1112745905 M * Bertl excellent ... 1112746769 M * Bertl hmm, scheduler and virt already done ... ;) 1112746828 M * micah you are faster than my kernel compile 1112746840 M * micah I had an error and had to restart it though 1112746854 M * Bertl me too, removed the wrong file ;) 1112746861 M * micah agh! you are still faster than my computer 1112746933 M * albeiro buy a faster computer or distract Bertl a bit to slow his down ;] 1112746965 M * Bertl lol 1112747028 M * micah hehe 1112747053 M * micah part of the problem is that debian kernels select a high number of modules to compile, so it takes a much longer time than if you create a very modular kernel 1112747142 M * Bertl ah, yeah, btw, you should figure out a way to test/compile the vserver debian kernel with minimal recompile times (just a note) 1112747164 M * Bertl I heard that there are some options to do that (even with kernel package building) 1112747167 M * albeiro msgfmt: command not found 1112747171 M * albeiro what do i lack ? ;p 1112747189 M * Bertl gettext 1112747189 M * albeiro (sorry for disturbing, i am just slowing Bertl down ;p) 1112747193 M * albeiro thx 1112747227 M * micah heheh 1112747246 M * micah Bertl: good point, I'll look into it 1112747302 M * Bertl isolation done ;) 1112747340 M * Bertl ah, memory looks like fun ... 1112747431 M * micah you are seriously fast 1112747443 M * micah It probably helps you know this code so well 1112747460 M * Bertl yeah, guess that helps ;) 1112748887 M * micah there is ccache, which can speed up recompiles significantly 1112748930 M * Bertl yeah, well, the kernel doesn't need much recompile (usually) just doing debian packages normally does ... but feel free to investigate/usr ccache 1112749128 M * micah if you manage to do all this before this one compile finishes... 1112749276 M * Bertl well, memory should be done now ;) 1112749331 M * micah I have a feeling you could optimize my CPU out from under me 1112749496 M * Bertl ahh, icky icky ... prepare to test nfs too ;) 1112749551 M * micah ohhh boy 1112749615 M * Bertl well, I'll do my best to avoid bugs ;) 1112749923 M * micah its a good philosophy :) 1112750217 M * Bertl hmm, tagxid done ... 1112750298 M * micah I've got some ways to speed up compilation 1112750478 M * Bertl iunlink done ... 1112750508 M * albeiro micah: make a public howto from what you learned about speeding it up 1112750569 M * Bertl inode done ... 1112750615 M * micah albeiro: 1112750623 M * micah if you're using the official debian image packages, you can nuke the extra configs., and drop 1112750631 M * micah a lot of modules out of the remaining config 1112750631 M * micah so, rm config/{386,k7,k7-smp,686-smp} 1112750631 M * micah cp config/686 linux-2.6.11/.config, cd linux-2.6.11 1112750631 M * micah && make menuconfig; cp linix-2.6.11/.config 1112750631 M * micah config/686 1112750642 M * micah additionally, ccache will speed up compiles significantly 1112750709 M * micah the basic thing is to remove as many unneeded modules as possible 1112750766 M * Bertl utsname done ... 1112750881 M * Bertl sock done ... 1112750889 M * micah because it is really unneeded to compile x25 or irda if you dont have those 1112750906 M * Bertl ah, no, you need to compile x25 in 1112750924 M * Bertl x25 is one of the typical bug sources in recent vserver kernels 1112750941 M * micah ah! ok, I just pulled that one as an example to distract you 1112750945 M * Bertl (no kidding, I'm serious ;) 1112751103 M * micah I am seeing that the kernel compile is coming to an end 1112751124 M * micah who will be the winner... 1112751126 M * micah coder or compiler? 1112751145 M * Bertl net done ... 1112751147 M * micah well, I prefer slower careful coder :) 1112751174 M * Bertl no way to make haste here ... it's a careful procedure ;) 1112751420 M * Bertl proc done ... 1112751479 M * Bertl micah: ah, did I mention that you have to test-cross compile for other archs too? 1112751554 M * micah kernel compile done finally :P 1112751587 M * micah Bertl: I am not sure if I know how to do that, but I will investigate how and do it 1112751607 M * Bertl that's actually easy ... 1112751624 M * Bertl well, I guess I can do that for you ... 1112751639 M * micah I would like to learn how, especially if it can help you 1112751665 M * micah but of course, sometimes it is just easier to do it yourself 1112751704 M * Bertl http://vserver.13thfloor.at/Stuff/Cross/ 1112751716 M * Bertl read the howto 1112751731 M * micah ok 1112751780 M * Bertl but the next few steps will require that you compile, boot and test on your platform (probably several times, with small changes) 1112751809 M * micah ok, first I must make sure that this 2.6.11.6 1.9.5.12 reboots and if it makes that sudo problem go away 1112751825 M * Bertl it would be good too if you had a look at the various features (vserver paper and changelogs) to know what to test ... 1112751916 M * Bertl syscalls done ... 1112752045 M * micah here goes reboot 1112752069 M * Bertl base done ... 1112752404 M * micah machine rebooted ok :) 1112752411 M * Bertl excellent! 1112752452 M * micah now I must get the up-to-date tools on as well 1112752549 M * Bertl of course, best would be 0.30.205 + 2 patches 1112752679 M * micah ah, tell me where the patches are 1112752685 M * micah I was doing 0.30.205 1112752707 M * Bertl sec, let me check where I put them ... 1112752751 M * Bertl can you unpack rpms? 1112752756 M * micah yes 1112752758 M * Bertl http://vserver.13thfloor.at/Experimental/MDK/util-vserver-0.30.205-3mdk.src.rpm 1112752771 M * Bertl should contain both patches 1112752782 M * micah I have a utility called "alien" 1112752785 M * micah that converts rpms :) 1112752817 M * Bertl yeah, I remember -- used it once on a deb ;) 1112752824 M * micah hehe 1112752862 M * micah actually 1112752865 M * micah I was using .204 1112752871 M * micah .205 is not on 13th floor 1112752881 M * Bertl not as releas, yes ... 1112752886 M * Bertl +e 1112752919 M * micah if I am to use it, where can I get it? 1112752941 M * Bertl it's in the rpm too 1112752956 M * Bertl http://www.13thfloor.at/~ensc/util-vserver/files/alpha/ 1112752966 M * micah ah ok 1112752984 M * Bertl limit done ... 1112753050 Q * ciphernaut Read error: Connection reset by peer 1112753064 M * micah I assume the two patches are the dynfix and vpsfix ones 1112753153 M * Bertl yep 1112753274 M * Bertl note: check ext2/ext3 xattr with disk limit accounting too 1112753375 M * micah got around the dietlib and beecrypt hurdles 1112753391 M * Bertl hopefully by installing that stuff ;) 1112753394 M * micah yes :) 1112753422 M * Bertl hmm, the ext3 xattr/dquot stuff is broken in debian 1112753449 M * Bertl submitted a patch a few months ago, has been included in mainline since ... 1112753538 M * Bertl let's see if I can find the fix in my outbox ;) 1112753564 J * ciphernaut ~a@61.88.18.130 1112753580 M * Bertl yep, so we'll add that one to debian ;) 1112753737 M * micah it was not fixed in debian? 1112753754 M * micah I could file a bug report against the kernel with the fix and they'll accept it 1112753839 M * Bertl http://lkml.org/lkml/2005/1/26/174 1112754148 M * micah hmm, I did ./configure --prefix=/usr --with-initrddir=/etc/init.d --with-vrootdir=/home/vservers --sysconfdir=/etc 1112754155 M * micah but it is not finding my vserver configs in /etc/vservers 1112754206 M * micah ah, these are legacy configs 1112754227 M * Bertl well, should work too ... 1112754231 M * micah hmm, is there an easy way to convert them? 1112754242 M * Bertl IIRC, somebody posted a script 1112754402 M * Bertl okay, dlimit done ... 1112754408 A * micah searches lsit 1112754410 M * micah list 1112754488 M * Bertl debug done ... 1112754547 M * micah hmm 1112754575 M * Bertl legacy done ... 1112754626 M * Bertl vshelper done ... 1112754636 M * micah I see it is a pending task for the tool to create converting script 1112754684 M * micah the configurations are easy to configure by hand if I can create the directory structure properly 1112754688 M * Bertl now to the last one .. the includes ;) 1112754702 M * Bertl micah: you can use the skeleton build method 1112754741 M * micah Bertl: that just creates the /etc/vservers directories and doesn't touch the /vserver stuff right? 1112754761 M * Bertl no, it also makes an empty dir in /vserver 1112754774 M * Bertl but it also makes a backup of your old files 1112754858 M * micah yes I am testing that now 1112754877 M * complexho Night guys - happy hacking :) BTW Bertl I've started documenting what I'm doing on a wiki. It's still quite incomplete but you can see it @ www.serverguru.co.uk 1112754892 M * Bertl complexho: night, and thanks! 1112754912 M * micah looks like the old version created more things in /dev than the new one 1112754931 M * Bertl old version meaning the old debian version, right? 1112754939 M * micah yes I suppose that is correct 1112754947 M * micah this is easy to convert 1112755171 A * complexho is AFK, sleep is a thing of beauty —I-n-v-i-s-i-o-n— 1112755364 M * Bertl okay, I guess I'm done with adapting ... now to the reassembling and first testrun ... 1112755499 M * micah 3 vservers converted, three left 1112755666 M * Bertl okay, I have to get something to eat, brb ... 1112755693 M * micah hm weird, this vserver complains about mounting on start 1112755693 M * micah /etc/vservers/margarita/fstab:3:1: failed to mount fstab-entry 1112755699 M * micah /etc/vservers/margarita/fstab:1:1: failed to mount fstab-entry 1112755789 M * micah it seems to work fine though 1112755829 M * micah vservers converted 1112755874 M * micah that might be because it was started before, I bet if I restart it will be fine 1112755925 M * micah and it looks like the sudo problem went away :) 1112756318 J * Filipok pholmes@c-24-1-185-209.hsd1.tx.comcast.net 1112756324 M * Filipok greetings 1112756344 M * micah greetings Filipok! 1112756357 M * Filipok zdrastya micah 1112756414 M * Filipok any python experts hanging around? 1112756677 M * Bertl back now ... 1112756703 M * micah welcome back! 1112756731 M * Bertl hey Filipok! isn't that the wrong channel for python? 1112756819 M * Filipok ya.. getting desperate 1112756861 M * Bertl i.c. 1112757145 M * Bertl micah: okay, now the first test build ... 1112757153 M * Bertl I expect a lot of issues ;) 1112757526 M * micah hehe 1112758812 M * Bertl ah, debian is missing another patch ;) 1112758883 M * Bertl one which actually fixes a funny bug ... 1112758892 M * Bertl http://www.ussg.iu.edu/hypermail/linux/kernel/0501.3/1026.html 1112759145 P * Filipok 1112759995 M * Bertl micah: still awake? 1112760936 M * Bertl lol, debian has atomic_inc_return for all archs, except x86 ;) 1112761137 M * micah Bertl: I am, I had to deal with an emergecny :( 1112761160 M * Bertl np, emergency handled now? 1112761217 M * micah almost finished 1112761366 M * micah need to file those bugs 1112761411 M * Bertl well, the last one isn't really a 'bug' because the kernel does not need inc_and_return() 1112761842 M * Bertl had to add some rcu stuff too ... 1112761883 M * Bertl but nothing critical so far ... 1112761945 M * micah i'm not sure their reasoning for not including some of those things, I think they try to backport minimal things, but critical things (such as security fixes) 1112761969 M * Bertl yes, obviously they missed some stuff ;) 1112761990 M * micah security stuff? 1112762003 M * micah would like to point those out to them so they can get them fixed! 1112762016 M * Bertl well, let's call it bugs ... 1112762027 M * micah I've been working on going through all acknowledged CANs for kernels to make sure those are all taken care of 1112762030 M * Bertl it's probably not a security issue if your kernel crashes, right? 1112762037 M * micah hehe 1112762070 M * micah I think security and crashing things they try to integrate 1112762123 M * Bertl well, the assert issue will crash your kernel when compiled with up/spinlock_debug ;) 1112762162 M * Bertl the ext3 xattr thingy just gets quota wrong ... 1112762300 A * micah notes these to bring to them 1112762361 M * Bertl note: task_state (in proc) has to be checked too 1112762377 M * Bertl (regarding initpid virtualization) 1112762635 A * micah makes the note 1112762705 M * Bertl note: virtual process uptime probably needs some fixing 1112763072 M * Bertl compile looks good so far ... 1112763630 M * Bertl oaky, first compile didn't boot in qemu ;) 1112763643 M * micah but it compiled! 1112763671 M * Bertl yeah, it did .. probably some issues with the split, I'm trying my typical qemu .config on that now 1112763785 M * micah I have had problems sometimes getting qemu to boot, second try sometimes works, I dont know why 1112763811 M * Bertl no, either it works here, or it doesn't ... the qemu-fast I use is pretty consistent ;) 1112763852 M * Bertl (and I guess I already found the issue) 1112764131 M * Bertl hmm, no wonder the debian kernels are _always_ bloated ... 1112764153 M * micah too many modules 1112764171 M * Bertl no, absolutely ridiculous dependancies ;) 1112764186 N * Doener|gone Doener 1112764189 M * Doener morning! 1112764193 M * Bertl morning Doener! 1112764253 M * Doener hey, timo seems to make progress :) at least the last mail looked somewhat better than the previous ones 1112764277 M * Bertl yeah, looks like it! 1112764329 M * Bertl ahem, I guess some debian person can explain how this happens: 1112764338 M * Bertl grep DVB .config 1112764338 M * Bertl # CONFIG_DVB is not set 1112764344 M * Bertl LD drivers/media/dvb/b2c2/built-in.o 1112764344 M * Bertl LD drivers/media/dvb/bt8xx/built-in.o 1112764344 M * Bertl LD drivers/media/dvb/dvb-core/built-in.o 1112764344 M * Bertl LD drivers/media/dvb/frontends/built-in.o 1112764349 M * Bertl make 1112764408 M * micah you mean, why does it compile dvb modules when it is not set in the .config? 1112764417 M * Bertl because I'm pretty sure that this doesn't happen with 2.6.8 (mainline) 1112764672 M * micah in my debian config it has CONFIG_DVB=y and CONFIG_DVB_CORE=m 1112764681 M * Doener hm, let's see... i got a debian box around... (i.e. i abuse my sister's box *g*) 1112764718 M * Bertl hmm, won't boot either ... 1112764761 M * Bertl I'm uploading the current state, as I consider it 'not so back' maybe debian folks managed to break the boot code ... somehow ... 1112764768 M * micah bertl: do you have an initrd? 1112764771 M * Bertl s/back/bad/ 1112764955 M * micah Bertl: because I believe to get it to boot you need the initrd I think 1112764971 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.8-15-vs1.9.5-1.diff 1112764992 M * Bertl well, I don't need an initrd for mainline so why for debian ... 1112765016 M * Bertl Doener: seems to lock up pretty hard on qemu/boot 1112765063 M * micah Bertl: well if you used the debian config, with all the modules I think you do 1112765072 M * micah Bertl: unless you made your own kernel config 1112765083 M * Doener micah: CONFIG_DVB would be on then ;) 1112765142 M * micah hmm 1112765196 M * Doener compiling now... will take some time, debian config includes kitchensink and more 1112765212 M * Doener (of course i disabled CONFIG_DVB ;) 1112765517 M * Bertl hmm, they do not have the FIXADDR mapping ... 1112765706 M * micah FIXADDR mapping came in a later kernel? 1112765723 M * Bertl hmm .. not sure ... somehow it seems 'missing' 1112765928 M * micah unfortunately I need to go :( still have things to clean up here from problems earlier, I'll be back tomorrow, I'll attempt to compile the patch you posted and make a summary of the notes and missing patchexs 1112765951 M * Bertl okay, probably will upload a -2 version before I go to bed ... 1112765973 M * micah ok, I'll look for a newer one 1112765998 M * Bertl so cya later then ... 1112765999 M * micah ok, goodnight, good day everyone! 1112766014 M * Doener night Bertl! 1112766038 M * Bertl ahem, yeah, but will take a few more minutes ;) 1112766047 M * Doener heh 1112766172 M * Doener # cat built-in.o 1112766172 M * Doener ! 1112766224 M * Doener that's what those files look like... 1112766255 M * Bertl hmm .. no idea what you are talking about ;) 1112766270 M * Doener 07:12:18 Bertl # CONFIG_DVB is not set 1112766271 M * Doener 07:12:24 Bertl LD drivers/media/dvb/b2c2/built-in.o 1112766293 M * Doener they look pretty 'empty' to me 1112766328 M * Bertl aha, okay, so why are the linked in at all? 1112766457 M * Bertl hmm hmm .. I remember some timer issues now ...which might be of interest here ... 1112766655 M * Bertl interesting, how did ACPI and PM get selected ... 1112766815 M * Doener Bertl: those seem to be unconditional... i get them with vanilla, too 1112766847 M * Bertl grep ACPI .config 1112766847 M * Bertl # Power management options (ACPI, APM) 1112766847 M * Bertl # ACPI (Advanced Configuration and Power Interface) Support 1112766847 M * Bertl # CONFIG_ACPI is not set 1112766847 M * Doener (the dvb built-in.o's) 1112766860 M * Bertl ah, okay ;) 1112766881 M * Bertl funny, I don't remember seeing them with 2.6.11 1112766917 M * Doener i get them with a 2.6.11-vs1.9.5-clrsid-qemu build 1112766929 M * Doener s/11/11.5/ even 1112766944 M * Bertl okay, probably I just ignored them ... 1112767648 M * Bertl ah, after disabling power management it boots fine ;) 1112767709 M * Bertl Linux-VServer Test [V0.11] (C) 2003-2005 H.Poetzl 1112767709 M * Bertl chcontext is working. 1112767709 M * Bertl chbind is working. 1112767709 M * Bertl Linux 2.6.8-vs1.9.5 i686/0.30.204/0.30.204 [Ea] 1112767709 M * Bertl VCI: 0001:0025 273 03110076 1112767711 M * Bertl --- 1112767714 M * Bertl [001]# succeeded. 1112767716 M * Bertl [011]# succeeded. 1112767719 M * Bertl [031]# succeeded. 1112767721 M * Bertl [101]# succeeded. 1112767724 M * Bertl [102]# succeeded. 1112767726 M * Bertl [201]# succeeded. 1112767729 M * Bertl [202]# succeeded. 1112767828 M * Bertl now trying with different split ... 1112767837 M * Bertl (different from default, that is) 1112767942 Q * Elchbulle Server closed connection 1112768264 M * Bertl okay, the split is broken somehow ... 1112768285 M * Bertl enough for me for today ... I'll upload the latest version and be gone ... 1112768305 M * Doener ok, good night then! 1112768443 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.8-15-vs1.9.5-1.1.diff 1112768470 M * Bertl and a good day to you Doener! 1112768482 M * Bertl cya all later ... 1112768505 N * Bertl Bertl_zZ 1112768729 J * AaronWL aaronwl@2002:44e4:80::dead:1337 1112769025 M * ciphernaut later Bertl 1112769571 M * maharaja hi there 1112771841 P * AaronWL 1112772107 Q * eyck Read error: Operation timed out 1112772141 J * erwan_ho ~erwan@lns-vlq-39f-81-56-133-136.adsl.proxad.net 1112773079 Q * erwan_ho Remote host closed the connection 1112773747 J * _BWare_ ~bware@212.26.196.195 1112773747 Q * BWare Read error: Connection reset by peer 1112776390 J * jsambrook ~jsambrook@aelfric.plus.com 1112777327 Q * Vudumen Read error: No route to host 1112778626 J * prae ~prae@134.106-14-84.ripe.coltfrance.com 1112779708 Q * alexx Quit: Bye 1112782658 P * jsambrook 1112784547 J * Schak ~abc@147.172.172.194 1112784566 J * Vudumen vudumen@perverz.hu 1112784664 M * Schak hi, i have a problem. since my host server was rebooted, i cannot start the vservers... before this reboot, everything works perfectly... following error: 1112784664 M * Schak kara:/# vserver --verbose v1 start 1112784664 M * Schak open(): No such file or directory 1112784664 M * Schak ipv4root is now 193.201.53.xx 1112784664 M * Schak secure-mount: chdir("/proc"): No such file or directory 1112784734 M * Schak (and yes, this time i executed /etc/init.d/vprocunhide start) :) 1112784821 M * Doener try with --debug instead of --verbose 1112784983 J * hvd ~takeagues@fw-grz.hollomey.com 1112785020 Q * Schak Quit: 1112785039 J * Schak ~abc@147.172.172.194 1112785133 M * hvd well sounds crazy, but i yust successfull software suspended (swsusp2) 2.6.11-vs1.9.5 on my mobile demonstration system running 5 vservers (oracle db included) .. this saves me a lot of time .. thx bertl ;-) 1112785456 M * Schak Doener: yes thx. it was my fault again... *sorrows* ... after reboot one should not to forget to mount the vserver-filesystems in their directories... %) 1112785470 M * Doener heh 1112787197 J * BWare ~bware@212.26.196.195 1112787197 Q * _BWare_ Read error: Connection reset by peer 1112788472 M * rs Doener: are you around ? 1112788476 M * Doener yep 1112788487 M * rs hi :) 1112788492 M * Doener hi rs :) 1112788517 M * rs is the clrsid-feat patch the fix for the odd issue with unterminated process at vserver stop ? 1112788621 M * Doener no, that's for http://archives.linux-vserver.org/200504/0029.html 1112788646 M * Doener what issue are you talking about btw? i don't remember having heard about such a problem... 1112788680 M * rs some time when someone stop it's vserver, there is one process left in the context (we are using fakeinit) 1112788719 M * rs it's often somthing like a mysqld zombie or something like that 1112788741 M * rs if you kill -9 the process the context is released, but util-vserver seems not able to do it during stop process 1112788761 M * Doener hm, zombies should be a dead end IIRC 1112788806 M * rs yeah it's pretty strange 1112788827 M * Doener but maybe that was fixed and i didn't take notice... a zombie without a parent in its context can't be cleaned out by init because it's not visible to init 1112788857 M * rs yeah I understand that, but why a kill -9 can clean it ? 1112788921 M * Doener that's a good question... in the past that wasn't possible... i'll take a look at that 1112788981 M * rs I have an example in my term history 1112788995 M * Doener but for the clrsid patch: that should solve the problem that a init-based vserver isn't restarted when you use 'reboot' from inside the vserver, because init gets killed and never initiates the real reboot 1112789030 M * rs # vserver 84.244.3.xx status 1112789031 M * rs Vserver '84.244.3.xx' is running at context '890' 1112789031 M * rs Number of processes: 14 1112789031 M * rs Uptime: 1 days, 06:53 1112789031 M * rs # vserver 84.244.3.xx exec ps aux 1112789033 M * rs USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND 1112789035 M * rs mysql 7516 0.0 0.0 0 0 ? Z Mar29 0:06 [mysqld] 1112789038 M * rs root 17892 8.0 0.3 2560 844 pts/20 R+ 11:37 0:00 ps aux 1112789041 M * rs # vserver 84.244.3.xx exec kill -9 7516 1112789063 M * rs then the context is released 1112789076 M * rs I don't have the proof of than in the history, but trust me :) 1112789081 M * rs than/that 1112789120 M * rs it's often happen with a mysqld process, I don't know why 1112789247 M * rs I have a reject with the clrsid-feat03 patch by the way 1112789271 M * rs I'm trying to apply it on a vs1.9.5 with the unhold-fix01 patch 1112789431 M * Doener the clrsid alone doesn't fix the problem yet 1112789440 M * rs oh 1112789469 M * rs is there something to add to util-vserver ? 1112789493 M * Doener although the sid/pgrp is set to 0, init is still killed 1112789559 M * Doener i don't remember which patches are needed to get the init problem done, and then you'd still have to patch util-vserver (with a not yet done patch ;) 1112789616 M * rs ok, maybe I could help to do the util-vserver patch? 1112789627 M * rs or enrico is already on it? 1112789682 M * Doener i'm not yet sure if the patches do the right thing, been a little tired when we did the last few things... basically clrsid is the result of testing 1 of 3 possibilities how to do it 1112789704 M * Doener don't know what Bertl thinks about the current approach 1112789791 M * rs he said me that it would be a good idea to test it, but I can't test it if it's not meant to work :) 1112789834 M * Doener hmm... well then... 1112790007 M * Doener you need the initvx-clean (not the fix) to have it apply 1112790059 M * hvd any idea where vserver-stat gets server name and description from ? 1112790117 M * rs where is it ? 1112790129 M * Doener in FOR-1.9.6 1112790132 M * rs ok 1112790311 M * Doener hvd: from the vserver configuration 1112790732 M * Doener rs: http://vserver.13thfloor.at/Experimental/delta-vkill.diff 1112790758 M * Doener that's a patch that protects fake inits from user signals and makes it killable only via vkill 1112790793 M * Doener that should make the reboot from inside the vserver work, but needs the tools to be patched, so that they use vkill to stop the vserver 1112790837 M * rs k 1112791081 Q * Schak Ping timeout: 480 seconds 1112795242 N * Bertl_zZ Bertl 1112795259 M * Bertl morning folks! 1112795303 M * erwan_taf lo berni 1112795305 M * erwan_taf grrrrrrrr 1112795308 M * erwan_taf lo Bertl 1112795381 M * Bertl hey erwan! 1112795401 M * Bertl hvd: heh, never tried that, good to hear that it works ;) 1112795540 J * SERKAN55 ~best29@212.156.186.118 1112795645 M * rs hey bertl! 1112795648 M * rs how are you ? 1112795653 M * Bertl welcome SERKAN55! 1112795669 M * Bertl rs: thanks fine ... just got up, but I'm fine ... 1112795678 M * rs hehe :) 1112795686 M * Bertl rs: looks like you're working on linux-vserver again ;) 1112795721 M * rs yeah, I try at least 1112795777 M * Bertl btw, do you by any chance have a list of things you needed to change/patch when you adapted the various distros ... (I mean things which linux-vserver couldn't handle back then ;) 1112795817 M * rs hmm yeah sure 1112795853 M * rs the biggest is the bind issue, that needs us to "port" a package for each distrib 1112795877 M * Bertl because one thing on my todo list is to reduce those issues to a minimum ... (bind should work by now without patching ;) 1112795919 M * rs yeah I know that, and it's a great improvement 1112795924 J * wurd ~kvlt@modemcable143.235-201-24.mc.videotron.ca 1112795927 M * rs so second thing is all network stuff 1112795941 M * rs but that will be fixed with NGNET 1112795947 M * Bertl welcome wurd! 1112795958 M * Bertl rs: precisely ... 1112795971 M * wurd thank you 1112795992 M * rs we just don't setup any network thing on the distribution 1112796019 M * Bertl yeah, anything else? 1112796042 M * rs nothing for network 1112796061 M * rs then we replace the sysinit entry in inittab by a script of our own 1112796104 M * rs in that script we port all necessary things that is done by sysinit in each distribution, all thing that don't deal with hardware 1112796206 M * rs it's mainly hostname configuration, /tmp cleaning and under debian /var/lock/subsys/ cleaning etc 1112796219 M * hvd morning bertl.. 1112796221 M * Bertl okay, so I take it that except for the bind (which should be worked around) and the network stuff, which will be covered by ngnet, no packages required modification, right? 1112796248 Q * SERKAN55 Quit: [ YûzûK ScripT -3- ] [ DownLoad: www.Yuzuk.Net ] [ Irc.Yuzuk.Net ] [ ßy PRoFeSSioNaL ] 1112796251 M * rs the makedev package under debian fail miserabily but it's not a big deal 1112796272 M * rs it just issue some warning and it's happening only when you want to upgrade a woody into sarge 1112796273 M * Bertl hvd: thanks! and a good afternoon to you (probably)! 1112796338 M * Bertl rs: hmm, yeah, makedev .. good point, but I guess that's beyond reach (atm) 1112796375 M * rs we also have to place the env var LD_ASSUME_KERNEL=2.2.5 1112796382 M * rs on redhats 1112796392 M * Bertl which ones? 1112796395 M * rs redhat 9.0 actually 1112796410 M * Bertl hmm, did you try without recently? 1112796416 M * hvd mhh as 3.0 seems to work fine here 1112796416 M * rs no 1112796452 M * hvd ok oracle wants me to set LD_ASSUME_KERNEL=2.2.5 but this is not a vserver problem ... the rest does fine without ... 1112796468 M * rs we also have to patch the syslog-ng package under debian to not try to read the kmsg proc entry 1112796514 M * Bertl ah, that should be solved too ... /proc/kmsg is virtualized in recent releases ... 1112796520 M * rs we replace the /dev/tty0 with a plain file too 1112796533 M * Bertl hvd: yes, IIRC it is/was NPTL related ... 1112796537 M * rs to fix some program that segfault if this file isn't present 1112796566 M * Bertl ah, good point, maybe we should add a virtual console to be logged/read on the host? 1112796621 M * rs yeah would be nice 1112796636 M * hvd syslog-ng does fine here on sarge with destination dp_xconsole { pipe(""); }; 1112796699 M * rs seems to be only "patch" we have to apply on images to work 1112796732 M * rs other things are just to make the image smaller 1112796820 M * hvd tomcat was some kind of problematic .. because it only accepts commands on 127.0.0.1 by default .. what be fine if they would do a gethostbyname for localhost first .. but i cannot complain .. sending killsignals to tomcat works too ;-) 1112796828 M * rs I'd like to test the initpid fix 1112796876 M * Bertl well, go ahead ... 1112796882 M * rs I compiled a 2.6.11.6-vs1.9.5 with unhold, initvx-clean01 and clrsid-feat03 patches 1112796903 M * rs but doener said that it needs an util-vserver patch too 1112796915 M * Bertl yep, didn't he provide it? 1112796922 M * rs no :) 1112796930 M * Doener Bertl: did we do one yet? 1112796935 M * rs he said that patch doesn't exist yet 1112796944 M * Bertl Doener: morning! 1112796951 M * rs re Doener 1112796960 M * Doener rs: you only need the patch if you got vkill patch, too 1112796968 M * Bertl Doener: well, you did one (in private of course ;) 1112796992 M * Doener without the vkill patch the behaviour with reboot doesn't change... i don't know why, yet 1112796999 M * Doener i did? 1112797022 M * rs yeah ok, but my concern is to fix the reboot bug :) 1112797036 M * Bertl simple, because without the vkill you cannot kill v-init 1112797045 M * rs because it generate 90% of or customers complain =) 1112797076 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-1.9.6/delta-vkill-feat01.diff 1112797084 M * rs 10 other % are guys that didn't understand what was a vserver, so I'd say that with this fixed, the linux-vserver product would be almost bugfree =) 1112797089 M * Doener Bertl: yeah, but from the information on the killall5 man page in deduced that init shouldn't be attempted to be killed in the first place... 1112797114 M * Bertl rs: and don't forget http://vserver.13thfloor.at/Experimental/FOR-1.9.6/delta-initvx-fix01.diff 1112797130 M * rs yeah I already applyed it 1112797136 M * Bertl Doener: yes, but the tools cannot kill it either (from outside) 1112797144 M * Doener rs: that reboot from inside only stops the vserver, or the bug with the process hanging around? 1112797155 M * rs Doener: both 1112797198 M * Doener Bertl: that's with the vkill patch. i'm talking about clrsid only... killall5 should not kill init, but the tools should be able to do so... 1112797334 M * Bertl yes, and that didn't work? 1112797370 Q * wurd Quit: brb 1112797416 M * Doener no, with clrsid only, on reboot init was still killed 1112797434 M * Bertl btw, my killall5 manpage says different 1112797457 M * Bertl killall5 is the SystemV killall command. It sends a signal to all processes except the processes 1112797460 M * Bertl in its own session, so it won't kill the shell that is running the script it was called from. 1112797513 M * Doener yes, and i expected the script to be in init's session if i issue reboot 1112797528 J * wurd ~kvlt@modemcable143.235-201-24.mc.videotron.ca 1112797546 M * Bertl Doener: well, we don#t know that, right? 1112797553 M * Doener right... 1112797592 M * Bertl so a printk() when the signal is sent might sched some light on it 1112797642 M * Doener hm, using the test program on the shell running the script that calls killall5 should also do 1112797712 A * Doener will do that later... got to go shopping now 1112797714 M * Doener cya later! 1112797717 N * Doener Doener|gone 1112797742 M * Bertl hmm, yeokay, but the tool patch? 1112797751 M * Doener|gone i didn't do one... 1112797774 M * Doener|gone IIRC i went to bed shortly after the vkill patch was done 1112797788 M * Bertl but you tested it, right? 1112797813 M * Bertl anyway, we'll figure it ... have fun! 1112797813 M * Doener|gone just the part that init is killable via vkill 1112797873 J * eyck eyck@81.219.64.71 1112797881 M * Bertl rs: the only required change is to issue a vkill to pid=1 when the reboot helper is called (to terminate the init) 1112797885 M * Bertl welcome eyck! 1112797895 M * Bertl wb wurd! 1112797909 M * rs from where ? 1112797925 M * Bertl from the vshelper 1112797932 M * rs ok 1112797950 M * Bertl basically send a kill, and loop until the context goes away 1112797981 M * rs with some sleep I guess :) 1112798000 M * Bertl yes, there is a syscall for that purpose, but it's not used by enricos tools yet ... 1112798059 M * Bertl vserver-info RUNNING (or so should give info) 1112798146 M * Bertl well, no it doesn't either ... 1112798165 Q * eyck Quit: 1112798189 J * eyck eyck@81.219.64.71 1112798249 M * Bertl rs: guess the best option is to check for the /proc/virtual/ entry 1112798261 M * Bertl something like: 1112798284 M * Bertl while [ -d /proc/virtual/$XID ]; do sleep 0.5; done 1112798312 M * rs k 1112798324 M * rs I guess I can put this in an external vshelper script 1112798376 M * rs or perhaps patch vserver.stop script 1112798395 Q * Doener|gone Quit: changing servers 1112798411 M * Bertl rs: whatever you consider appropriate 1112798427 M * rs there is already something related there: 1112798429 M * rs if test -n "$_IS_FAKEINIT" && \ 1112798429 M * rs $_VSERVER_INFO - FEATURE vkill && \ 1112798429 M * rs initpid=$($_VSERVER_INFO "$S_CONTEXT" INITPID 2>/dev/null); then 1112798429 M * rs $_VKILL -s INT --xid "$S_CONTEXT" -- 1 "$initpid" || fail=1 1112798431 M * rs ## HACK: remove the 'initpid' stuff above when PID virtualization 1112798433 M * rs ## is implemented 1112798491 M * rs I just have to add the wait loop then? 1112798493 M * Bertl hmm ... interesting ... 1112798509 M * rs it's a patch added recently (not in the last 205 release) 1112798515 M * Bertl this is latest tools I guess? 1112798520 M * rs cvs 1112798522 M * Bertl yeah, okay, should work then ;) 1112798583 M * Bertl hmm, guess I really have to talk with enrico about that 'HACK' ;) 1112798673 J * Doener|gone doener@193.24.208.125 1112798681 M * rs so is it needed to add the loop ? 1112798710 M * Bertl the loop can't hurt .. and it ensures that the context goes away ... 1112798724 M * Bertl sometimes sockets stay around for a few seconds ... 1112798745 M * ntrs Bertl, by the way we also have to add "export LD_ASSUME_KERNEL=2.2.5" to all fedora servers 1112798747 M * rs yeah but if it doesn't I don't want infinit loops ;) 1112798784 M * Bertl ntrs: with 2.6.11* kernels? are you sure? 1112798796 M * rs ntrs: we don't need it for fedoras here 1112798799 M * Bertl rs: if the context doesn't go away, you have an issue anyways, right? 1112798835 M * rs Bertl: yeah but I will have more issues with a stop commends that never returns :) 1112798841 M * Bertl you might consider adding a nifty $[n++] timeout counter and mail a warning after that ... 1112798843 M * rs command 1112798861 M * ntrs hmmm, what happened? it really does work fine with 2.6.11 1112798869 M * Bertl ntrs: ;) 1112798884 M * rs ntrs: really ? 1112798894 M * ntrs I only tried now with FC# 1112798896 M * ntrs FC3 1112798900 M * rs that's really strange, we never got the issue with FC 1112798902 M * ntrs let me try with FC2 1112798930 M * Bertl rs: '_does_' work fine 1112798936 M * Bertl ;) 1112798949 M * ntrs how do we make bind work fine in vservers without recompiling? 1112798954 M * rs Bertl: haha :) 1112799059 M * Bertl ntrs: sec 1112799066 M * ntrs By the way, cfq and deadline io schedulers make no difference in the IO load or the overall latency of the server while under heavy disk IO load 1112799082 M * Bertl interesting ... 1112799118 M * ntrs especially when doing a lot of swap IO 1112799128 M * Bertl ad bind ... should work with VXC_SET_RLIMIT 1112799140 M * Bertl #define VXC_SET_RLIMIT0x00000002 1112799158 M * ntrs Do I need to add this to the patch? 1112799169 M * ntrs Or it's already there? 1112799228 M * Bertl it's already there, basically you just have to make sure the cap is set, but it's not 100% perfect yet (it gives too much permissions for vserver users) 1112799250 M * ntrs ok. I'll use it when it becomes perfect then. :) 1112799296 M * Bertl okay :) 1112799309 M * rs which one ? 1112799320 M * ntrs one more thing... 1112799324 M * Bertl well, I'm off for now .. back in a few hours ... 1112799338 M * Bertl ntrs: yep? 1112799346 M * ntrs on fedora updating the filesystem package gives this 1112799348 M * ntrs Transaction Listing: 1112799348 M * ntrs Update: filesystem.i386 0:2.3.0-1 - base 1112799348 M * ntrs Total download size: 15 k 1112799348 M * ntrs Is this ok [y/N]: y 1112799348 M * ntrs Downloading Packages: 1112799349 M * ntrs Running Transaction Test 1112799353 M * ntrs error: failed to stat /proc/bus/usb: No such file or directory 1112799355 M * ntrs Finished Transaction Test 1112799357 M * ntrs Transaction Test Succeeded 1112799359 M * ntrs Running Transaction 1112799361 M * ntrs error: failed to stat /proc/bus/usb: No such file or directory 1112799363 M * ntrs Updating: filesystem 100 % done 1/2 1112799365 M * ntrs error: unpacking of archive failed on file /proc: cpio: chown 1112799367 M * ntrs Updated: filesystem.i386 0:2.3.0-1 1112799369 M * ntrs Complete! 1112799373 M * ntrs so, I guess we'll have to exclude that one in yum.conf on all vservers 1112799390 M * ntrs unless there is aworkaround. 1112799398 M * Bertl well, it doesn't make to much sense to actually update it 1112799412 M * ntrs right. but it complains and people are asking questions. 1112799418 M * Bertl but you can update the database and ignore the install 1112799434 M * ntrs yes, that's what we normaly tell people. 1112799461 M * Bertl both issues are a weird behaviour of the package ... 1112799492 M * Bertl a) you cannot assume that /proc/bus/usb is mounted (don't mount it!) 1112799516 M * Bertl b) you cannot assume that /proc is mounted rw ... 1112799553 M * Bertl if you don't mount /proc/bus/usb and mount /proc ro on a normal fc, the update will probably give the same issues ... 1112799597 M * Bertl okay, really off for now ... 1112799601 N * Bertl Bertl_oO 1112799736 M * Snow-Man Bertl_oO: You didn't break networking stuff in 1.9.5, did you? :) 1112800344 M * wurd in order to install vserver, i need to build a vanilla kernel, right ? 1112800429 M * Snow-Man No, you need a vserver-patched kernel. 1112800452 M * wurd i was told i *HAD* to build a vanilla kernel.. 1112800480 M * eyck re 1112800498 M * wurd where can i get this vserver-patched kernel, Snow-Man ? 1112800507 M * wurd http://www.13thfloor.at/vserver/d_rel26/v1.9.5/ there ? 1112800516 M * Snow-Man Download the patch and patch the vanilla kernel. 1112800534 M * wurd Download the patch and patch the vanilla kernel. 1112800542 M * wurd do i need, or do i not need, a vanilla kernel ? 1112800555 M * wurd No, you need a vserver-patched kernel. 1112800564 M * wurd i dont know which sentence to follow between these two 1112800613 M * Snow-Man You need the source. 1112800623 M * Snow-Man Get the source to 2.6.11 1112800627 M * Snow-Man Then get the 2.6.11.6 patch 1112800630 M * wurd where 1112800632 M * Snow-Man Then get the vs1.9.5 patch 1112800634 M * Snow-Man kernel.org 1112800667 M * wurd http://www.isurf.lkams.kernel.org/pub/linux/kernel/v2.6/ <- are those vanilla kernels ? 1112800685 M * Snow-Man Probably. 1112800860 M * wurd so i only need linux-2.6.11.tar.bz2 and patch-2.6.11.6.bz2 ? 1112801255 M * wurd hello ? 1112801517 M * Snow-Man And the vs1.9.5 patch 1112802528 M * complexho •Snow-Man• I have been using vserver fine for about4 years now, only ever patching against the major releases (eg linux-2.6.11.tar.bz2). I keep seeing reference to these additional patches (eg patch-2.6.11.6.bz2) but have never applied any of these patches... 1112802554 M * complexho What I was wondering is, is it just a "latest greatest" thing, or is there some other added benefit? 1112802613 M * complexho to putting these patches in too... or is this something specific to 2.6.x[.y] 1112802661 M * complexho am I missing out on something by not applying these patches? 1112802667 M * Snow-Man God damn was that fucking ugly. 1112802691 M * Snow-Man The added benefit of the .y patches is improved stability. 1112802694 N * Doener|gone Doener 1112802710 M * Snow-Man Basically they're important bugfixes backported to the last 'stable' kernel.org release. 1112802719 M * Snow-Man security wise and stability wise. 1112802720 M * complexho ah right ok 1112802753 M * Doener that scheme was introduced just a few weeks ago, starting with 2.6.11.1 1112802758 M * micah Doener: did you have luck compiling on your sister's computer :) 1112802765 M * complexho funny thing kernels - on one hand we try to keep them up for 450+ days, on the other we have to get the last 3 days worth of patches in when building a new kernel. Quite funny really :) 1112802784 M * Doener micah: yep, and it also produced those built-in.o's, as does the vanilla kernel... 1112802797 M * complexho thanks Doener that's probably why I've only just noticed them... having only recently started working with 2.6.x 1112802805 M * Doener didn't do any further testing since she had a bunch of processes running i didn't want to kill 1112802841 M * micah makes sense 1112802881 N * kevinp|gone kevinp 1112802984 M * kevinp complexho, looks like you changed the location of the wiki 1112802985 A * complexho is AFK, sleep is a thing of beauty. Gone now for:«hrs 17mins» (Pager is On, /ctcp complexho Page ) —I-n-v-i-s-i-o-n— 1112803039 M * micah complexho: the ansi colors... they look terrible 1112803055 A * Doener likes xchat's color-stripping :) 1112803068 A * complexho is back from sleep is a thing of beauty. I was gone for hrs 18mins —I-n-v-i-s-i-o-n— 1112803112 M * Doener hm, guess i need away-announcement-stripping now ;) 1112803114 M * complexho sorry :) nothing worng with a little bit of colour, but I agree Invision is pretty colourful :) Now where did I put that knoppix disk... 1112803149 M * complexho hi kevin, yes, it's just www.serverguru.co.uk now (and it won't move again) 1112803163 M * kevinp found it, thanks 1112803380 J * Pazzo ~thomas@host130-250.pool8172.interbusiness.it 1112803387 M * Doener hi Pazzo 1112803394 M * Pazzo hi doener! 1112803399 M * Pazzo hi @ll! 1112803703 Q * complexho Read error: Connection reset by peer 1112804232 J * complexho ~complexmi@funk.gotadsl.co.uk 1112804267 M * complexho ok ok on x-chat now if that keeps the colours down ;) 1112804324 M * Doener xchat should be color-free by default ;) but what i meant, was that xchat can also remove all colors that are used by others 1112804341 M * Doener i.e. your away stuff showed up simply black-on-white here 1112804357 M * complexho :) So now I can give everyone else less headache as well.. 1112804532 M * kevinp So what is the difference between /etc/vservers/.distributions and /usr/lib/util-vserver/distributions? 1112804571 M * kevinp I just created a centos4 folder in /usr/lib/util-vserver/distributions based off of fc3, but now I see there isn't even a fc3 in /etc/vservers/.distributions 1112804689 M * kevinp I'm also planning on using yum instead of apt, but there are no examples of yum in there 1112804776 M * kevinp Looking at the flower page it just looks like it is a repeat of what I put in /usr/lib/util-vserver/distributions 1112804809 M * Doener kevinp: what _you_ put in there? i don't think you're supposed the change stuff there 1112804818 M * Doener configuration happens below /etc 1112804839 M * kevinp Yeah, I added a centos4 folder with the same structure as fc3, but with the centos yum repos 1112804861 M * kevinp and config 1112805238 M * kevinp I also want to have yum validate off of a gpg key, is this know to work inside a vserver? 1112806661 A * DaCa votes for chanmode +c 1112806924 M * Doener DaCa: you got to tell Bertl (once he's back) ;) 1112807453 M * albeiro +c ? 1112807465 M * albeiro right, colours ;p 1112808366 J * alexx ~alexx@82.225.136.176 1112808382 M * micah can someone help me understand how to fix Hunk offsets in patching? 1112808405 M * micah although they succeed, I want to make them not require the offset 1112808533 M * micah I know how to make the code changes, I just dont know the best way to re-work the diff 1112808550 M * micah because if I just change the line numbers in the diff, I think I will get errors 1112808615 M * daniel_hozac what makes you think that? 1112808712 M * Zoiah micah: why not just rediff after applying it? 1112808744 M * micah hmm I suppose if I do it right, I dont have a problem :) 1112809443 J * muadda ~muadda@gar31-1-82-234-50-167.fbx.proxad.net 1112810325 Q * prae Quit: Client exiting 1112810848 J * sim0n ~simon@212.76.255.43 1112810859 M * sim0n hi all! 1112810886 M * sim0n anybody know if there is a patch for kernel 2.6.11.6 ? 1112810904 M * Pazzo sim0n: patch for 2.6.11.5 should work 1112810916 M * sim0n i'll try thnx 1112810928 M * Doener sim0n: 1.9.5 should apply, if you want the latest and greatest development stuff look at http://vserver.13thfloor.at/Experimental 1112810997 M * sim0n oh no thnx... this is for a production server :-) 1112811043 M * Doener do you use the hardcpu scheduler? 1112811067 M * sim0n uhm haven't as of yet 1112811094 M * Doener ok, if you intend to do so, apply: http://vserver.13thfloor.at/Experimental/FOR-1.9.6/delta-unhold-fix01.diff 1112811104 M * sim0n will look at the new fancy stuff when i have this new machine running. my old vservers still run with very basic options 1112811114 M * sim0n ok thnx Doener 1112811148 M * Doener applying http://vserver.13thfloor.at/Experimental/FOR-1.9.6/delta-namei-fix01.diff is also a good idea 1112811159 M * Doener (independent of what features you use) 1112811236 M * sim0n hmm what patches from the experimental dir do you recommend me to apply? 1112811248 M * sim0n namei...unhold... 1112811263 M * Doener those two (from Experimental/FOR-1.9.6 1112811266 M * Doener ) 1112811277 M * sim0n yup...ok 1112811318 M * Pazzo Doener: namei & unhold - what will they do/change? 1112811377 M * sim0n is there anything special i should pay attention to, when configuring a kernel for a dual-opteron-64bit system ? 1112811413 M * sim0n AFAIK the kernel shouldn't be Preemptible.... is that still correct ? what else ? 1112811440 M * Doener namei: with namespaces, /foo is different from /../foo without the patch if you didn't chroot yet (which causes problems f.e. with yum if used from tools IIRC) 1112811470 M * Doener unhold: fixes some problem with the hardcpu scheduler and contexts that are put on hold (i.e. tokens used up), don't know the exact problem 1112811504 M * Doener sim0n: preemptible should be off (unless you build a kernel for a desktop system ;) 1112811518 Q * sim0n Quit: Client exiting 1112811529 J * sim0n ~simon@212.76.255.43 1112811532 M * sim0n oh ok 1112811547 M * sim0n does selinux work with vserver? 1112811585 M * Doener according to timo's problems with it, i'd say yes (you don't have to understand my thoughts here ;) 1112811612 M * Doener but _always_ enable linux capabilities, no matter waht 1112811616 M * Doener what even 1112811646 M * sim0n yeah i know...else you aren't able to do something, right ? 1112811653 M * sim0n ok 1112811702 M * Doener well, actually you're able to do way to much ;) (except dropping capabilities ;) 1112811743 M * Pazzo Doener: thnx! 1112811746 M * sim0n if there are no caps enabled? hmm 1112811763 M * Doener sim0n: yep, because that also disables the checks (at least IIRC) 1112811776 M * sim0n kay 1112814587 M * micah Bertl_oO: the patch-2.6.8-15-vs1.9.5-1.1.diff has a bunch of FAILED lines when I patch, as a handful of offset/fuzzes, I've fixed a bunch of them, but I'm wondering why I am getting these when you werent 1112814772 J * wfamy bill@lns-p19-16-idf-82-65-96-116.adsl.proxad.net 1112814777 M * wfamy hi 1112815926 M * wfamy i just install the debian tool but i can't use my last vserver. How can i recreate my config folder. i try vserver pg build -m skeleton --hostname pg --interface pg=eth0:192.168.0.103 but it does not run anyidea? 1112815955 M * Doener what does it say? 1112816223 M * wfamy ecure-mount: chdir("/tmp"): No such file or directory 1112816223 M * wfamy /etc/vservers/pg/fstab:2:1: failed to mount fstab-entry 1112816237 M * wfamy find: var/run: Aucun fichier ou répertoire de ce type 1112816237 M * wfamy fakerunlevel: open("/var/run/utmp"): No such file or directory 1112816263 M * Doener so, is there a /tmp directory in the vserver's filesystem? 1112816344 M * wfamy no tmp folder in /var/lib/vservers/ 1112816368 M * Doener hm, i'd look in /lar/lib/vservers/ ;) 1112816476 M * wfamy no; my vserver seem to be broker only dev etc and proc ! 1112816498 M * wfamy is it possible that the vserver build -m skeleton crash it? 1112816504 M * Doener well, you only build a skeleton, you need to fill in some data 1112816526 M * wfamy it was my old vserver 1112816547 M * Doener hm, normally the build script should refuse to build a vserver then IIRC 1112816565 M * wfamy IIRC? 1112816573 M * Doener if i remember/recall correctly 1112816583 M * wfamy ok 1112816644 M * wfamy is it possible to create a new config folder for my old vserver 1112816652 M * wfamy not the pg an other one 1112816673 M * Doener move the vserver out of the way, build a skeleton, put the vserver back in place 1112816708 M * wfamy ok i'll try this 1112817046 M * wfamy yep ! 1112817069 M * wfamy thanks 1112818436 M * wfamy see you bye 1112818439 Q * wfamy Quit: Leaving 1112819122 J * erwan_ho ~erwan@lns-vlq-39f-81-56-133-136.adsl.proxad.net 1112819688 Q * sebd Remote host closed the connection 1112820222 M * kevinp Has anyone built a new vserver from an ISO over a network? 1112820701 N * Bertl_oO Bertl 1112820711 M * Bertl evening folks! 1112820720 M * Pazzo hi bertl! 1112820751 M * Bertl hey Pazzo! 1112820781 M * Bertl micah: sounds strange, because it's a delta to the patch you gave me (and we verified with md5sum ;) 1112820805 M * Doener evening Bertl 1112820825 M * Bertl hey Doener! 1112820920 M * sim0n hi Bertl 1112820957 M * Bertl hey sim0n! LTNS! 1112820975 M * sim0n LTNS? 1112820989 M * sim0n (long time not seen?) 1112821007 M * Bertl yup 1112821021 M * sim0n :-) yeah... well i was quite busy 1112821046 M * sim0n got a new dual opteron server today... i'm currently setting it up 1112821062 M * Bertl great! 64bit? 1112821073 M * sim0n yup 1112821089 M * Bertl maybe you're interested in helping development out a little? 1112821091 M * sim0n i'm installing gentoo on it 1112821145 M * sim0n i can try to.... but once this beast is installed in the DC, i can't reboot it too often cos it's mainly a storage/backup server 1112821189 M * Bertl well, there are two issues with x86_64 atm 1112821205 M * Bertl a) dietlibc and util-vserver won#t play along 1112821221 M * Bertl b) the 32bit compatibility layer is not done (for the tools) 1112821253 M * Bertl if you're interested in improving either of that .. just let us know ... 1112821348 M * sim0n i won't be able to do coding...if that's what you need ? 1112821380 M * Bertl no, actually it's not the coding part, it's more the testing and figuring stuff out part .. because we have no x86_64 ;) 1112821550 M * Bertl micah: btw, excellent report! congrats! 1112821631 M * sim0n what would need me to figure out :-) ? 1112821653 M * sim0n maybe i could give shell access, if that would help you too ? 1112821669 Q * Loki|muh_ Ping timeout: 480 seconds 1112821992 M * Bertl sim0n: things like trying stuff out, especially kernel patches 1112822019 M * Bertl userspace (i.e. account might help too, not for me, more for enrico I guess) 1112822091 M * sim0n Bertl: trying stuff out should be ok... kernel patches, will be a bit difficult because i can't reboot the machine too often, and maybe risc to move to the DC 1112822118 M * Bertl ah, well, was just asking, no problem ... 1112822144 M * sim0n sorry :-/ 1112822155 M * sim0n i'll mail enrico.. 1112822196 M * sim0n Bertl: do you have his mail address ? 1112822236 M * sim0n Bertl: should be Enrico Scholz ? 1112822252 M * Bertl yes, that's it ... 1112822265 M * kevinp Hi Bertl! 1112822277 M * Bertl sim0n: but you can find his 'new' email on the ml too ... 1112822281 M * Bertl hey kevinp! 1112822294 M * kevinp Finally got all my drive problems sorted out and got the tools installed as well! 1112822306 M * Bertl excellent! 1112822320 M * kevinp I just needed more sleep and then I was able to figure it out. 1112822431 M * kevinp I am stuggling to find the "correct" way to create a new vserver though 1112822483 M * Doener http://linux-vserver.org/alpha+util-vserver 1112822486 M * kevinp I tried creating the necessary folders and files for centos4, but can't seem to get vserver-build to create a new vserver 1112822513 M * Doener create a skeleton vserver and fill the vserver's filesystem with your stuff 1112822528 M * kevinp I've tried through yum, but I would really prefer to use a ISO I have shared on an NFS server 1112822556 M * Doener well, does centos support being installed in a chroot environment from that iso? 1112822563 M * Doener should cause no problems then 1112822566 M * kevinp I was able to make a skeleton, so then do a copy? 1112822622 M * kevinp I'm coming from the older version where I used vskel to do this before 1112822629 M * Doener you just have to find a way how to get your centos stuff into /vservers/ 1112822643 M * kevinp Does it work with the new tools? 1112822697 M * Doener vskel? don't know, don't think so (it won't use the new config style, old style will work, but without the new features)... i don't even remember what it does... 1112822720 M * kevinp All the examples on the URL you sent are using apt, yum, or rpm - what's the point of using those? 1112822758 M * Bertl well, you want to use the packages on the CD, no? 1112822769 M * kevinp Sure, why not? 1112822772 J * Loki|muh loki@satanix.de 1112822781 M * micah Bertl: I agree, thats why I stopped after fixing about 20 of the problems because it didn't seem right 1112822789 M * Bertl kevinp: yes, but how dou you thing you are going to install them? 1112822816 M * kevinp Well, they aren't they already installed on the root server? 1112822838 M * Bertl don't know, did you install everything? 1112822856 M * kevinp No, I installed the minimum 1112822874 M * kevinp I now want to create a vserver based off this minimum install 1112822895 M * kevinp and then install which ever apps I want later with yum or whatever 1112822905 M * Bertl micah: damn you're right! I backported it to vanilla, obviously forgot to apply the debian patch ... 1112822914 A * Bertl is looking for a brown paperbag now ... 1112822924 M * micah Bertl: woooo... I was just going to say I tried again and same result 1112822947 A * micah hands Bertl a brown paper bag 1112822966 M * Doener Bertl: hmm... brown paper bag? you know what Michael Mittermeier said about brown paper bags, right? 1112822974 A * Bertl now with brown paperbag: okay, but that's not so bad actually, we just have to select the right patches ... 1112822983 M * Doener (that's a german comedian btw) 1112822989 M * Bertl Doener: no? 1112823002 M * micah I actually dont know what the brown paper bag is for :) 1112823017 M * Doener micah: nobody should recognize him ;) 1112823023 M * micah ahah :) 1112823048 M * Bertl micah: can't you see, it's over my head ;) 1112823068 A * micah puts eyeholes in bag for Bertl 1112823073 M * Bertl Doener: so what did he say? 1112823115 M * Doener Bertl: he said that in xxx shops you get a inconspicious brown paper bag, so that nobody knows that you've been to such a shop... and that those shops seem to be the only ones giving out such brown paper bags... [sic] 1112823157 M * Doener so now that you've got a brown paper bag.... =) 1112823176 M * Bertl hmm, the shops here have perfectly neutral black bags, you can recognize them from far ... ;) 1112823177 Q * lilo Read error: Connection reset by peer 1112823190 J * lilo ~lilo@lilo.usercloak.oftc.net 1112823199 M * Doener hm, damn country-specific humor ;) 1112823204 M * Bertl the brown bags are used for groceries and such ;) 1112823277 M * micah well, darn... I had fixed about 20 Hunk fuzz/offset patching problems and 3 or 4 FAILED problems thinking I was helping make the patch better :P 1112823291 M * micah well, it was good practice! :) 1112823295 M * Bertl ;) 1112823312 M * Bertl 14 Apr 6 01:07 patch-2.6.8-15-debian.diff.bz2 1112823322 M * Bertl this should have given me _some_ hint ... 1112823410 M * Bertl ah well, nothing which cannot be fixed ... 1112823454 M * Bertl and I'm pretty confident it will be much faster now ... 1112823473 M * Bertl (the patching/adapting) 1112823517 M * micah you were fast before, I'm afraid to see how fast this will be 1112823519 Q * erwan_ho Remote host closed the connection 1112823550 M * micah be back in a few... 1112823551 J * erwan_ho ~erwan@lns-vlq-39f-81-56-133-136.adsl.proxad.net 1112823556 M * Bertl k, cya 1112823833 M * sim0n Bertl: i can't find his new one in the archives... 1112823887 M * Bertl same name @ sigma-chemnitz.de 1112824024 M * kevinp So back to the new vs old config, vskel put files in a template of certain folders, global, static, local, etc - is that all a thing of the past? 1112824037 M * kevinp (http://linux-vserver.org/VSkel) 1112824134 Q * erwan_ho Quit: Leaving 1112824137 J * erwan_ho ~erwan@lns-vlq-39f-81-56-133-136.adsl.proxad.net 1112824220 M * Bertl kevinp: well, you can still use the tool but, correct me if I'm wrong, it just creates legacy configs ... 1112824255 M * kevinp right, I'm just trying to understand how to convert what I already understand to the new way 1112824260 M * Bertl (of course, you can either combine that with the script to convert legacy configs, or adapt it to new style configs) 1112824277 M * kevinp Is there a vskel type script for new configs? 1112824342 M * Bertl not really, because this will ultimately be done by the v(server-)copy script/tool 1112824342 M * kevinp Is Simon Garner still active in development - looks like he wrote the original 1112824371 M * Bertl I've not heard from him for some while now ... you might contact him ... 1112824382 M * Bertl s/might/might want/ 1112824501 M * kevinp Sorry about all the stupid questions, but let me re-ask this one: 1112824528 M * kevinp If I want to make a copy of my root server into a vserver, what is the best way to do it? 1112824601 M * Bertl vserver - build --help says 1112824605 M * Bertl copy ... the copy-all-from-host method which uses the recent 1112824605 M * Bertl configuration scheme 1112824629 M * Bertl so I would do: 1112824675 M * Bertl vserver shinynew build -m copy ... .. -- ... ... 1112824701 M * Bertl where it seems that the 'copy' method has no args ;) 1112824731 M * Bertl a complete example could be: 1112824774 J * lilo_ ~lilo@lilo.usercloak.oftc.net 1112824838 M * Bertl vserver shinynew build -m copy --context 42 --hostname shiny.new.org --interface shiny=eth0:192.168.0.2/24 --initstyle plain 1112824861 Q * lilo Ping timeout: 480 seconds 1112824892 N * virtuoso zaphod 1112824903 N * zaphod virtuoso 1112824909 M * virtuoso d@mn 1112824914 M * Bertl hmm? 1112824921 M * kevinp why the shiny=eth0 part? 1112824939 M * Bertl because I'm pretty sure you want one of those old style aliases ;) 1112824948 M * sim0n does anybody have some experience with lsi megaraid controllers ? 1112825001 M * virtuoso Bertl: Zaphod is already taken. :/ 1112825017 M * Bertl how unexpected ;) 1112825028 M * kevinp Can not find util-vserver installation (the file '/usr/local/lib/util-vserver/util-vserver-vars' would be expected); aborting... 1112825059 M * Bertl which tools, how did you install them? 1112825078 M * kevinp I compiled twice, once with the defaults, removed all the files, second time with parameters to install in /usr instead of /usr/local/ 1112825099 M * kevinp 204 1112825101 M * Bertl seems there is something left pointing to /usr/local ... 1112825107 N * lilo_ lilo 1112825127 M * kevinp ah, yes, vserver! 1112825145 M * kevinp I have it in /usr/sbin and /usr/local/sbin 1112825218 M * kevinp removed the old copies, now I get 1112825224 M * kevinp /usr/lib/util-vserver/vserver-build: line 125: /usr/lib/util-vserver/vserver-build.copy: No such file or directory 1112825248 M * kevinp It's true, I have no such file 1112825299 M * Bertl hmm, me neither ... please check with 0.30.205 and if the issue remains, file a bug report to the ml, cc to enrico 1112825327 M * kevinp I didn't see the release of 205 1112825359 M * Bertl http://www.13thfloor.at/~ensc/util-vserver/files/alpha/ 1112825637 Q * erwan_ho Remote host closed the connection 1112825748 M * Medivh hi 1112825755 M * Bertl evening Medivh! 1112825768 M * Medivh i got a question again... :P 1112825787 M * Bertl ah new one? 1112825793 M * Bertl s/ah/a/ 1112825798 M * Medivh probably one asked too often :> 1112825805 M * Medivh how again can i specify multiple ip addresses in different subnets (also different netmask/bcast) in a vserver's config file? 1112825836 M * Medivh currently running vs1.27 and util-vserver 0.29.5 on the box i need it on 1112825846 M * Bertl .../interfaces/0/{ip,..} 1112825858 M * Bertl ah, ancient tools too ... 1112825876 M * Medivh hehe, yeah i should really upgrade the machine 1112825889 M * Bertl then you have to use IPROOT="eth0:/ eth0:/" 1112825899 M * Medivh ah... thx 1112825906 M * Medivh broadcast is calculated automagically? 1112825914 M * Bertl yeah, but wrong IIRC ;) 1112825922 M * Medivh humn :> 1112826000 M * Medivh renumbering that machine, so i'll need to use old + new ip for now due to stupid customers not updating their dns in time :p 1112826023 M * Bertl what about NAT? 1112826065 M * Medivh guess that'd work, too 1112826085 M * Bertl but would be too simple I guess *G* 1112826088 M * Medivh inet 195.140.211.97/24 brd 85.197.125.255 scope global secondary eth0:vs00003 1112826089 M * Medivh ieks :> 1112827455 M * albeiro interesting, so fat no single coruption on ext3 1112827463 M * albeiro and i had some on xfs i like much more 1112827503 M * albeiro not to mention speed is also good... very good in fact... our of couriosity Bertl, which fs are you using ? 1112827528 M * Bertl ext2/ext3 1112827537 M * albeiro lol. kernel unpacked when i was writing it - i am shocked 1112827556 M * albeiro _no_ hdparm tweaks, just DMA on. 1112827564 M * albeiro there is much to learn for me i see :/ 1112827804 Q * muadda Quit: Leaving 1112827825 M * Bertl hmm, while I agree, as there is always plenty of stuff to learn, why did you get to that conclusion right now? 1112827888 M * albeiro i always thought my own settings based on gained knowledge are just right 1112827903 M * albeiro and now i see something done exactly oposite to my own ideas 1112827907 M * albeiro and am amazed 1112827931 M * albeiro so i come to the point i still do not know everything ;] 1112827941 M * Bertl well, the following 'tunings' are advisable (on new disks) 1112827949 M * Bertl hdparm -u 1 -c 3 -m 16 1112827959 M * Doener well, it would be boring to know everything, right? 1112827967 M * albeiro sure 1112827973 M * albeiro that's live anyway 1112828011 M * albeiro Bertl: why c3 and not 1 ? 1112828072 M * Bertl just to make sure ... 1112828086 M * Bertl 3 to enable 32-bit data transfers with a special sync 1112828086 M * Bertl sequence required by many chipsets. The value 3 works with nearly all 32-bit IDE 1112828089 M * Bertl chipsets, but incurs slightly more overhead. 1112828109 M * albeiro right, nothing to loose 1112828112 M * Bertl if you _know_ 1 works fine too, no problem with that 1112828123 M * albeiro i do not know for sure ;p 1112828293 M * albeiro for disk i was also using line similar to this (acording to drive possibilites) 1112828295 M * albeiro -a 256 -A 1 -c 1 -d 1 -k -K -M 254 -m 16 -P -u 1 -X udma5 1112828324 M * albeiro but i did not feel my system so responsive as it is now 1112828364 M * albeiro it was just giving me big (not necesary right ;) numbers in hdparm bencharm (worth nothing i think) 1112828480 M * Bertl bonnie++ is a good test, and io* (don't remember the name ;) 1112828506 M * Bertl hdparm -tT is fun if you want to see funny numbers ;) 1112828517 M * Doener o.O i just had a look at my hdparm settings, io support was off (sata disks) and google found this: http://hulllug.principalhosting.net/archive/index.php/t-52291.html 1112828527 M * Doener quote: "libata is hardcoded to use 16bit PIO data transfers." 1112828545 M * Doener this seems erhm... strange? 1112828617 M * albeiro what do you think about some settings i am not sure about - like -A 1 ? 1112828650 M * albeiro you mean iozone i think 1112828660 M * Bertl well, they are usually default, and yes ad iozone ;) 1112828698 M * Bertl -a 256 doesn't make much sense 1112828735 M * albeiro because... ? 1112828739 M * Bertl (unless you're streaming video only ...) 1112828747 M * albeiro looks like it is default, heh 1112828797 M * albeiro right, too much readahead. will downgrade it a bit and see how it goes. question is always the same - what is new magical number i should start with ;] 1112828799 M * Bertl hdparm /dev/hd0/disc 1112828799 M * Bertl /dev/hd0/disc: 1112828799 M * Bertl multcount = 16 (on) 1112828799 M * Bertl IO_support = 1 (32-bit) 1112828799 M * Bertl unmaskirq = 1 (on) 1112828801 M * Bertl using_dma = 1 (on) 1112828804 M * Bertl keepsettings = 0 (off) 1112828806 M * Bertl readonly = 0 (off) 1112828809 M * Bertl readahead = 8 (on) 1112828817 M * Bertl so no, 8 is default which means 4k on i386 which makes sense 1112828830 M * albeiro same with ppc i think... 1112828853 M * albeiro (it is Apple PowerBook G4 i am proud of :) 1112828862 M * Bertl Titanium? 1112828891 M * albeiro nah, silver or how they were called... 1112828903 M * albeiro 1.33 ghz, 15' screen 1112828915 M * Bertl have a Titanium here .. what distro do you ahve on that? 1112828979 M * albeiro i used to have gentoo but now i am testing yellow dog linux (www.yellowdoglinux.com) and am quite impressed 1112828993 M * Bertl could you test a few things for me? 1112828996 M * albeiro sure 1112829029 M * Bertl I'm trying to compile the tools (I have a mandrake rpm for them) with dietlibc for some time now, and it fails for various reasons ... 1112829057 M * Bertl http://vserver.13thfloor.at/Experimental/MDK/util-vserver-0.30.205-3mdk.src.rpm 1112829080 M * Bertl I'm pretty sure the mandrake headers and probably dietlibc is broken ... 1112829106 M * Bertl well, I guess there is no distro where it is not broken, actually ... 1112829128 M * albeiro it was a _long_ time i was using rpm (around redhat 5 up to 7.3 ;) how do i rebuild it ? 1112829145 M * Bertl use alien to get a tar with the sources and two patches 1112829159 M * Bertl but isn't yellowdog rpm based? 1112829172 M * albeiro it is 1112829177 M * albeiro i can just install it :] 1112829189 M * Bertl ah, then try with 'rpm --rebuild ' 1112829206 M * Bertl if that fails, then rpmbuild --rebuild 1112829260 M * albeiro [root@silver src]# rpm --rebuild 1112829260 M * albeiro --rebuild: unknown option 1112829261 M * albeiro lol 1112829278 M * albeiro with one "-" ;p 1112829295 M * albeiro nah, it is different option 1112829303 M * Bertl try the rpmbuild 1112829322 M * Bertl if you don't have that, try to install rpm-build* 1112829359 M * albeiro figured it out, installing... 1112829536 M * albeiro rebuilding... 1112829548 M * kevinp Do I need vhashify? 205 needs beecrypt for it to work 1112829568 M * Bertl if you do not want to use hashification, no ;) 1112829602 M * kevinp Since I have no idea what it is, I guess I don't need it! :) 1112829618 M * Bertl it's basically unification but without a reference server 1112829749 M * albeiro Bertl: failed - http://pastebin.de/94 1112829859 M * Bertl okay, thanks ... 1112829898 M * albeiro if it gave you something, good ;p 1112829981 M * kevinp "reference server" is that the same as the old skel? 1112830024 M * kevinp Before I had a skel that I made all of my vservers from, will I still have that or should I make them all copies of the root server? 1112830073 J * sig ~sig_miche@S0106000f66254626.cg.shawcable.net 1112830080 M * Bertl basically the asnwer should be: neither nor .. but yes you can have skels ... 1112830085 M * Bertl welcome sig! 1112830094 M * sig hello 1112830155 M * kevinp so instead you have a "distribution" with a list of rpms that you want installed when you create it? 1112830190 M * Bertl yes, this or one of the yum/apt-get/debootstrap methods 1112830217 M * kevinp which are all the same thing, just through different methods? 1112830233 M * albeiro interesting - default kernel configuration has [*] Set custom user task size enabled with value (0x80000000) Size of user task space. i wonder what can it give to them 1112830239 M * Bertl yes, they use either yum or apt-rpm or debootstrap 1112831144 M * kevinp FYI, after rebuilding 205, got the same error on the copy. I sent an email to the ml and enrico 1112831191 M * Bertl thanks! 1112831233 M * kevinp thanks again for all your help 1112831240 N * kevinp kevinp|gone