener hi nayco 1114027737 M * nayco Do you know if Enrico is available these days ? 1114027751 M * Doener on irc? i don't know... 1114027892 M * nayco no, in general ? I mean, I sent two emails to him, and he seems busy... I can't even conctact him here. Well it has a capital importance: I sent hom a new stylesheet for the "flower page". Ok, this is a joke ;-), it's not really important, but I was wondering 1114027924 M * nayco [...]I sent him[...] 1114028146 M * Doener maybe he still hasn't got permanent (affordable) internet access 1114028182 M * nayco oh, ok. Do you know where he is from (country) ? 1114028228 M * Doener germany 1114028284 M * nayco thx 1114028290 M * Doener eastern germany, and unfortunately the phone system there is using fibre optics, and thus no dsl is available... just isdn, which is pretty expensive 1114028302 M * nayco arggg... 1114028339 M * nayco yes, in france it has always been reserved for enterprises: Too expensive for people. Few had isdn 1114028416 J * Nik ~Nik@cable-153-130.online.bg 1114028421 M * Nik hi all 1114028431 M * nayco 'llo ! 1114028433 M * Doener isdn for voice isn't that expensive here (IIRC about 10 Euros more than standard analog phone), but you can't get a flat rate for internet access :/ 1114028485 M * nayco You mean you have to pay your isdn account, then for internet (Youre from germany too ?) ? 1114028499 M * Doener yes, yes :) 1114028535 M * nayco ;) 1114028564 M * Doener you can get an isdn account, and then you still have to pay for calls and internet... prices are around 1 euro cent per minute for internet access over isdn 1114028608 M * nayco whaoooh. 1114028612 M * Doener same for standard phone accounts... only with dsl you can get a flat rate... 1114028623 M * nayco vserver hackers are brave. 1114028629 M * nayco ;) 1114028669 M * Doener AFAIK Bertl is on isdn, too... (he's from austria btw ;) 1114028707 M * Nik try to get internet in my country ;-) (and its also Europe...) 1114028710 M * nayco here, we have dsl 2048 Kb + free national phone for 30¤. We think it took veeeery long, but it's not that bad... If you live in a town, that is, because in the countryside :(... 1114028759 M * nayco ok, gotta eat. see you soon ! 1114028771 M * Doener enjoy your meal! 1114028772 M * Doener cya 1114028776 M * nayco thx 1114028877 Q * nayco Quit: Bonne nuit ! 1114029477 Q * Nik Ping timeout: 480 seconds 1114029939 Q * wurd Read error: Operation timed out 1114031302 J * Taube ~Taube@247.227.adsl.i4surf.net 1114031626 M * Doener i'm off to bed now... too tired... have a good day/night/in-between everyone! 1114031632 N * Doener Doener_zZz 1114032459 Q * Taube Quit: Using KVIrc 3.0.1 'System Virtue' 1114032906 N * Bertl_oO Bertl 1114032925 M * Bertl evening folks! 1114033471 J * nayco ~nayco@82.249.51.190 1114033492 J * erwan_ho ~erwan@lns-vlq-39f-81-56-133-136.adsl.proxad.net 1114033499 M * Bertl evening nayco! erwan! 1114033509 M * nayco 'llo, Bertl :) ! 1114033607 M * nayco Bertl: I finally found the problem with Samba and Cups... What annoys me is that I couldn't find that in the logs. But it wasn't Vserver related ;) 1114033631 M * Bertl yup, read that ... 1114033660 M * nayco I updated the wiki. Hope I did it right... 1114033709 M * erwan_ho hey Bertl 1114033718 M * Bertl nayco: if it is readable, I guess so ;) 1114033810 J * sepp1 ~|Rogue@port-212-202-71-11.dynamic.qsc.de 1114033814 M * sepp1 hi 1114033837 M * sepp1 any got experience with tun devices and vservers 1114033838 M * sepp1 ? 1114033848 M * nayco Oh, and I tried to run a teamspeak server in a Vserver, It needs more testing (Partly because the client - my machine - seems to have problems with oss), but still I managed to connect to the server and create a chatroom, without any modification to the server conf. So, I think I will be able to add it to the "Working Programs" :) 1114033900 M * nayco sepp1: er... no. 1114033956 M * Bertl welcome sepp1! 1114033974 M * Bertl what's up with tun and vserver? 1114033982 M * sepp1 problem is i can use the tun device in the host system 1114033995 M * sepp1 i use it as an openvpn server 1114034019 M * sepp1 however when i try to bind a service to the tun ip it does not work 1114034066 M * Bertl on the host or inside a vserver? 1114034103 M * sepp1 on the host 1114034121 M * sepp1 tun devices in a vserver did not work at all 1114034170 M * Bertl that sounds very unusual ... 1114034188 M * Bertl ould you provide some details regarding your system, maybe the output of testme.sh? 1114034198 M * sepp1 let me check 1114034245 M * Bertl http://vserver.13thfloor.at/Stuff/SCRIPT/testme.sh 1114034253 M * sepp1 a thx 1114034261 M * Bertl you#re welcome ;) 1114034333 M * sepp1 Linux-VServer Test [V0.12] (C) 2003-2005 H.Poetzl 1114034333 M * sepp1 chcontext is working. 1114034333 M * sepp1 chbind is working. 1114034333 M * sepp1 Linux 2.4.29-vs1.2.10-grsec i686/0.30/0.30 [E] 1114034333 M * sepp1 VCI: 1114034335 M * sepp1 --- 1114034338 M * sepp1 [000]# succeeded. 1114034341 M * sepp1 [001]# succeeded. 1114034343 M * sepp1 [011]# succeeded. 1114034346 M * sepp1 [031]# succeeded. 1114034348 M * sepp1 [101]# succeeded. 1114034351 M * sepp1 [102]# succeeded. 1114034353 M * sepp1 [201]# succeeded. 1114034356 M * sepp1 [202]# succeeded. 1114034358 M * sepp1 that would be it 1114034369 M * sepp1 openvpn is working just fine in servermode on the host 1114034403 M * Bertl okay, and how do you verify that binding to an ip doesn't work? 1114034443 M * sepp1 actually it try to bind a server process to the tun0 ip 1114034541 M * sepp1 this works fine on system without vserver 1114034553 M * Bertl okay, and that fails with the bind() or with what? 1114034566 M * sepp1 jep 1114034570 M * sepp1 yes 1114034585 M * Bertl could you provide an strace -fF of that (via pastebin.com or so) 1114034636 M * sepp1 just give me a second 1114034658 Q * yarihm Quit: Leaving 1114034726 Q * romke Read error: Connection reset by peer 1114034760 Q * erwan_ho Remote host closed the connection 1114034773 M * sepp1 http://pastebin.com/273881 1114034782 M * sepp1 just the end of the output 1114034794 M * sepp1 10.23.24.1 is the tun0 device ip 1114034858 M * Bertl hmm, now the important question: 1114034867 M * sepp1 go ahead :) 1114034891 M * Bertl do you logon via ssh and use the v_ssh script for restricting ssh to the host ips? 1114034899 M * sepp1 yes 1114034909 M * Bertl well, then that is the reason ;) 1114034914 M * sepp1 if have to vservers running on the system 1114034928 M * sepp1 I can access all of them via ssh 1114034939 M * Bertl it's much better to configure the host ssh to bind to a subset of ips 1114034966 M * Bertl the v_ssh script is depreciated because of the unwanted restriction to the host 1114034984 M * sepp1 unwanted restrictions ? 1114034996 M * Bertl well, let me give an example 1114035011 M * Bertl you configure the host to use 10.0.0.1 okay? 1114035017 M * sepp1 ok 1114035019 J * romke romke@procyon.romke.net 1114035027 M * Bertl now you use the v_ssh script to limit sshd to 10.0.0.1 too 1114035033 M * sepp1 yes 1114035045 M * Bertl after a logon, your set of ips is _restricted_ to 10.0.0.1 1114035051 Q * Vudumen Ping timeout: 480 seconds 1114035059 M * Bertl any attempt to bind to other ips will be rejected 1114035066 M * sepp1 uhoh 1114035073 M * Bertl (that's the purpose of such a chbind ;) 1114035086 M * sepp1 ok i get it 1114035102 M * sepp1 so i should bind the sshd to the ip via the ssh config file ? 1114035128 M * Bertl yep, that's a much better solution, especially if you want to administer the vservers via ssh 1114035159 M * sepp1 right now i have a sshd running in every vserver bound to its individual ip 1114035220 M * Bertl yes, that's fine, they will work fine too 1114035259 M * Bertl just use ListenAddress in the hosts sshd_config 1114035271 J * Vudumen vudumen@perverz.hu 1114035271 M * sepp1 ok 1114035273 M * Bertl and do not use the v_ssh wrapper ... 1114035293 M * sepp1 do you think that this will resolve the bind problem of the other service 1114035304 M * Bertl definitely 1114035327 M * Bertl bind(3, {sin_family=AF_INET, sin_port=htons(15667), sin_addr=inet_addr("10.23.24 1114035330 M * Bertl .1")}}, 16) = -1 EADDRNOTAVAIL (Cannot assign requested address) 1114035347 M * Bertl that's what you get when you try to bind to an ip out of the chbind range 1114035384 M * albeiro out of couriosity _only_ - why not -EACCESS or -EPERM ? 1114035412 M * Bertl I'd say, out of tradition ;) 1114035429 M * albeiro hehe, ok, 1:0 for you 1114035431 A * Bertl never bothered to change it ... 1114035477 M * albeiro not that there would be some understable difference between -EACCESS and -EPERM ;] 1114035519 M * Bertl well, most of the error codes have very (often unfortunately) specific meanings 1114035550 M * sepp1 I have change the sshd config of the host system to bind to a specific ip 1114035554 Q * nayco Quit: Bonne nuit ! 1114035570 M * sepp1 still i can not bind something to the tun0 ip 1114035586 M * sepp1 maybe it just don't get the problem ... 1114035591 M * Bertl you sure that you have restarted the sshd and logged on again? 1114035604 M * sepp1 i did that 1114035612 M * Bertl (and more important, that you did restart the sshd from outside the chbind jail?) 1114035658 M * sepp1 well i restarted the sshd on the host system 1114035670 M * Bertl via ssh? 1114035675 M * sepp1 yes 1114035679 M * Bertl lol 1114035681 J * kalou ~kalou@AToulouse-203-1-3-208.w80-14.abo.wanadoo.fr 1114035703 M * kalou 'night 1114035704 M * albeiro Bertl: it does work until you change bind ip. done that many times, does not interrupt work. 1114035735 M * sepp1 there is no other way i can only access the system via ssh 1114035752 M * Bertl sepp1: okay, viable 'retart' options include: system reboot, cron job, logon service different to ssh 1114035772 M * sepp1 ok i try that tomorrow ... thx for the help 1114035780 M * Bertl you're welcome! 1114035791 M * sepp1 cu 1114035793 M * Bertl night kalou! 1114035800 Q * sepp1 Quit: leaving 1114036535 M * kalou I'm reading sched.c right now, and would have a little question about the "idle task" .. 1114036546 M * Bertl go ahead! 1114036560 M * kalou does it have a pid ? 1114036579 M * Bertl yes and no, it has a pid entry, but that is zero 1114036580 J * yarihm ~yarihm@80-218-3-32.dclient.hispeed.ch 1114036584 M * kalou ok 1114036600 M * Bertl and there is one per cpu ;) 1114036613 M * Bertl welcome yarihm! 1114036681 M * albeiro per physical / logical cpu (think hyperthreading) ? 1114036689 M * Bertl logical cpu 1114036697 M * yarihm yo Bertl 1114037443 M * kalou what about the task_struct *child_reaper ? :) 1114037489 M * Bertl what about it? ;) 1114037493 M * kalou I understand that init_task is the link to the first task created on the system. 1114037513 M * kalou So I'm wondering, first, how can I browse through the list of all processes in the system 1114037536 M * Bertl there are some macros for doing that 1114037538 M * kalou should I follow init_task, next, next, next ? 1114037558 M * kalou ok heard about current() 1114037592 M * Bertl that's a hack to get you the task 'context' you're currently in 1114037687 M * Bertl for_each_process(p) will walk all the 'processes' 1114037799 M * kalou ok, each call to for_each_process(p) moves p forward ..? 1114037840 M * Bertl hmm, you have a tool to search in the kernel sources yet? 1114037870 M * kalou yep, find, xargs, grep 1114037900 M * Bertl yeah, well, add cscope, ctags, vim adn/or emacs to that ... 1114037908 M * albeiro speaking about moving around kernel sources, is there any way to join vim and cscope ? so i could walk functions just like hyperlinks ? 1114037910 M * kalou in 2.4, for_each_process does not seem to appear in any .c file 1114037930 M * Bertl yes, 2.4 it was named differently .. for_each_task IIRC 1114037938 M * Bertl +in 1114037960 M * Bertl albeiro: yup, there is a vim plugin, right from the cscope page ;) 1114037963 M * kalou ok. Let's take a look at cscope and ctags 1114037998 M * albeiro Bertl: sometimes i think you should publish a book about kernel hacking ;] 1114038007 M * albeiro the right one ;p 1114038038 M * Bertl hehe, well, maybe I will at some point ... 1114038375 M * albeiro strings `which vim` | grep cscope test passed ;p 1114038581 Q * yarihm Quit: Leaving 1114038737 M * kalou sorry .. my body is asking for sleep. 1114038749 M * kalou see you tomorrow, guys ! 1114038758 N * kalou kalou_Zz 1114038808 M * Bertl night kalou_Zz! 1114038863 M * albeiro does powerbook has sysrq ? 1114038911 M * Bertl yup, should be there ... 1114038944 M * albeiro how can i invoke it ? cannot find :/ 1114039001 M * Bertl F13 IIRC 1114039079 M * Bertl I know there is no such key on the built in keyboard ... 1114039119 M * Bertl (but it can easily be remapped to other keys, well, you ahve to do some remapping anyway ...) 1114039151 M * albeiro i will dog into remaping, thx 1114039191 M * albeiro that's much better answer comparing to what i heard recently. it was "sysrq is on your external keyboard" ;p 1114039223 M * Bertl yeah, is probably the powerbook way to solve this ;) 1114039272 M * albeiro WOW ! cscope plugin is amazing 1114039283 M * albeiro and i was about to go sleep, no way ;p 1114039312 M * Bertl yeah, it's quite cool ... 1114039345 M * Bertl although cscope gets some kernel definitions wrong ... 1114039356 M * albeiro never seen it 1114039376 M * Bertl so for example after doing an unsuccessful global definition search, you should always look for C sybol too 1114039386 M * Bertl symbol even 1114039458 M * albeiro interesting... in fact i bet there were some unsucesfull searching for symbols in my career ;] 1114039490 M * albeiro one more question - having CTRL-\ s pressed i got this nice and truly amazing list of occurences of this symbol. nice. but how can i move upward this list ? (downwards is easy ;) 1114039582 M * Bertl I don#t think there is an 'up' binding ... 1114039699 M * albeiro so many shortcuts to remember, will of course get used to them, but for now i will prepare small card so i could cheat ;] 1114041040 M * Bertl okay, off to bed now ... night folks! 1114041058 N * Bertl Bertl_zZ 1114041106 M * albeiro good night Bertl_zZ :]