1170979238 M * Bertl i.c. 1170979254 A * meandtheshell is off to bed now 1170979265 M * meandtheshell bye folks 1170979270 Q * meandtheshell Quit: Leaving. 1170980289 A * ard guesses mkxauth is a shell script 1170980315 M * ard a simple dd /dev/random bs=1 count=16|hexdump or something will suffice 1170980341 A * ard used that a lot of times in his diskless workstation images 1170981182 N * TrueBrain night 1170981216 N * night TrueBrain 1170981495 P * stefani I'm Parting (the water) 1170981986 M * Bertl does anybody know if you can run linux on an xbox 360 by now? 1170983541 J * hatoon ~musis@189.12.136.26 1170983552 M * Bertl welcome hatoon! 1170983565 M * hatoon ok 1170983724 M * ntrs Bertl, no you cannot run linux on xbox360 yet. 1170983805 M * Bertl ntrs: unfortunate ... would be an interesting platform SMP powerpc for 300 USD 1170983822 M * ntrs Yes, indeed. 1170984026 M * hatoon no speak english ? 1170984214 M * Bertl hatoon: well, give it a try ... 1170984415 M * hatoon sei la 1170984418 M * hatoon rsrs 1170984438 P * hatoon Saindo 1170984456 M * baldy lol 1170984459 M * baldy ;-) 1170984533 M * baldy Bertl: mein pps count works fine ;) 1170984541 M * baldy for ddos detection 1170984577 M * Bertl okay :) 1170985151 M * cehteh ARRGS! 1170985155 M * cehteh # vserver nebulae start 1170985155 M * cehteh /lib/util-vserver/vserver.start: line 133: /usr/bin/nice: No such file or directory 1170985170 M * cehteh still ... but testme.sh shows all working 1170985183 M * Bertl well, 'which nice' ? 1170985211 M * cehteh chbind --ip 10.20.30.40 -- echo test 1170985212 M * cehteh ncontext: vc_net_create(): Invalid argument 1170985223 M * Bertl --nid 42 1170985231 M * cehteh nice is there .. everywhere .. it just a loong line 1170985246 M * cehteh ah --nid works 1170985260 M * cehteh is there a nid config missing? 1170985327 M * Bertl it takes the xid (in /etc/vservers//context) if not specified otherwise 1170985406 M * cehteh cat /etc/vservers/nebulae/context 1170985407 M * cehteh 10 1170985415 M * cehteh .. ok i am going to debug this further 1170985423 M * cehteh likely some fucked up config 1170985522 M * xp_prg2 X11 connection rejected because of wrong authentication. 1170985535 M * xp_prg2 Bertl I have xauth on the vserver now, any ideas how to fix this next problem? 1170985541 M * Bertl ah, now you are a step further 1170985551 M * Bertl first do 'xauth list' 1170985556 M * Bertl and echo $DISPLAY 1170985564 M * Bertl (upload all to paste.linux-vserver.org) 1170985709 J * _mcp ~hightower@wolk-project.de 1170985709 Q * mcp Read error: Connection reset by peer 1170985715 M * xp_prg2 http://paste.linux-vserver.org/1121 1170985753 M * xp_prg2 Bertl? 1170985757 M * cehteh vattribute: execvp("/lib/util-vserver/"): Permission denied 1170985800 M * Bertl xp_prg2: change the sshd_config regarding Localhost for X forwarding 1170985802 M * cehteh aha .. there is something missing 1170985814 M * xp_prg2 Bertl change it to what? 1170985826 M * Bertl well, what is it now? 1170985860 M * Bertl also doube check that localhost is entered as the guest ip in /etc/hosts 1170985974 M * xp_prg2 Bertl I don't see anything about Localhost in sshd_config :( 1170985998 M * Bertl X11UseLocalhost no 1170986002 M * Bertl or X11UseLocalhost yes 1170986038 M * xp_prg2 man I don't know have that entry 1170986050 M * Bertl well, then try the first one :) 1170986100 M * Bertl and add localhost to /etc/hosts :) 1170986124 M * xp_prg2 ok did that, restarted ssh and relogged in and got: 1170986134 M * xp_prg2 /usr/bin/X11/xauth: (stdin):1: bad display name "libero:10.0" in "remove" command 1170986174 M * xp_prg2 I think I need an entry for libero 1170986203 M * xp_prg2 X11 connection rejected because of wrong authentication. 1170986214 M * xp_prg2 what now? 1170986216 M * Bertl then try X11UseLocalhost yes :) 1170986270 M * xp_prg2 rejected because of wrong authentication. 1170986292 M * Bertl then your xauth is producing wrong authentications 1170986304 M * Bertl check that you have a 1170986317 M * Bertl localhost/unix:10 MIT... with xauth list 1170986329 M * Bertl and that this matches your host (you ssh from) 1170986346 M * xp_prg2 k one sec 1170986373 M * Bertl this is an xauth/ssh issue which seems to happen now with debian ... at least it was reported several times in the last few days 1170986427 M * xp_prg2 libero/unix:11 MIT-MAGIC-COOKIE-1 2945fd270b98c201ee43bfe2fd722819 libero/unix:10 MIT-MAGIC-COOKIE-1 4f1186abbd7f66f6f256e5d9469562b7 1170986442 M * xp_prg2 Bertl I don't see it do you? 1170986458 M * Bertl I don't know your magic cookies :) 1170986492 M * xp_prg2 so I have to compare the magick cookie to the cookie on the host I am sshing from? 1170986546 M * Bertl yep, and they have to match 1170986553 M * Bertl i.e. the following has to be true: 1170986568 M * Bertl $DISPLAY list a host/port pair 1170986580 M * Bertl this one must be in xauth list, and the cookie there 1170986589 M * Bertl must match the one you are ssh-ing from 1170986629 M * xp_prg2 Bertl where can the cookie on the host be located? 1170986649 M * Bertl xauth list will show them, it's in ~/.Xauthority 1170986678 M * xp_prg2 xauth> [root@engtest7 ssh]# xauth list engtest7/unix:0 MIT-MAGIC-COOKIE-1 93d434ef059b7c53ecbbaef5f991a839 1170986683 M * xp_prg2 that is the one on the host 1170986690 M * xp_prg2 I mean the client logging into the vserver 1170986708 M * Bertl okay, that one must be on the target too 1170986725 M * Bertl ssh will transfer and add that one with the :10 port 1170986735 M * xp_prg2 how do I make it do that? 1170986765 M * Bertl it does that by default on my systems :) 1170986893 M * Bertl you can do that by hand too 1170986939 M * Bertl xauth extract - $DISPLAY | ssh remotehost xauth merge - 1170986955 Q * hiaslboy Ping timeout: 480 seconds 1170986981 M * Bertl you might need to adjust the string to match the $DISPLAY on the remote 1170987000 A * xp_prg2 is afraid 1170987019 M * cehteh + exec /sbin/vnamespace --new -- /sbin/vserver ----nonamespace --debug nebulae start 1170987033 M * cehteh is the ---- already a bug? 1170987092 M * Bertl no, I don't think so, but double check with daniel_hozac 1170987167 A * cehteh just compares the log with one from a working server 1170987206 M * xp_prg2 http://paste.linux-vserver.org/1122 1170987209 M * xp_prg2 Bertl help 1170987226 M * Bertl xp_prg2: sorry, I'm neither an expert on debian nor X11 mit magic cookie authentication ... 1170987227 M * xp_prg2 can I just manually alter the .Xauthority file? 1170987263 J * AndrewLe1 ~andrew@tnlug.linux.org.tw 1170987272 M * Bertl maybe, no idea ... here it 'just' works, and I'm very happy about that :) 1170987281 M * xp_prg2 ya I would be too 1170987283 M * Bertl welcome AndrewLe1! 1170987296 M * Bertl xp_prg2: maybe ask debian or X11 folks? 1170987304 M * xp_prg2 yes I will thanks 1170987369 Q * AndrewLee Ping timeout: 480 seconds 1170987615 M * Bertl let us know what it was, once you resolved it ... 1170987767 M * cehteh mhm .. these debug logs are meant for mental masochists 1170987922 M * Bertl well, it is simply what is executed 1170988182 M * xp_prg2 can I turn Xauthentication off? 1170988219 M * cehteh except slight differences they are both the same .. even the command which finally fails is the same .. just works on one machine and fails on the other 1170988293 M * Bertl well, it says, no such file 1170988305 M * cehteh yes but it is there 1170988312 M * Bertl so whatever it is trying to reach, isn't there or is not reachable (permissions) 1170988346 M * cehteh starbase:/usr/src/linux-2.6.20# /vservers/nebulae/usr/bin/nice 1170988346 M * cehteh 0 1170988346 M * cehteh starbase:/usr/src/linux-2.6.20# /usr/bin/nice 1170988346 M * cehteh 0 1170988371 M * Bertl check all the pathes recursively upwards, regarding normal permissions and xattrs (it is not the nice which was not found here :) 1170988377 M * cehteh on the root and on the guest both are reachable and executeable .. guest is 32 bit elf root is 64 bit 1170988422 M * cehteh i even executed the formed commandline partially by hand .. but that obliviously missing some set up thing 1170988425 M * cehteh s 1170988491 M * cehteh vcontext: open("/dev/null"): No such file or directory 1170988541 M * Bertl have you verified the directory permissions as suggested? 1170988640 M * cehteh mhm i didnt changed anything and it worked before ... which dirs to check anyways 1170988676 M * Bertl 03:32 < Bertl> check all the pathes recursively upwards, regarding normal 1170988685 M * Bertl permissions and xattrs 1170988746 M * cehteh upwards from where? 1170988787 M * Bertl from the file which cannot be accessed inside the guest 1170988792 M * cehteh i tried this whole line manually taking out the save_ctxinfo and use -- echo test at the end instead /sbin/init works 1170988896 M * cehteh do i need /sbin/vcontext in the guest? 1170989043 A * cehteh is not very familar with the util-vserver .. 1170989215 Q * FireEgl Quit: ... 1170989235 M * xp_prg2 ok figured it out, sshd_config must have a Listen ip address, and X11UseLocalhost no, the guest must be able to resolve its own ip address 1170989252 M * xp_prg2 just put an entry in the /etc/hosts 1170989254 M * Bertl cehteh: no, you do not need any tools inside the guest 1170989274 M * Bertl xp_prg2: the listen address sounds fishy 1170989284 M * Bertl none of my guest here have that ... 1170989296 M * cehteh /sbin/vcontext --migrate-self --endsetup --chroot --silent --initpid --disconnect --personality-type linux_32bit -- /bin/echo test 1170989296 M * cehteh vcontext: vc_set_cflags(): No such process 1170989303 M * cehteh is that supposed to be ok? 1170989313 M * Bertl xp_prg2: the host/resolv part I buy ... that was what I suggested in the first place ... 1170989335 M * cehteh mhm yes 1170989348 M * Bertl cehteh: you are missing another sequence of commands 1170989368 M * cehteh yes ... i try to test them in isolation 1170989412 M * cehteh this thing is really ugly to debug .. i dont know all the new commands and for what they are good for .. and they depend on things i dont know eihter 1170989460 M * Bertl you have two guests, one is working as expected, the other isn't, right? 1170989473 M * cehteh the working one is on another computer 1170989484 M * cehteh this problem is on a computer with only one guest 1170989497 M * Bertl maybe with a different kernel and different tools :) 1170989501 M * cehteh yes 1170989511 M * Bertl so why do you mention that at all? 1170989512 M * cehteh diffrent processor too 1170989527 M * Bertl it is of absolutely no relevance then, no? 1170989554 M * cehteh well i mentioned that i compared the debug logs ... 1170989558 M * cehteh thats all 1170989580 M * Bertl try to build a new guest 1170989581 M * cehteh if there would be some glitch in the scripts i had a chance to see it 1170989584 M * Bertl see if that works 1170989596 M * cehteh whats the easiest way to build a new guest in debian? 1170989619 M * Bertl you know the wiki? 1170989623 A * cehteh used dbootstrap and manual config .. that might be the problem already :) 1170989637 M * cehteh heh in the time when that wasnt well documented 1170989738 M * cehteh never used newvserver ...lemme try that 1170989792 M * Bertl forget that debian specific tool 1170989815 M * Bertl I'm going to delete the debian install pages soon on the wiki 1170989845 M * Bertl micah: please rework that, feed back the newvserver specific stuff upstream and remove it ASAP! 1170989891 M * cehteh haha ... first you recommend the wiki ... then you say i shouldnt use what i read there :P 1170989896 M * Bertl vserver foo build -m debootstrap --context 42 --hostname=foo.lan --interface eth0:192.168.1.10/24 -- -d edgy 1170989908 M * cehteh oki 1170989918 M * Bertl well, the debian folks keep really annoying me with the newvserver stuff 1170989919 M * cehteh -d etch ? 1170989931 M * Bertl or sid or whatever 1170989950 M * cehteh yeah i build tools by myself and dont use the debian stuff ... that was just the first thing i found on the wiki 1170990022 M * cehteh ok building 1170990072 M * cehteh thangs for your help 1170990110 M * Bertl you're welcome! 1170990883 M * cehteh # vserver test start 1170990883 M * cehteh /lib/util-vserver/vserver.functions: line 685: /usr/bin/find: No such file or directory 1170990884 M * cehteh /lib/util-vserver/vserver.functions: line 686: xargs: command not found 1170990912 M * cehteh hmpf .. with a new debootstrapped server ... 1170991014 M * cehteh thats now really weird 1170991550 M * Bertl so I'd say your setup is somewhat broken 1170991570 M * Bertl did you do the testme.sh? if so, redo with -d and upload that 1170991578 M * Bertl also do vserver-info - SYSINFO 1170991587 M * cehteh yes ... with new kernel all success 1170991685 M * cehteh looks all ok 1170991761 M * cehteh dietlibc yes .. gcc3.x everything fine/yes/available etc 1170991851 M * cehteh the rest is a normal debian/etch system 64 bit ... the now debootstrapped vserver should be 64 bit too .. i just build it and didnt changed anything 1170991868 M * Bertl maybe you have a mix of different tools installed? 1170991911 M * cehteh mhm lemme check if i once had some with --prefix=/ussr 1170992003 M * cehteh make install should overwrite anything shoudlnt it .. and usually i make uninstall .. but i may forgotten it once 1170992060 M * Bertl well, it will overwrite stuff, but not remove existing stuff from curious places 1170992080 M * Bertl also, did you check the perms and attributes for the guest pathes now? 1170992103 M * cehteh yes .. but it should not depend on exiisting stuff at curious places either 1170992129 M * cehteh yes .. never unified/hashed them either 1170992168 M * cehteh and the new vserver build should set up all correct shoudnt it? 1170992197 M * Bertl it will not change flags/permissions in your system 1170992211 M * cehteh ah crap there are some tools in /usr/local 1170992453 M * cehteh that wasnt it ... 1170992681 M * Bertl well, please wait for daniel_hozac, and try to investigate 1170992842 M * cehteh yeah ... not urgend only annoying ... i now wiped the tools and reinstalled .. and debootstrap again .. last try 1170992927 M * cehteh ok same problem .. i doing something other now 1170995443 M * cehteh huh .. a friend reports he has a kernel without /proc/kmsg .. which kernel option is that? 1170995613 M * Bertl none I would know of :) 1170995665 M * cehteh CONFIG_PRINTK=y ... maybe 1170995845 M * cehteh #ifdef CONFIG_PRINTK 1170995845 M * cehteh entry = create_proc_entry("kmsg", S_IRUSR, &proc_root); 1170995848 M * cehteh gotcha 1170995873 M * Bertl interesting ... well, nothing I would disable ... 1170995914 M * cehteh a small terminal client -- almost embedded system with linux on a CF card 1170996886 J * olivierk_ ~olivier@olivierk.org 1170996903 M * Bertl welcome olivierk_! 1170996993 Q * olivierk Ping timeout: 480 seconds 1170997116 J * Aiken_ ~james@ppp41-126.lns1.syd6.internode.on.net 1170997433 Q * Aiken Ping timeout: 480 seconds 1170998711 J * DoberMann_ ~james@AToulouse-156-1-129-116.w90-30.abo.wanadoo.fr 1170998817 Q * DoberMann[ZZZzzz] Ping timeout: 480 seconds 1170999066 J * Aiken__ ~james@ppp103-30.lns1.bne1.internode.on.net 1170999180 M * daniel_hozac cehteh: /usr/bin/nice is a relative symlink to /bin/nice and /bin/nice is missing from the guest? 1170999259 M * cehteh ls /vservers/nebulae/usr/bin/nice /usr/bin/nice -lh 1170999259 M * cehteh -rwxr-xr-x 1 root root 23K Aug 31 14:25 /usr/bin/nice 1170999259 M * cehteh -rwxr-xr-x 1 root root 17K Aug 31 13:57 /vservers/nebulae/usr/bin/nice 1170999263 M * cehteh no symlinks 1170999301 M * daniel_hozac hmm, is it the other way around maybe? 1170999318 M * daniel_hozac what does grep _NICE /lib/util-vserver/util-vserver-vars return? 1170999352 J * Aiken ~james@ppp103-30.lns1.bne1.internode.on.net 1170999354 M * cehteh _NICE="/usr/bin/nice" 1170999389 M * cehteh did you seen the other problem with the freshly bootstrapped vserver? 1170999402 M * cehteh # vserver test start 1170999402 M * cehteh /lib/util-vserver/vserver.functions: line 685: /usr/bin/find: No such file or directory 1170999402 M * cehteh /lib/util-vserver/vserver.functions: line 686: xargs: command not found 1170999403 Q * Aiken_ Ping timeout: 480 seconds 1170999417 M * daniel_hozac is /usr or /usr/bin a symlink? 1170999427 M * cehteh not that i know 1170999432 M * daniel_hozac that looks more like your host doesn't have findutils installed. 1170999487 M * Bertl IIRC, we did hit that dependancy previously? 1170999496 M * cehteh # file `which find` 1170999496 M * cehteh /usr/bin/find: ELF 64-bit LSB executable, AMD x86-64, 1170999519 M * daniel_hozac hmm... is /lib a symlink? 1170999538 M * cehteh drwxr-xr-x 17 root root 6144 Feb 9 04:39 lib 1170999539 M * cehteh lrwxrwxrwx 1 root root 20 Feb 7 03:22 lib32 -> /emul/ia32-linux/lib 1170999539 M * cehteh lrwxrwxrwx 1 root root 4 Feb 7 03:23 lib64 -> /lib 1170999553 M * cehteh negative houston 1170999622 M * cehteh want a --debug log from the freshly instantiated vserver? 1170999623 Q * Aiken__ Ping timeout: 480 seconds 1170999667 M * daniel_hozac an strace would probably be even more interesting. 1170999686 M * cehteh strace stucks because at some point it isnt allowed anymore 1170999713 M * cehteh but i can can add the CAP .. which one was it? CAP_SYS_ADMIN ? 1170999719 M * daniel_hozac yeah, it can't enter contexts, but we want the stuff before that anyway. 1170999739 M * cehteh it hangs and i have to hard kill it ... 1170999747 M * daniel_hozac there's no capability for that, AFAIK. 1170999766 M * cehteh strace -f .. i tihnk without -f its worthless isnt it? 1170999778 M * daniel_hozac yep. 1170999873 M * cehteh http://www.pipapo.org/vserver.start.strace 1170999910 M * daniel_hozac and that's with the find thing, yes? 1170999972 M * cehteh yes 1170999994 M * daniel_hozac hmm, maybe try with -F too. 1171000003 M * daniel_hozac because that one ends way too early. 1171000034 M * cehteh there 1171000047 A * cehteh reads it too 1171000100 M * daniel_hozac huh, now it doesn't follow forks at all? 1171000126 M * cehteh -fF ? 1171000131 M * daniel_hozac yeah.- 1171000155 M * cehteh again :) 1171000198 M * cehteh hmm 1171000209 M * daniel_hozac why does it seem incapable to follow the clone in vnamespace? 1171000226 M * cehteh why do you ask me? :) 1171000245 M * cehteh maybe it cant follow into other namespaces? 1171000275 M * cehteh vserver ... to tight for debugging :) 1171000299 M * daniel_hozac the filesystem namespace vnamespace is using is vanilla though. 1171000322 M * daniel_hozac and AFAIK all of the ptrace checks are done against the xid. 1171000329 M * Bertl yep 1171000340 M * Bertl at least the ones we introduce 1171000358 M * cehteh mhm ... maybe i rebuild the kernel without filesystem xid's i dont need them anyways 1171000374 M * daniel_hozac that likely wouldn't help... 1171000390 M * cehteh would help debugging or? 1171000426 M * daniel_hozac well, just not mounting with tagxid should have the same effect. 1171000434 M * Bertl why not add the strace to the command sequence after chainging the ccontext? 1171000472 M * cehteh yeah .. but i didnt want to modify the utils at first 1171000500 M * cehteh a vserver --strace 'strace options' would be nice somehow ... 1171000518 M * cehteh but util-vserver already looks like there is too much duct tape 1171000718 M * cehteh maybe i move /vservers out of the way, delete /etc vservers and reinstall it completely .. 1171000775 M * cehteh since noone else has problems with it, i think its rather some bug here than some vserver bug and maybe not of interest and only a waste of time 1171000830 M * daniel_hozac an strace of /lib/util-vserver/vserver.functions: line 685 1171000912 M * cehteh ok http://www.pipapo.org/vserver.test.log << debug log maybe you find something 1171000920 M * cehteh i prepare the strace 1171000949 M * daniel_hozac do you have mounts anywhere, btw? 1171000966 M * cehteh in the guest? no 1171000975 M * daniel_hozac heh, guess i see it. 1171000980 M * daniel_hozac ++ /bin/umount -l -n /usr 1171000995 M * cehteh /data/fast/vservers on /vservers type none (rw,bind) 1171001009 M * cehteh but that worked that way before 1171001046 M * cehteh mhm 1171001053 M * cehteh umount /usr .. wtf ... 1171001054 M * daniel_hozac i guess i should just hardcode /usr/bin in the namespace cleanup exclusion list. 1171001106 M * daniel_hozac you could echo /usr >> /etc/vservers/.defaults/namespace-cleanup-skip 1171001144 M * cehteh woah 1171001146 M * cehteh works 1171001165 M * cehteh /lib/util-vserver/vserver.stop: line 85: 22839 Killed "${NICE_CMD[@]}" ${USE_VNAMESPACE:+$_VNAMESPACE --enter "$S_CONTEXT" -- } $_VCONTEXT $SILENT_OPT --migrate --chroot --xid "$S_CONTEXT" -- "${INITCMD_STOP[@]}" 1171001168 M * cehteh when stopping 1171001186 M * daniel_hozac that's expected with Debian guests. 1171001196 M * cehteh ok 1171001206 M * cehteh ok thanks .. works now 1171001230 M * daniel_hozac you're welcome! 1171001494 M * Bertl hum, so what did actually happen here? 1171001503 M * daniel_hozac the /usr mount is cleaned up. 1171001544 M * Bertl and why? 1171001559 M * cehteh in guest namespace before chroot or what .. i dont understand the thing eiher :) 1171001580 M * daniel_hozac because neither the guest nor the utils live below /usr ;) 1171001605 M * cehteh vserver balack magic 1171001608 M * cehteh black 1171001618 M * daniel_hozac the guest namespace is created as one of the first steps. 1171001652 M * cehteh i think i have a glimpse whats going on ... but why? :)O 1171001661 M * daniel_hozac it's prepped by removing mounts, mounting the stuff the guest needs, rbind'ing the guest to /, and then finally the guest is started. 1171001676 M * Bertl so far so good ... 1171001692 M * Bertl how does that affect stuff mounted inside the guest? 1171001702 M * daniel_hozac ? 1171001713 M * Bertl you say, /usr is removed, no? 1171001717 M * daniel_hozac right. 1171001721 M * Bertl on the host? 1171001726 M * daniel_hozac yep. 1171001730 M * cehteh in host .. yes ... but guest namespace 1171001755 M * Bertl i.c. so the tools unmount their directory :) 1171001759 M * daniel_hozac no. 1171001768 M * daniel_hozac the tools are installed in /. 1171001775 M * daniel_hozac that's why /usr is unmounted in this case. 1171001784 M * cehteh so we do rbinding and not chrooting anymore? 1171001786 M * daniel_hozac (it's usually kept around because the utils live somewhere in there) 1171001787 M * Bertl /.? 1171001800 M * daniel_hozac followed by a period ;) 1171001812 M * daniel_hozac (i swear i typed a / before that...) 1171001818 M * Bertl the tools are in /? who would install them there? 1171001826 M * cehteh --prefix='' 1171001884 M * Bertl okay, so you use stuff from /usr but the tools are in /* 1171001901 M * cehteh .. --prefix=/opt will also break 1171001911 M * daniel_hozac http://svn.linux-vserver.org/projects/util-vserver/changeset/2492 should fix it. 1171001959 M * Bertl I guess, it would be smart to make a list of all tools/binaries/configs? the tools use, and except all pathes leading there from bein unmounted, no? 1171001977 M * cehteh well .. at least it wasnt my error finally :P 1171001992 M * cehteh Bertl: right 1171002001 M * daniel_hozac sure, but i'd like to keep the list as minimal as possible... 1171002007 M * cehteh or use a busybox? 1171002053 M * daniel_hozac and certain things might not be required in certain configurations. 1171002106 A * cehteh wonders that noone else experienced that bug ... seems noone installed outside of /usr 1171002162 M * daniel_hozac _and_ has /usr as a separate filesystem. 1171002176 M * Bertl daniel_hozac: okay, maybe the scripts should get a before/after check then, so that the tools can actually tell, oops my binary is gone now? 1171002206 M * daniel_hozac a check to see if all the binaries are available? 1171002209 M * Bertl daniel_hozac: what if I --bind mount /etc/vservers into place? 1171002213 M * cehteh well .. it told that .. the problem is that the reason wasnt clear 1171002226 M * Bertl daniel_hozac: no, just before they are first used 1171002247 M * daniel_hozac it will unmount the other one, but /etc/vservers will stick around. 1171002268 M * Bertl daniel_hozac: what is the other one? 1171002275 M * daniel_hozac the "real" mount. 1171002280 M * cehteh why do we rebind to / anyways ... 1171002286 M * Bertl I'm thinking about having /etc/vservers on /opt 1171002290 M * cehteh why not somewhere deeper 1171002302 M * Bertl and doing a --bind /opt/vservers /etc/vservers 1171002317 M * daniel_hozac right. that should work fine. 1171002332 M * Bertl ah? so that one is not cleaned up? 1171002335 M * daniel_hozac /opt/vservers would be unmounted in the guest's namespace though. 1171002341 M * cehteh /etc/vserver/*/rootfs 1171002346 M * daniel_hozac /etc/vservers/ is kept. 1171002354 M * Bertl ah, okay, good to know 1171002391 M * daniel_hozac as is /vservers//*, the tools' sbindir, libdir, and statedir. 1171002418 M * daniel_hozac and any path that has /dev or /proc in it. 1171002463 M * Bertl interesting, so, if there is some /proc mount into some subdir, (not related) it will be kept? 1171002476 M * daniel_hozac yes, which is rather unfortunate. 1171002500 M * Bertl is there a reason for doing so? 1171002508 M * daniel_hozac to be perfectly honest, i'm not sure why /proc is kept at all, or why /dev can be anywhere in the path, but... 1171002611 M * daniel_hozac i guess it's just that this test is easier than checking if the path starts with /proc or /dev. 1171002670 M * Bertl i.c., well tx for the info 1171002721 J * FireEgl Proteus@68.220.222.136 1171003071 Q * ntrs Read error: Connection reset by peer 1171003073 J * ntrs ~ntrs@68-188-55-120.dhcp.stls.mo.charter.com 1171005558 M * Bertl okay, off to bed now ... have a good one everyone! cya! 1171005564 N * Bertl Bertl_zZ 1171005945 J * ntrs_ ~ntrs@68-188-55-120.dhcp.stls.mo.charter.com 1171005945 Q * ntrs Read error: Connection reset by peer 1171007291 N * DoberMann_ DoberMann 1171009080 J * Daniel15 ~dansoftau@server.daniel15.com 1171009477 J * DavidS ~david@chello062178045213.16.11.tuwien.teleweb.at 1171009752 J * dna ~naucki@45-223-dsl.kielnet.net 1171009922 N * DoberMann DoberMann[PullA] 1171010924 J * cdrx ~legoater@blueice2n1.uk.ibm.com 1171013268 J * dlezcano ~dlezcano@blueice4n2.uk.ibm.com 1171013567 M * matti Morning folks. 1171013648 M * Daniel15 Hello :) 1171013747 M * daniel_hozac Daniel15: probably better to just remove the pages when in doubt ;) 1171013790 M * Daniel15 Yeah, I thought it might have had a use (no idea what, though :P) 1171013809 A * Daniel15 runs off to delete the page :) 1171013824 M * Daniel15 What does this "Mark this article as patrolled" link do? 1171013871 M * Daniel15 Oops... I clicked it, and have no idea what I did :P 1171013891 M * Daniel15 Anyways, deleted the page :-) 1171013950 M * Daniel15 Wow, someone fixed the grammar issue :P 1171013955 M * Daniel15 "We are currently migrating to MediaWiki from our old system, but not all content has been migrated yet. " 1171013967 M * Daniel15 Used to say something like "We currently migrate to MediaWiki" 1171014029 M * Daniel15 I've got to go, I may be back later 1171014032 Q * Daniel15 Quit: ( www.nnscript.de :: NoNameScript 4.02 :: www.XLhost.de ) 1171014136 M * matti Who wants some 0xc0ffee? 1171014389 M * derjohn daniel_hozac, can you tell me, what Redhat Enterprise Linux AS is, and if it is opensource? I see support packs in their shop, and d/l only possible with trial foobar or subscription.... but they dont seem to charge for linux distri itself. is redhat free, but simply not available? I.e. could i copy from someone? 1171014448 M * daniel_hozac i believe the EULA is what limits it. 1171014461 M * daniel_hozac the software is free to redistribute, that's why centos.org exists ;) 1171014483 N * DoberMann[PullA] DoberMann 1171014839 M * eyck you can't use RedHat AS taken from someone else... binaries are restricted, 1171014860 M * eyck but the source is available, and binaries built from that source are freely distributable, and that's why centos exists. 1171015953 J * meandtheshell ~markus@85-124-175-169.dynamic.xdsl-line.inode.at 1171016005 M * daniel_hozac that's what i was trying to say. 1171016421 Q * m`m`h Ping timeout: 480 seconds 1171016660 M * pflanze On http://linux-vserver.org/util-vserver:Documentation, /etc/vservers/vserver-name/apps/vunify/hash= appears to be a typo 1171016681 M * pflanze Also, it's not clear what *id and *method is. 1171016706 M * pflanze The page says, ../hash is "A directory which will be used as the storage place for the vhashify command." 1171016715 M * pflanze So one would think there are no su-entries. 1171016722 M * pflanze Maybe just the wording is confusing. 1171016788 M * daniel_hozac "su-entries"? 1171016797 M * pflanze The flower page is a bit better (no "=", and "id ->" suggests a hardlink so explains it better), but the wording of the first sentence is still in conflict with reality. 1171016812 M * daniel_hozac s/hardlink/symlink/ 1171016844 M * daniel_hozac personally, i'd like to remove util-vserver:Documentation, as it's just the flower page in another format. 1171016853 M * pflanze The problem is, the "hash" directory is *not* the storage place, but it's entries designate it. 1171016868 M * pflanze if i understand correctly. 1171016870 M * daniel_hozac and i can't update it. 1171016903 M * pflanze yeah, good idea, except maybe that others can't edit the flower page? 1171016929 M * pflanze Is it in a versioning system so one could provide patches easily? 1171016933 M * daniel_hozac yes. 1171016945 M * daniel_hozac it's in doc/configuration.xml in the util-vserver tree. 1171017037 M * pflanze I'm missing information about the CVS/whatever repository on the wiki. 1171017123 M * pflanze ah, "Archives" 1171017316 J * Piet hiddenserv@tor.noreply.org 1171017886 M * pflanze Do the hashify/unify tools expect the exclude list containing paths starting with or without slash? 1171017948 M * pflanze And it looks like hashify excludes some paths (/etc, /home, /root) anyway by default? Where can this be found? 1171017985 M * pflanze Are entries starting with # ignored? 1171017999 M * pflanze Are empty lines ignored? 1171018305 A * pflanze pokes daniel_hozac and points to the above questions 1171018388 Q * Aiken Quit: Leaving 1171018410 M * DavidS pflanze: yes, there are default excludes, but they are completely replaced when specifying custom excludes AFAIK 1171018452 M * pflanze Do you know more about the format of paths? 1171018461 M * pflanze Maybe I should search in the sources in the end 1171018579 M * pflanze The problem with letting users writing docs may be that the users don't know the intents or future directions of the authors. 1171018654 M * pflanze So what I'm collecting here can just be what the status quo is, and that might break tomorrow if the author doesn't (want to) care about the docs. 1171019011 M * daniel_hozac i tend to update all the docs i know of when i break things 1171019077 M * daniel_hozac as for your questions, i don't have any source trees with me. 1171019244 M * TrueBrain what is wrong if 'top' or 'vtop' tells you that it is using 9999% CPU? IN fact, the CPU indicator has become a counter, is starts with 1, and depending how much a processes uses, climbs to 9999.... 1171019277 M * daniel_hozac kernel? 1171019310 M * TrueBrain 2.6.17-vs2.0.2.1-gentoo SMP x86_64 1171019402 M * TrueBrain I have to add that it happened after adding "acpi=noirq noapic" to the kernel options 1171019437 M * daniel_hozac sounds really weird... 1171019469 M * daniel_hozac doesn't really sound like a vserver problem though. 1171019472 M * TrueBrain it is, and annoying. but okay, as it is a development system, it isn't the end of the world :) 1171019485 M * TrueBrain hmm, I could try to load a normal kernel and find that out yes 1171019500 M * TrueBrain normal = vanilla :) 1171019547 M * pflanze TrueBrain: you could check whether normal top differs, and check in ctx 1 or a vserver context. 1171019559 M * TrueBrain top and vtop both show this behavoir 1171019566 M * pflanze ah 1171019578 M * pflanze top on the host too? 1171019581 M * TrueBrain yes 1171019607 M * pflanze even if no vserver is running? 1171019610 M * TrueBrain cpu-times are also really wrong, haha, didn't know I had that much cpu-time :p 1171019615 M * TrueBrain pflanze: even then 1171021827 M * TrueBrain okay, I rebooted to 2.6.19-vs2.2.0-rc9-gentoo, and now it is all running fine :) 1171022375 Q * shedi Quit: Leaving 1171022850 Q * bronson Ping timeout: 480 seconds 1171023764 J * m`m`h ~simba@deb30.mgts.by 1171026078 M * daniel_hozac pflanze: so, filenames start with /, entries with # are ignored, as are empty lines. 1171026096 M * pflanze ok, thanks! 1171026114 M * daniel_hozac the default exclude list is in distrib/misc/vunify-exclude, and when installed in /usr/lib*/util-vserver/defaults/vunify-exclude 1171026139 M * pflanze why dash and not slash? 1171026153 M * pflanze I thought its in vunify/exclude, that's what flower page says 1171026162 M * daniel_hozac but if you want to modify it, you should put it in /etc/vservers/.defaults/apps/vunify/exclude. 1171026797 M * daniel_hozac TrueBrain: except for the bugs ;) 1171027429 Q * infowolfe Read error: Connection reset by peer 1171028710 J * infowolfe ~infowolfe@c-67-164-195-129.hsd1.ut.comcast.net 1171029432 J * kir_home ~kir@swsoft-mipt-nat.sw.ru 1171029906 J * lilalinux ~plasma@80.69.41.2 1171030170 J * ema ~ema@lart.galliera.it 1171030899 J * URfriend ~URfriend@125.112.203.58 1171030908 P * URfriend 1171033174 Q * gab Quit: Leaving 1171033814 Q * TrueBrain Ping timeout: 480 seconds 1171034393 J * TrueBrain truelight@openttd.org 1171034507 Q * duckx Ping timeout: 480 seconds 1171034905 Q * infowolfe Quit: Leaving 1171036222 Q * mountie Remote host closed the connection 1171036289 J * mountie ~mountie@CPE0080c6fe323f-CM000a739acaa4.cpe.net.cable.rogers.com 1171036561 J * duckx ~Duck@tox.dyndns.org 1171037181 Q * Piet Quit: Piet 1171037212 J * Piet hiddenserv@tor.noreply.org 1171037530 J * shedi ~siggi@ftth-237-144.hive.is 1171038677 Q * shedi Remote host closed the connection 1171040631 J * shedi ~siggi@ftth-237-144.hive.is 1171041193 Q * kir_home Ping timeout: 480 seconds 1171041698 Q * cdrx Read error: Connection reset by peer 1171041738 J * olivierk ~olivier@olivierk.org 1171041848 Q * olivierk_ Ping timeout: 480 seconds 1171044363 J * kir_home ~kir@213.152.157.70 1171044461 J * stefani ~stefani@208.99.202.5 1171045344 Q * kir_home Quit: Ухожу я от вас 1171045406 N * _mcp mcp 1171045577 M * pflanze daniel_hozac: is there a way to get the list of excludes which the hashify tool(s) is/are building for a particular "vserver foo hashify" call? 1171045677 M * nebuchadnezzar by default it uses /usr/lib/util-vserver/defaults/vunify-exclude 1171045848 Q * dlezcano Read error: Connection reset by peer 1171045904 Q * ema Quit: leaving 1171046082 Q * gerrit Read error: Operation timed out 1171046397 J * cdrx ~legoater@cap31-3-82-227-199-249.fbx.proxad.net 1171046795 M * daniel_hozac pflanze: increasing the verbosity might do it, but i don't think it ever outputs the list. 1171046971 T * daniel_hozac http://linux-vserver.org/ | latest stable 2.0.2.1, 2.0.3-rc1, 2.2.0-rc12, devel 2.1.1.7.1, 2.3.0.9, stable+grsec 2.0.2.1, 2.2.0-rc12 | util-vserver-0.30.212 | libvserver-1.0.2 & vserver-utils-1.0.3 | He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the Wiki, and we'll forget about the minute ;) 1171047913 Q * lilalinux Remote host closed the connection 1171048037 M * nebuchadnezzar is there an openvpn.org man here ? 1171048064 M * nebuchadnezzar arf 1171048065 M * nebuchadnezzar openvps 1171049945 M * daniel_hozac i wondered that myself way-back-when :) 1171050015 M * TrueBrain is their website working by now? :) 1171050250 M * nebuchadnezzar their web site has a proxy problem 1171050295 M * nebuchadnezzar I wrote to root@ because webmaster do not exists 1171050321 M * TrueBrain they have that for ages now 1171050333 M * nebuchadnezzar I just arrive :-) 1171050349 M * daniel_hozac i guess that shows whoever was interested in it isn't anymore.. 1171050352 N * Bertl_zZ Bertl 1171050356 M * daniel_hozac morning Bertl! 1171050430 M * TrueBrain hmm.. is there anywhere documented what sched_hard does? 1171050460 M * TrueBrain I assumed it only limited the VPS to a certain limit, but it kills processes too, and in fact in kills many processes at a crazy rate ;) (even so it can use 50% CPU) 1171050474 M * daniel_hozac "kills"? 1171050488 M * TrueBrain dmesg shows: VM: killing process 1171050495 M * TrueBrain (which in fact could be a bit more verbose ;)) 1171050501 M * daniel_hozac that sounds more like OOM rather than the hard CPU scheduler... 1171050534 M * TrueBrain but why does the OOM do it? 1171050542 M * TrueBrain oh wait... hmm.. 1171050544 M * Bertl because you are hitting a memory limit 1171050548 M * daniel_hozac because you're Out Of Memory? 1171050552 M * TrueBrain becuase of the CPU limiter, httpd starts using memory as a maniac 1171050579 M * TrueBrain (as in, currently it uses 1.5 GiB :( 1171050707 M * TrueBrain k, tnx guys. Now I need to find out why Apache is doing this :( (I hate Apache) 1171051191 M * daniel_hozac Bertl: the Fedora oops is in choose_new_parent, BUG_ON(p == reaper). 1171051210 M * daniel_hozac Bertl: appears to be Fedora-specific though. 1171051367 M * Bertl 2.6.20 or 19? 1171051371 M * daniel_hozac 2.6.19. 1171051407 M * Bertl hmm, so we end up self reaping here? 1171051428 M * Bertl that might not be a bug after all, as was discussed on the ML 1171051439 M * daniel_hozac hmm? 1171051444 M * Bertl lkml 1171051459 M * daniel_hozac ah, i have yet to subscribe there... 1171051486 M * Bertl you should have received that one, it was during the nfs thread 1171051505 M * Bertl let me pick out the IMHO relevant details ... sec 1171051544 M * daniel_hozac oh, Oleg pointing out that we can reap ourself if we're a thread? 1171051568 M * daniel_hozac (or something along those lines...) 1171051623 M * Bertl I agree, but please note that the child_reaper is not the only issue. Think 1171051623 M * Bertl about sub-thread which auto-reaps itself. 1171051635 M * Bertl (Oleg Nesterov) 1171051645 M * Bertl yep, precisely 1171051662 M * daniel_hozac in this case, i think it's because the guest's init is exiting. 1171051670 M * Bertl maybe we are hitting this one and it goes through choose_new_parent() too? 1171051695 M * Bertl ah, good point 1171051710 M * Bertl we should special case that, for init dying 1171051740 M * daniel_hozac or, hmm... it's vcontext exiting that causes the oops. 1171051746 M * Bertl but hmm, that works in my test cases quite fine, with something like 'true' as init 1171051776 M * Bertl do we have the oops somewhere? (not fully awake yet :) 1171051778 M * daniel_hozac yeah, it works fine on 2.6.19.3-vs2.2.0-rc12. it's just in the Fedora version. 1171051800 M * Bertl ah? on the same machine? 1171051805 M * daniel_hozac yeah. 1171051820 M * Bertl that indeed points to an FC port issue 1171051832 M * Bertl how far is that kernel from mainline? 1171051834 M * daniel_hozac i'm guessing it's due to utrace. forget_original_parent looks completely different. 1171052042 M * daniel_hozac the diff from mainline is almost 10 MiB. 1171052056 M * daniel_hozac (though that includes 2.6.19.3) 1171052432 M * daniel_hozac in forget_original_parent, why doesn't the first choose_new_parent use reaper? 1171052491 M * Bertl are we talking mainline code now? 1171052518 M * daniel_hozac yeah. 1171052534 M * Bertl sec, looking 1171052568 M * Bertl we start out with the father 1171052588 M * Bertl and look through the sibblings ... 1171052613 M * Bertl if we do not find one (i.e. we return to the father) 1171052627 M * Bertl we get the vx_child_reaper() 1171052667 M * Bertl then we start to reparent children 1171052689 M * Bertl this handles cases where we have a multithreaded app 1171052706 M * Bertl with one tread spawning a cchild and exiting before the child finished 1171052816 M * Bertl in theory, that could end up with 'reaper = father' too 1171052920 M * Bertl but it should never end up with reaper = p 1171052950 M * Bertl maybe you are hitting the 1171052956 M * Bertl reaper->exit_state ? 1171052964 M * daniel_hozac no, i verified by separating them. 1171053030 M * Bertl who are the callers of choose_new_parent() in the FC codebase? 1171053045 M * daniel_hozac forget_original_parent only. 1171053060 M * Bertl okay, could you upload the code for that somewhere? 1171053149 M * Bertl and I'd like to see the oops, if possibe :) 1171053152 M * daniel_hozac http://paste.linux-vserver.org/1123 1171053193 M * daniel_hozac http://paste.linux-vserver.org/1124 1171053254 M * Bertl ah, fc doesn't care about the ptraced children? 1171053269 M * daniel_hozac utrace reimplements ptrace. 1171053310 M * daniel_hozac the traced children should be just regular children, IIUC. 1171054052 M * daniel_hozac Bertl: changing it to http://paste.linux-vserver.org/1125 seems to have fixed the problem. 1171054338 Q * xp_prg2 Quit: Ciao! 1171054852 M * baldy Bertl: huhuhu 1171054859 M * baldy dod i saw the email? 1171054861 M * baldy did 1171054969 M * Bertl yep, I did :) 1171054972 M * baldy fine 1171055003 A * baldy hates perl 1171055019 M * baldy and every other progammer language hehe 1171055024 M * Bertl daniel_hozac: hum ... 1171055065 M * Bertl daniel_hozac: that will always be true? 1171055090 M * Bertl ah, no, sorry 1171055121 M * Bertl so somehow FC is electing a child as reaper? 1171055171 M * daniel_hozac hmm, why won't that always be true? 1171055175 Q * DreamerC Read error: Operation timed out 1171055286 J * DreamerC ~dreamerc@125-225-101-221.dynamic.hinet.net 1171055321 M * daniel_hozac all of father->children should have ->parent == father, shouldn't they? 1171055410 M * Bertl right, so my first shot was correct after all 1171055432 M * Bertl I'm confused by the naming though, as p is always used for any type of process 1171055463 M * daniel_hozac yeah, isn't child usually used for these things? 1171055463 M * Bertl so basically you are disabling the second do .. while() loop 1171055482 M * Bertl and replace reaper with vx_child_reaper() 1171055522 M * Bertl i.e. you could simply cut down the function to 1171055544 J * bonbons ~bonbons@83.222.37.103 1171055547 M * Bertl the last list_for_each_safe() and use vx_child_reaper() 1171055573 M * daniel_hozac yeah. 1171055590 M * Bertl but I wonder, what that will break with threads :) 1171055620 M * Bertl I would suggest to special case the error situation, and only then replace the suggested reaper with the vx_child_reaper() 1171055650 M * Bertl as this will leave certain threading setups intact 1171055696 Q * kaner Ping timeout: 480 seconds 1171056102 M * daniel_hozac hmm, this has to be a host process exiting with a guest child, right? 1171056136 M * daniel_hozac or, hmm. i can't imagine what's causing the BUG_ON. 1171056176 M * daniel_hozac two guest process, the child being the init? 1171056181 M * daniel_hozac +es 1171056187 M * daniel_hozac and the parent exiting 1171056477 Q * DavidS Quit: Leaving. 1171057467 J * xp_prg2 ~xp_prg2@ftp.microvu.com 1171057502 M * xp_prg2 ok, I want to duplicate a vserver, this is my first time doing it, does the vserver have to be stopped? 1171057517 M * Bertl probably better if it is 1171057532 M * Bertl (to have a correct internal state :) 1171057540 M * FaUl is the grsec-vserver patch incl. grsec or does it need grsec as prequisite? 1171057562 M * daniel_hozac it includes grsec, AFAIK. 1171057569 M * Bertl FaUl: you have to ask harry, but IIRC, it is an all-in-one now 1171057586 M * FaUl thx 1171057590 M * FaUl for that answer 1171057620 M * Bertl daniel_hozac: maybe init itself forks in a special way? 1171057637 M * Bertl daniel_hozac: then the 'parent' exits within the start scripts or so? 1171057650 M * daniel_hozac i thinks it's vcontext that's doing the naughty. 1171057655 M * Bertl nah, that would kill any real system :) 1171057688 M * daniel_hozac would it? 1171057705 M * Bertl yes, if init exits after forking, that would panic 1171057720 M * daniel_hozac ah, i thought you meant my vcontext theory. 1171057732 M * Bertl but it could wait for exit of the child though 1171057744 Q * meandtheshell Quit: Leaving. 1171057752 M * Bertl does vcontext fork after setting the init process? 1171057766 M * daniel_hozac no, but before. 1171057775 M * Bertl withing hte context? 1171057787 M * daniel_hozac yep. 1171057791 M * Bertl why? 1171057827 M * daniel_hozac for plain init (--disconnect), the process needs to return. 1171057849 M * Bertl okay, but why inside the context? 1171057861 M * daniel_hozac i'm not positive about inside the context just yet... 1171057888 J * Piet_ hiddenserv@tor.noreply.org 1171057895 M * Bertl okay, np, we'll see ... have to leave now ... will be back later ... 1171057901 N * Bertl Bertl_oO 1171057911 M * TrueBrain Hi! Some time ago I reported here that sched_hard was failing on 2.0.2.1 in the Gentoo repos. I wanted to let you guys know that the 2.2.0-rc9 and 2.2.0-rc11 works fine with sched_hard :) So I am all happy ;) 1171057963 J * kaner kaner@strace.org 1171058073 M * daniel_hozac Bertl_oO: ok, setting initpid and disconnecting is done by the final process in the sequence. 1171058159 Q * Piet Ping timeout: 480 seconds 1171058251 M * daniel_hozac Bertl_oO: i assume because disconnecting before that would mean errors after that point wouldn't get passed to the scripts. 1171058285 M * xp_prg2 vserver postgres8_2_3 build -m debootstrap --context 8020 --hostname postgres8_2_3 --interface postgres8_2_3=eth1:172.16.1.20/24 -- -d etch 1171058308 M * xp_prg2 ok this is the command I used to create the vserver, what is the command to duplicate this and give it another ip address etc...? 1171058345 M * daniel_hozac including changes you've done since installing it? 1171058396 M * xp_prg2 yes 1171058425 M * daniel_hozac depending on your version, vserver ... build -m rsync or vserver ... build -m clone. 1171058441 M * daniel_hozac (the first should be available in 0.30.212+, the latter in 0.30.213-pre4+) 1171058473 M * xp_prg2 vserver --version vserver 0.30.212 -- manages the state of vservers 1171058569 M * xp_prg2 daniel_hozac can you help me to understand the exact command I need? 1171058581 M * daniel_hozac just read vserver ... build --help 1171058593 M * xp_prg2 I should use clone or rsync? 1171058607 M * daniel_hozac given that your version doesn't have clone, rsync seems like a good idea. 1171058625 M * xp_prg2 heh 1171058661 M * Radiance is it possible to have an interface like tun available in a vserver but not assign it an ip ? 1171058669 M * daniel_hozac no. 1171058691 J * dhansen ~dave@pool-72-90-117-15.ptldor.fios.verizon.net 1171058692 M * hallyn Bertl: . 1171058701 M * hallyn uh, 1171058703 M * daniel_hozac _oO 1171058704 M * daniel_hozac ;) 1171058704 M * hallyn Bertl_oO: . 1171058706 M * hallyn yeah 1171058713 M * hallyn will try later i guess - thx 1171058733 M * Radiance i can make it available but openvpn will complain that the interface is already in use and tries to go to tun1... but tun1 doesn't exist, and if i make tun1 available then i am required to give it an ip in the vserver which will start the shit again :) 1171058738 M * Radiance Hi btw :) 1171058778 M * daniel_hozac and why doesn't OpenVPN just use tun0? 1171058862 M * Radiance because rtnetlink says it exists already 1171058867 M * Radiance so openvpn exists 1171058907 M * Radiance if i say dev tun in server.conf openvpn will automatically check..."oh tun0 is exists already and has an ip ok i'll go to tun1...oh shit tun1 doesn't exist" hehe 1171058918 M * daniel_hozac so tell it to use tun0? 1171058984 M * nebuchadnezzar I saw a howto openvpn+Vserver somewhere 1171058990 M * nebuchadnezzar maybe google now about it :-) 1171059015 M * Radiance trust me, whatever howto it is i saw it :) 1171059032 M * nebuchadnezzar ok 1171059040 M * Radiance i'm just trying to make something working using tun mode, talked to derjohn about it and i think it would be a nice addition to the wiki too 1171059117 M * Radiance daniel_hozac, yes i can tell it to use indeed tun0 , would cause a problem if tun0 has already an ip ? 1171059120 M * Radiance well 1171059123 M * Radiance i did it ... 1171059131 M * Radiance here is the result when i do an ip a: 1171059143 M * Radiance inet 10.88.88.1/24 brd 10.88.88.255 scope global tun0 1171059143 M * Radiance inet 10.88.88.1 peer 10.88.88.2/32 scope global tun0 1171059174 M * Radiance i assume the first is the one i assigned to the interface and started the vserver, the second is the one assigned by openvpn which is now listening in server mode 1171059210 M * Radiance lemme see if this nastiness works :) 1171059466 M * Radiance looks like it went fine, authenticated 1171059477 M * Radiance time to get the routing working 1171059699 M * Radiance hummm, From 10.88.88.1 icmp_seq=1 Destination Host Unreachable 1171059753 M * Radiance can't ping each other, but using tcpdump on the server side i can see that something is received, length 60 packet ? icmp ? 1171059986 M * Radiance client has ip 10.88.88.6 / 255.255.255.252 , the server has 10.88.88.1 255.255.255.0 1171060035 M * Radiance this is weird, inet 10.88.88.1 peer 10.88.88.2/32 scope global tun0, peer is the client right ? so why has it 10.88.88.2 instead of 10.88.88.6 which is shown in windows by ipconfig ? 1171060064 M * daniel_hozac because your client assigned the wrong address? 1171060099 M * daniel_hozac they don't even seem to agree on the type of addressing used. 1171060140 M * Radiance well openvpn is running in server mode, so it should give out ip addresses to the client 1171060153 M * Radiance server 10.88.88.0 255.255.255.0 (this is in the server.conf file) 1171060180 M * Radiance in the client.conf i have no manual setting for a specific ip (ifconfig is not used) 1171060250 M * Radiance unless something evil....(lemme see) 1171060933 Q * phreak`` Quit: leaving 1171061053 J * phreak`` ~phreak``@deimos.barfoo.org 1171061334 Q * bonbons Quit: Leaving 1171061350 M * daniel_hozac Bertl_oO: who is the guest init's parent normally? child_reaper? 1171062056 M * Radiance looks like the win32 tap driver has a limitation with tun causing it use specific pairs within a subnet, but anyway, they're now connected...but when i ping from each side, i can see packets are received on both sides, but ping output shows no reply 1171062083 M * Radiance yet tcpdump and monitoring packets on client and server show that the icmp packet is received 1171062223 M * daniel_hozac Bertl_oO: does http://paste.linux-vserver.org/1126 look better? 1171062596 Q * dna Quit: Verlassend 1171062799 J * olivierk_ ~olivier@olivierk.org 1171062832 Q * olivierk Read error: Operation timed out 1171063336 J * meandtheshel1 ~markus@85-124-38-46.dynamic.xdsl-line.inode.at 1171063497 Q * ensc Killed (NickServ (GHOST command used by ensc_)) 1171063507 J * ensc ~irc-ensc@p54B4ECC7.dip.t-dialin.net 1171064802 N * Bertl_oO Bertl 1171064806 M * Bertl back now ... 1171064829 M * baldy Bertl: can u help me with a perl prob? 1171064839 Q * duckx Remote host closed the connection 1171064846 M * waldi i think its time to resubmit the followint patch: 1171064848 M * waldi http://svn.debian.org/wsvn/kernel/dists/trunk/linux-2.6/debian/patches/features/all/vserver/bindmount-dev.patch?op=file&rev=0&sc=0 1171064859 M * Bertl baldy: a Linux-VServer related perl issue? 1171064888 M * baldy nope ;) 1171064892 M * xp_prg2 Bertl is it possible to just copy the /etc/vserver/vserverdir into another dir and change the config files to duplicate a vserver? 1171064942 M * Bertl baldy: sorry then, besides that I hate perl, that would just add to my already tight schedule, and cost valuable Linux-VServer time ... 1171064967 M * Bertl xp_prg2: yes, you just have to get all the symlinks right 1171064984 M * Bertl daniel_hozac: no, doesn't look better to me :) 1171065015 M * xp_prg2 Bertl the some links where? 1171065015 M * daniel_hozac how come? 1171065020 M * xp_prg2 symlinks I meant 1171065057 M * Bertl the config contains symlinks to the guest and various other stuff 1171065139 M * baldy Bertl: oke no prob 1171065150 M * daniel_hozac Bertl: did i forget some important case again? 1171065159 M * xp_prg2 what files are symlinked? 1171065163 M * Bertl no, but it just looks weird ... 1171065171 M * Bertl (that was for daniel_hozac) 1171065186 M * Bertl xp_prg2: just look at the config you want to copy 1171065200 M * Bertl xp_prg2: you should see _what_ is a symlink and what not, no? 1171065237 M * xp_prg2 ok 1171065270 M * daniel_hozac Bertl: sure. 1171065292 M * Bertl first, why child_reaper and not vx_child_reaper()? 1171065304 M * daniel_hozac hehe, that was my next question :) 1171065336 M * Bertl second, why not put the special case into choose_new_parent() 1171065346 M * Bertl something like: 1171065383 M * Bertl if (p == reaper) 1171065403 M * Bertl reaper = vx_child_reaper(p); 1171065425 M * Bertl and remove the BUG_ON() part with p == reaper 1171065447 M * Bertl still I think the entire case _is_ a bug somewhere 1171065470 M * Bertl as a process cannot really reap itself IMHO 1171065474 M * daniel_hozac well, init having a parent is certainly an odd scenario. 1171065593 M * Bertl well, let me put it this way, if that really is the cause, why doesn't it happen on mainline?