1140652806 M * mugwump mnmr: you've never had a process stuck in D state that kill -9 doesn't kill? how long have you been using Linux? :) 1140652834 M * mnmr uhm, since.. 2000 or so, can't recall exactly. 1140652839 M * Aiken mnmr I have never been able to kill -9 a D state process 1140652913 M * mnmr hum.. well, I've just never seen it before, and certainly not from apache. 1140653012 M * mugwump I've lost count of the number of times I've seen bugs like that, even on stock kernels 1140653023 M * Aiken I think D state is wating for some IO to finish. 1140653045 M * mugwump D Uninterruptible sleep (usually IO) 1140653046 M * ebiederm D state is a short term disk wait (and so doesn't setup so the process will exit from the kernel if it receives a signal). Normally the D state last such a short while it is rarely observed. 1140653089 M * mugwump basically whenever you're in an uninterruptable system call 1140653103 M * mnmr are you saying it's normal (if rare) behavior for a process to get stuck like that? like, if I restart apache at a bad time? 1140653113 M * mnmr to me it's a bug.. 1140653129 M * Doener it is a bug... 1140653129 M * mugwump yes, it is a kernel bug. 1140653136 M * ebiederm It is never correct behaviour for a process to stay for a noticable amount of time in the D state. 1140653157 M * Doener the only times i've seen such stuck processes were when disk failures happened... i.e. you might want to check your logs ;) 1140653180 M * ebiederm mugwump: How has your patch play gone? 1140653211 M * mugwump ebiederm: still working on getting the test suite going, otherwise I'll go nuts with the debugging 1140653230 M * ebiederm Good plan. 1140653239 M * mnmr trust me, there are no disk failures, I know that even without looking. brand new setup and it's been working flawlessly up until yesterday where I rebooted with 2.6.15.4-vs2.1.1-rc7 1140653274 M * mnmr my guess is that either of those two components are broken, most likely its the vs patch. 1140653326 M * mugwump is the process still running? 1140653330 M * ebiederm Have you looked at what wchan reports for the stuck process? 1140653357 M * mnmr I rebooted the server. It's a prod server, can't have the apache guest not running! 1140653388 M * mugwump which mpm are you using, mnmr ? 1140653395 M * mnmr prefork 1140653411 M * ebiederm I'm off for a bit... 1140653414 M * mugwump ok. and which kernel version did you say was stable beforehand? 1140653415 N * ebiederm ebiederm_oO 1140653433 M * mnmr 2.6.15 (plain, with vs2.1.0.4) 1140653489 J * dothebart ~willi@xdsl-213-196-225-82.netcologne.de 1140653490 M * mnmr bertl recommended I went from that to 2.6.16-rc4 or 2.6.15.4 (either of them with vs2.1.1-rc7), but 2.6.16-rc4 had other issues (couldnt get apache to server any local files) 1140653544 M * mugwump ok. sadly, we would need some debugging information which is impossible to get now in order to triage this bug. 1140653576 M * mugwump ie, what system call was the process running when it went into D state (available via strace -p XXX) 1140653585 M * mnmr I would've helped find it, but couldn't wait around.. my patience is limited when services are down :) 1140653615 M * mnmr I'll run an strace next time it happens (7-9-13).. 1140653782 M * mugwump knowing things like what that child was doing (eg, serving a request, any system calls the handler does for that path, etc) can help a lot, too 1140653827 M * mugwump another thing you could have done was change the XID for that context, then it could have started again 1140653832 M * mugwump on the new XID 1140653950 M * mnmr you mean by editing the context id somewhere inside /etc/vserver/guest? 1140653954 Q * tudenbart Ping timeout: 480 seconds 1140653955 M * Doener unless using tagxid 1140654466 J * mkhl mkhl@200-153-181-248.dsl.telesp.net.br 1140654533 M * mugwump mnmr: yes, in that config area. as Doener points out, if you are using the XID for anything important this might not be practical 1140654701 M * mnmr I don't think I am.. did even have it enabled in the old kernel, afaicr. 1140654708 M * mnmr dint, that is. 1140654726 M * mnmr didn't (sheesh; something is eating my keystrokes :) ) 1140655702 M * Skram haha 1140655728 A * Skram is hungry, sowwy, mnmr 1140656075 Q * matta Ping timeout: 480 seconds 1140658200 P * mnmr 1140658355 J * vrwttnmtu ~eryktyktu@82-69-161-137.dsl.in-addr.zen.co.uk 1140658367 M * vrwttnmtu Hey all 1140658370 J * entroposcope ~entroposc@user-0c992og.cable.mindspring.com 1140658566 J * Wandrer ~wandrer@70.88.89.70 1140658642 M * vrwttnmtu Is there a way to hide routes from the host from being seen in a vserver? 1140658650 P * Wandrer 1140658652 J * Wandrer ~wandrer@70.88.89.70 1140658656 P * Wandrer 1140658954 J * Smutje_ ~Smutje@xdsl-84-44-241-194.netcologne.de 1140659064 Q * Smutje Ping timeout: 480 seconds 1140659064 N * Smutje_ Smutje 1140659464 Q * vrwttnmtu Quit: Leaving 1140659646 Q * gerrit Ping timeout: 480 seconds 1140661403 J * fwl ~fwl@83.215.237.1 1140661458 J * brisho ~brians@adsl-66-142-59-236.dsl.kscymo.swbell.net 1140661723 Q * fwl Quit: This computer has gone to sleep 1140661905 Q * brisho Quit: using sirc version 2.211+KSIRC/1.3.11 1140663191 J * matta ~matta@c-68-81-35-243.hsd1.pa.comcast.net 1140664148 J * anonc ~anonc@staffnet.internode.com.au 1140664157 Q * matta Quit: Download Gaim: http://gaim.sourceforge.net/ 1140664225 N * Skram foo 1140664242 N * foo woo 1140664262 N * woo Skram 1140665111 Q * fh Quit: changing servers 1140665589 Q * matti Quit: 8-X 1140668101 J * fh fiona@cheeto.net 1140669304 Q * fh Ping timeout: 480 seconds 1140669877 J * fh fiona@cheeto.net 1140670454 Q * fh Read error: No route to host 1140670667 J * fh fiona@cheeto.net 1140671750 Q * Vudumen Ping timeout: 480 seconds 1140671753 J * Vudumen ~vudumen@perverz.hu 1140674089 J * Hmmmm ~Hmmmm@221.135.51.19 1140674110 M * Hmmmm hi guys 1140674189 J * fwl ~fwl@83.215.237.1 1140674259 M * Hmmmm anyone home? 1140674363 M * mugwump hello 1140674416 M * Hollow morning 1140674453 M * Hmmmm hi i need some help 1140674465 M * Hollow just ask ;) 1140674469 M * Hmmmm iv been using vserver@2.4 for a while now. and i was pretty happy with it 1140674486 M * Hmmmm recently i tried the new version on 2.6 1140674491 M * Hmmmm and im pretty muc hstuck 1140674502 M * Hmmmm i was hoping u folks to tell me where im going wrong 1140674509 M * Hmmmm here's what i did 1140674528 M * Hmmmm installed centos4, download linux 2.6.14.3, patched 1140674547 M * Hmmmm then i got the vserver-util source and compiled and installed it. 1140674557 M * Hollow vserver-utils or util-vserver? 1140674568 M * Hmmmm ran the testme.sh and it showed "Succeeded" in all cases 1140674571 Q * Loki|muh Ping timeout: 480 seconds 1140674583 M * Hmmmm util-vserver-0.30.209 1140674585 M * Hollow ok, then it was util-vserver probably :) 1140674589 M * Hollow ok 1140674603 M * Hmmmm did i get it all right so far? 1140674608 M * Hollow yeah 1140674623 M * Hmmmm now what i was doing in the 2.4 version was this 1140674640 M * Hmmmm vserver vstemplate build 1140674662 M * Hmmmm what i need is to build many images of the base system, so that pretty much did the trick for me 1140674673 M * Hmmmm on the new version of vservers that command didn't work 1140674692 M * Hollow the command is still the same, but with some additions, depending on the distribution you want to install 1140674694 M * Hmmmm so i tried '-m legacy' and that worked, but then i couldn't shut down the vserver... 1140674696 M * Hollow (for the guest) 1140674714 M * Hollow i'm not too familar with all the -m switches of build 1140674727 M * Hmmmm Hollow, what do u think is the method that i should use if i do NOt want to download the image data from the internet 1140674728 M * Hollow but there are howtos for nearly all distros 1140674735 M * Hollow and how to install them as vserver guest 1140674753 M * Hollow well, which distro? 1140674761 M * Hmmmm im on centos 1140674768 M * Hollow and you want centos for guests too? 1140674799 M * Hmmmm yup, i ust want the vservers to be ditto copies of the base OS instalation 1140674823 M * cehteh derjohn? 1140674858 M * Hollow never heard of duplicating the base system, and i wouldn't do that as well 1140674868 M * mugwump the original vserver utilities worked like that 1140674887 M * mugwump they'd just cp -r / /vservers/foo with some exclusions 1140674896 M * Hmmmm well, then what is the best alternative? considering that i have NO internet access on that machine 1140674896 M * Hollow but look at http://linux-vserver.org/CentOS_HowTo paragraph C, it should help you with centos integration 1140674907 M * Hollow hm 1140674927 M * Hmmmm checking 1140674934 M * Hollow well, it needs internet.. 1140674967 M * Hmmmm yup thats been my problem so far 1140674990 M * Hmmmm that's why i came to this channel, hopiong that herbert or someone else might be able to advise me 1140675084 M * Hollow well, i'm sure you can just copy the base system, but you'll probbaly need to adjust init scripts or so 1140675148 N * Bertl_zZ Bertl 1140675152 M * Bertl morning! 1140675172 M * Hollow morning Bertl 1140675245 M * mugwump Bertl: is there a way to get qemu to notice that a guest image has shut down? 1140675365 M * mugwump http://pastebin.com/568152 # check that out :) 1140675448 M * Hmmmm hi Bertl 1140675459 M * Hmmmm can you help me? 1140675643 M * Bertl Hmmmm: in what way? 1140675825 M * Bertl mugwump: well, you mean so that qemu exits once the guest is done? 1140675848 M * mugwump yes, that's it 1140675857 M * Hmmmm Bertl, i want to create vservers without using the internet 1140675868 M * Bertl mugwump: I have that on my todo list :) 1140675872 M * Hmmmm basically i want my vservers to be images of my base installation 1140675875 M * mugwump aha 1140675877 M * Bertl Hmmmm: where is the problem? 1140675889 M * Hmmmm instead of downloading packages from the net 1140675912 M * Bertl Hmmmm: steps are pretty easy: 1140675926 M * Hmmmm which method should i use? 1140675927 M * Bertl - make a directoy (e.g. template01) 1140675931 M * Hmmmm ok 1140675940 M * Hmmmm mkdir /vservers/template01 1140675951 M * Bertl - copy over all your stuff you want to see in a guest (with cp, tar, dump, rsync, whatever) 1140675958 M * Hmmmm ok 1140675973 M * Hmmmm so should i cp -r /usr/ ? 1140675988 M * Bertl - cleanup the template if necessary (e.g. remove unused stuff like ekrnels and hardware drivers, etc) 1140676010 M * Hmmmm one sec 1140676020 M * Bertl that was preparation, now the actual guest install: 1140676033 M * Hmmmm in the 2.4 version of vserver i would use "vserver vstemplate build" that seemed to do the copying 1140676038 M * Bertl - create a new guest with the -m skeleton build method 1140676058 M * Hmmmm how exactly would i do the copying here in the new version? 1140676059 M * Bertl (make sure to add all --interface and --hostname stuff) 1140676083 M * Bertl - copy over all directories from the template01 except for /dev and /proc 1140676091 M * Bertl that's it 1140676103 M * Hmmmm but how do i get all the stuff into template01? 1140676120 M * Hmmmm do i have to "cp -r " all the directories fro mthe base system myself? 1140676121 M * Bertl I prefer to use dump/restore for that 1140676138 M * Hmmmm could u please give me a syntax for that? 1140676150 M * Bertl e.g. dump 0f - / | (cd /path/to/template && restore rf -) 1140676173 M * Bertl (if you have more partitions, e.g. /usr /var then do the same for them) 1140676179 M * Hmmmm ok 1140676193 M * Bertl e.g. dump 0f - /usr | (cd /path/to/template/usr && restore rf -) 1140676226 M * Bertl don't forgot to cleanup the template, as this will save you time and space 1140676240 M * Hmmmm ok 1140676250 M * Bertl mugwump: but maybe something like that was already added 1140676250 M * Hmmmm trying 1140676346 M * Bertl mugwump: # In order to exit cleanly from qemu, you can do a shutdown inside qemu. qemu will automatically exit when the Linux shutdown is done. 1140676364 M * Bertl (from fabrice' docu page) 1140676377 M * mugwump hmm, wonder why it isn't working for me. nevermind 1140676390 M * Bertl mugwump: guess it requires acpi or apm 1140676403 M * Bertl mugwump: you are working on automated testing, right? 1140676413 M * Bertl (well that is obvious) 1140676435 M * Bertl mugwump: I had some ideas for that you might like to 'borrow' :) 1140676449 M * mugwump well, I almost have a completely automated test suite :P 1140676538 Q * fwl Quit: This computer has gone to sleep 1140676584 M * Bertl mugwump: ah, well, then ... 1140676605 M * mugwump the only thing I haven't got it doing is harnessing multiple kernel config/patchlevel builds and testing them 1140676625 M * Hmmmm Bertl, i think it's working os far. I'm dumping /usr /var /etc 1140676631 M * mugwump by the time it is doing that, I think I'll start needing a 2X2 system :) 1140676679 M * mugwump anyway, I'm running very late for kung fu ... later folks 1140676692 M * Bertl cya 1140676715 M * Hmmmm Bertl, so what should i do when the dump is finished? 1140676731 M * Bertl move to the next step (cleanup) 1140676754 M * Hmmmm ok 1140676812 M * Hmmmm vserver psycho build -m skeleton --hostname psycho.lk.com --interface eth0:psycho:192.168.1.75 1140676814 J * fwl ~fwl@83.215.237.1 1140676824 M * Hmmmm would this be a correct next step after cleanup? 1140676837 M * Bertl you should at least add --context to that 1140676851 M * Bertl but yes, that looks good so far 1140676884 M * Hmmmm what's the xid? 1140676908 M * Bertl the context number, it is an unique identifier between 2 and 49151 (for now) 1140676932 M * Hmmmm and how do i know what it it? 1140676936 M * Bertl choose one, and keep it unique for your guest 1140676960 M * Hmmmm so its basically something i just choose? like 500 for example? 1140676967 M * Bertl yup, precisely 1140676976 M * Hmmmm ah okay, may i ask what it does? 1140677000 M * Bertl it is the number which is used to create the guest context and if tagging is enabled to tag the files 1140677006 M * Hmmmm ic 1140677011 M * Hmmmm vserver psycho build -m skeleton --hostname psycho.lk.com --interface eth0:psycho:192.168.1.75 --context 500 1140677015 M * Hmmmm that looks okay? 1140677037 M * Bertl you should consider this small change: 1140677046 M * Bertl --interface eth0:psycho:192.168.1.75/24 1140677055 M * Bertl or, if you are used to using ifconfig 1140677062 M * Bertl --interface karli=eth0:psycho:192.168.1.75/24 1140677071 M * Hmmmm ic 1140677080 M * Bertl which will make an eth0:karli (which can be seen by ifconfig) 1140677092 M * Bertl ahem 1140677098 M * Bertl just noticed the psycho there 1140677100 M * Hmmmm got it 1140677104 M * Hmmmm :) 1140677111 M * Bertl --interface psycho=eth0:192.168.1.75/24 1140677116 M * Bertl that's the correct one 1140677122 M * Hmmmm righto 1140677169 M * Hmmmm ok im gonna skip the cleanup so i can run thru the entire process with u 1140677178 M * Hmmmm the ni'll do it again when i get more time 1140677213 M * Bertl k, the skeleton creation will be quite fast 1140677231 M * Hmmmm it was 1140677235 M * Hmmmm next? 1140677242 M * Bertl after that, you can use 'cp -a' or rsync or whatever 1140677252 M * Bertl to copy over all except for /proc and /dev 1140677264 M * Bertl (you can remove proc and dev fom the template firsttoo) 1140677319 M * Hmmmm "cp -a /vservers/template 01/* /vservers/psyhco" 1140677357 M * Hmmmm is that right? 1140677415 M * Bertl well, the space won't work out this way 1140677430 M * Bertl and the psyhco is probably wrong too 1140677439 M * Hmmmm ya, what abotu the space? 1140677461 M * Bertl shell doesn't like spaces without escaping 1140677480 M * Bertl cp -a /vservers/template\ 01/. /vservers/psyhco/ 1140677480 M * Hmmmm ya got it 1140677501 M * Hmmmm i just rename it to template01 1140677583 M * Hmmmm somethings happening :) 1140677620 J * slither ~doivoid@c220-237-72-130.kelvn1.qld.optusnet.com.au 1140677627 P * slither 1140677714 M * Hmmmm ok its copying, what od i do next? 1140677721 M * Hmmmm er, what do i do next? 1140677745 M * Bertl once it's finished, you start your guest, that's it 1140677759 M * Hmmmm nice. i hope it works. 1140677761 M * Hmmmm thanks a ton 1140677767 M * Bertl you're welcome! 1140677774 M * Hmmmm BTW u hang out here a lot? 1140677794 M * Bertl it will probably write a ton of error messages, because you didn't clean up the tempalte, but it should work nevertheless 1140677813 M * Bertl yes, I hang out here a lot :) 1140677872 M * Hmmmm im working on a load-balancer for my vservers 1140677897 M * Hmmmm when its done i would like to maybe become a contributor to the vserver project 1140677923 M * Bertl would be nice ... 1140677936 M * Hmmmm what kind of help do u folks require? 1140677941 M * Hmmmm i'm good with documentation 1140677947 M * Bertl excellent! 1140677962 M * Bertl wiki always needs improvement ... 1140677972 M * cehteh buyed :) 1140677982 M * Hmmmm wud be glad to help 1140677999 M * Bertl we also plan to move to mediawiki sooner or later, so a cleanup and restructuring could be useful too) 1140678013 M * Hmmmm i'm also going to make a doc wit hthe stuff u explained to me today and add it to the wiki 1140678021 M * cehteh bah, use moinmoin wiki, less PHP, no database hazzles 1140678044 M * Hmmmm ic 1140678048 M * Bertl cehteh: I'm pro php :) 1140678068 M * cehteh heh 1140678081 M * cehteh oh well :) 1140678106 M * Hmmmm Bertl, got some errors 1140678115 M * Hmmmm Error: an inet prefix is expected rather than "boing:192.168.1.74" 1140678130 M * Hmmmm secure-mount: chdir("/tmp"): No such file or directory 1140678131 M * Hmmmm /etc/vservers/boing/fstab:2:1: failed to mount fstab-entry 1140678131 M * Hmmmm find: var/lock: No such file or directory 1140678131 M * Hmmmm Invalid IP number or host name 'boing:192.168.1.74' 1140678155 M * Bertl well, you got the --interface wrong 1140678189 M * Hmmmm ouch! 1140678206 M * Bertl and your guest is missing a /tmp dirt 1140678210 M * Bertl *dir even 1140678215 M * Hmmmm how cna i change it? my vserver is called boing 1140678227 M * Bertl you can edit the config tree 1140678258 M * Bertl (or recreate the guest from your template) 1140678275 M * Hmmmm then i would have to copy the stuff agina, right? 1140678294 M * Bertl yup 1140678329 M * Hmmmm done, got it in /etc/vservers/boing/interfaces/name 1140678353 M * Hmmmm where does the /tmp have to be? 1140678362 M * Hmmmm in /vservers/boing/tmp ? 1140678363 M * Hollow Bertl: btw, has anyone decided on the logo yet? 1140678404 M * Hollow one week of holiday is emerging.. so i could spend some time on the wiki 1140678447 M * Bertl nope, haven't even checked the logo page if it contained all the contributions 1140678460 M * Bertl (which wasn't the case last time I checked) 1140678482 M * Hollow ok 1140678505 J * Aiken_ ~james@tooax8-064.dialup.optusnet.com.au 1140678514 M * Hmmmm Bertl, i still get the same errors 1140678523 M * Hmmmm Error: an inet prefix is expected rather than "boing:192.168.1.74". 1140678552 M * Bertl show me your build command 1140678600 M * Hmmmm vserver boing build -m skeleton --hostname boing.lk.com --interface boing=eth0:boing:192.168.1.74 --context 500 1140678606 M * Bertl now make that 1140678624 M * Bertl --interface boing=eth0:192.168.1.74/24 as I suggested 1140678630 M * Hmmmm ah okay 1140678683 M * Hmmmm i'll just rm -rf boing first 1140678798 M * Hmmmm done. now copying the data 1140678825 M * Hmmmm do i have to explicitly create a /tmp in /vservers/boing/ ? 1140678849 Q * Aiken Ping timeout: 480 seconds 1140678853 M * Bertl well, your template should contain that 1140678860 M * Bertl (as your host does) 1140678870 M * Hmmmm ah okay 1140678911 M * Hmmmm [root@asterisk vservers]# ls -all template01/ | grep tmp 1140678912 M * Hmmmm lrwxrwxrwx 1 root root 10 Feb 23 12:00 tmp -> ../var/tmp 1140678930 M * Bertl hmm, well, better make it just a dir 1140678939 M * Hmmmm i have a feeling that i shud delete it and create a new one 1140678940 M * Hmmmm okay 1140679042 M * Hmmmm so what kind of help do u need with the wiki? 1140679053 M * Hollow well, util-vserver mounts tmpfs to /tmp anyway 1140679064 M * Hmmmm ok 1140679096 M * Bertl ad wiki: well, there are many faqs and howtos on the wiki 1140679115 M * Bertl looking through them and combining information in one place might make sense 1140679121 M * Hmmmm i would really like to redo the flower page :-) 1140679126 M * Hmmmm it really really hurts the eye 1140679138 M * Bertl well, choose a diffferent stylesheet? 1140679138 M * Hollow Hmmmm: you can choose a different css in your browser 1140679164 M * Hmmmm i kow, but why not make the default one a little less hurtful to the eye? 1140679171 M * Bertl Hmmmm: the thing with the flower page is, it is kind of auto generated from the xml 1140679182 M * Hmmmm i have saved a copy of it on my desktop, and iv removed the background 1140679192 M * Hmmmm ah okay 1140679196 M * Bertl Hmmmm: so, you might want to do a new stylesheet for that, and submit it to enrico 1140679201 M * Hollow well, i like the background, but you're probably right ;) 1140679204 M * Hmmmm will do 1140679218 M * Hmmmm Hollow, it don't hurt ur eye? 1140679226 M * Hollow it does, but i nevertheless like it ;) 1140679236 M * Hmmmm :-) 1140679243 M * Bertl btw, weedpage is very nice to use .. only a few minor issues there 1140679287 M * Hmmmm a unification of all vserver documentation will really help a lot 1140679294 M * Hmmmm something like ubuntu 1140679332 M * Hollow Bertl: do you have time to discuss a basic structure of the new wiki this/next week? 1140679341 M * Hmmmm https://wiki.ubuntu.com 1140679380 M * Hmmmm Bertl, tried starting the vserver now. a few less errors, but still no joy 1140679389 M * Bertl Hollow: yeah, we should find some time 1140679389 M * Hmmmm find: var/lock: No such file or directory 1140679389 M * Hmmmm No command given; use '--help' for more information. 1140679414 M * Bertl what does your guest's /var contain? 1140679424 M * Bertl (I assume you made an error with the copy) 1140679429 M * Hmmmm [root@asterisk boing]# ls -all var/ 1140679429 M * Hmmmm total 12 1140679429 M * Hmmmm drwxr-xr-x 3 root root 4096 Feb 23 09:53 . 1140679429 M * Hmmmm drwxr-xr-x 20 root root 4096 Feb 23 09:53 .. 1140679429 M * Hmmmm drwxr-xr-x 5 root root 4096 Feb 23 12:46 run 1140679445 M * Hmmmm yup looks like it 1140679470 M * Hmmmm copying from base machine now 1140679476 M * Bertl that also explains why the symlink failed (/tmp) 1140679519 M * Hmmmm so i gotta dump all dirs 1140679541 Q * anonc Quit: adios 1140679631 M * Hmmmm dump 0f - / | (cd /vservers/template/ && restore rf -) 1140679646 M * Hmmmm thats what im doing so i dont miss anything 1140679647 M * Bertl will copy the root fs 1140679657 M * Hmmmm not a good idea? 1140679657 M * Bertl will not copy mounted filesystems 1140679662 M * Hmmmm ok 1140679676 M * Bertl that's why I also told you to do the same for /usr /var ... :) 1140679683 M * Hmmmm so i should do var usr tmp individually? 1140679701 M * Bertl .o( hmm, why am I talking :) 1140679744 M * Hmmmm okay, starting all over again 1140679744 M * Bertl 07:29 < Bertl> (if you have more partitions, e.g. /usr /var then do the same for them) 1140679747 M * Bertl 07:29 < Hmmmm> ok 1140679750 M * Bertl 07:29 < Bertl> e.g. dump 0f - /usr | (cd /path/to/template/usr && restore rf -) 1140679793 M * Hmmmm did that 1140679836 M * Hmmmm in fact thats what i did the first time over 1140679874 M * Hmmmm btw my / and /usr are mounted on 2 dif partitions. maybe that why it didn't work 1140679894 M * Hmmmm so now im doing : 1140679896 M * Hmmmm dump 0f - /usr | (cd /vservers/template02/ && restore rf -) 1140679897 M * Hmmmm and 1140679902 M * Hmmmm dump 0f - /var | (cd /vservers/template02/ && restore rf -) 1140679910 M * Hmmmm dump 0f - /tmp | (cd /vservers/template02/ && restore rf -) 1140681079 J * jkl eric@c-67-172-156-116.hsd1.co.comcast.net 1140681109 M * Bertl welcome jkl! 1140681128 M * jkl hey bertl! 1140683795 Q * ComplexMind Ping timeout: 480 seconds 1140684082 M * Hmmmm Bertl, i creatde a new skeleton vserver and dunped all the data in it (this time ALL the data) 1140684098 M * Hmmmm however, i get an error regarding /proc 1140684109 M * Hmmmm secure-mount: chdir("/proc"): No such file or directory 1140684110 M * Hmmmm /etc/vservers/boing2/fstab:1:1: failed to mount fstab-entry 1140684110 M * Hmmmm secure-mount: chdir("/dev/pts"): No such file or directory 1140684110 M * Hmmmm /etc/vservers/boing2/fstab:3:1: failed to mount fstab-entry 1140684385 J * vrwttnmtu ~eryktyktu@82-69-161-137.dsl.in-addr.zen.co.uk 1140684445 Q * shedi Quit: Leaving 1140684648 M * Bertl Hmmmm: well, you want to create that directories, as your host has them too 1140684655 M * Bertl s/that/those/ 1140684660 M * vrwttnmtu Hello Bertl 1140684664 M * Hmmmm okay 1140684674 M * Hmmmm i created a /proc 1140684684 M * Hmmmm and i also cretade a /dev/pts 1140684689 M * Bertl Hmmmm: btw, your skeleton (the one you created before) should contain them 1140684702 M * Bertl Hmmmm: so you must have done something wrong there, I guess 1140684720 M * Hmmmm Hmmmmm 1140684738 M * Hmmmm any clue as to what i could have done wrong this tiem over? 1140684766 M * Bertl maybe you removed the contents of the dir after the -m skeleton finished 1140684772 M * Hmmmm vserver boing build -m skeleton --hostname boing.lk.com --interface 1140684773 M * Hmmmm boing=eth0:192.168.1.74/24 1140684791 M * Hmmmm nope i don't think so 1140684799 M * Hmmmm maybe i should just start from scratch 1140684875 M * arnaud hi! 1140684896 J * bonbons ~bonbons@83.222.39.180 1140685038 M * arnaud when i connect from a vserver to another server (tcp, udp...), the source ip is the host ip 1140685047 M * arnaud and not the guest ip 1140685106 M * Bertl arnaud: well, is the guest IP able to reach your server? 1140685123 M * arnaud yes, it's on the same subnet as the host ip 1140685147 M * arnaud (/24) 1140685177 M * Bertl care to give some details regarding kernel/tool version and guest setup? 1140685203 M * Bertl (e.g. testme.sh output would be nice, uploaded somewhere) 1140685393 M * arnaud http://greffier.net/temp/test.sh.txt 1140685418 M * Bertl tx 1140685453 M * arnaud :) 1140685473 M * Bertl btw, there is a newer version of the script, but the output is fine 1140685498 M * Bertl what IPs are assigned to your guest? 1140685681 M * arnaud well, i have a local ip (127.0.0.100) on the first interface and an external/eth0 ip (193.189.143.100) on the second interface 1140685703 M * Bertl well, that pretty much explains it 1140685714 M * arnaud if i exchange the two interfaces the route stays the same 1140685733 M * arnaud and i still connect to external servers by the host ip 1140685754 M * Bertl probably the host IP is marked primary on the interface 1140685762 M * Bertl and the guest is secondary on the same IP 1140685769 M * Bertl s/IP/network/ 1140685777 M * Bertl you have two solutions there: 1140685805 M * Bertl a) use a /32 prefix (will basically make it a host entry) 1140685828 M * Bertl b) set up proper rules to use the 'right' source address on the host 1140685845 M * Bertl (b is a little trickier) 1140685975 A * arnaud is testing 1140686111 M * vrwttnmtu Hey Bertl :) 1140686119 M * Bertl hey vrwttnmtu! 1140686128 J * balbir ~balbir@59.145.136.1 1140686129 M * vrwttnmtu I've set up an amd64 host 1140686136 M * vrwttnmtu And a sarge amd64 guest 1140686138 M * Bertl great! 1140686148 M * vrwttnmtu But when I start the guest, I lose all keyboard input on the host 1140686156 M * vrwttnmtu Caps lock lights, everything 1140686170 M * vrwttnmtu I checked that the gettys were commented out in inittab 1140686177 M * Bertl means that your guest somehow has too much power 1140686191 M * vrwttnmtu I suspect it's the unofficialness of the Debian AMD64 stuff 1140686197 M * vrwttnmtu So I'm going to try it with Gentoo 1140686199 M * Bertl no 1140686201 M * vrwttnmtu Really, how? 1140686226 M * vrwttnmtu It's the same config I use elsewhere 1140686231 M * Bertl this was reported by a few folks lately, I suspect either some kind of brokeness in the setup or a real issue ... 1140686231 M * vrwttnmtu No capabilities 1140686252 M * Bertl (real issue means Linux-VServer kernel problem) 1140686255 M * vrwttnmtu If I run an Gentoo amd64 guest, and it's OK 1140686257 M * vrwttnmtu Oh :/ 1140686276 M * vrwttnmtu 2.6.15-vs2.1.0.4-gentoo 1140686278 M * Bertl so, let's investigate this a little more if you've got some time 1140686278 M * vrwttnmtu Is the host 1140686282 M * vrwttnmtu Yeah, sure 1140686296 M * vrwttnmtu To clarify 1140686299 M * bonbons vrwttnmtu: do you have /dev/tty* device nodes in the guest? 1140686311 M * vrwttnmtu It doesn't lose the keyboard on starting the vserver 1140686316 M * vrwttnmtu But on stopping it 1140686320 M * Bertl first thing, your host is running X, right? 1140686325 M * vrwttnmtu In fact, I'm not sure it's not random 1140686332 M * vrwttnmtu Yep, it's this current machine 1140686359 M * Bertl so, I further assume it doesn't happen when your are not using/running X? 1140686378 M * vrwttnmtu This is the only amd64 box I have 1140686389 M * vrwttnmtu I have 2 x x86 boxes that don't run X 1140686397 M * vrwttnmtu But they're headless 1140686400 M * Bertl let's stay with the x86_64 for now 1140686402 M * vrwttnmtu OK 1140686433 M * Bertl what if you logout of X, and do 'telinit 3' on the host before starting the guest 1140686459 M * Bertl i.e. when X is stopped, does anything bad happen to the keyboard too? 1140686459 M * vrwttnmtu If I start the vserver from a console login, it seems to be OK 1140686465 M * vrwttnmtu It seems to be X related 1140686469 M * Bertl okay, good ... 1140686476 M * vrwttnmtu Want me to drop to non-X and try it? 1140686492 M * Bertl well, if you know that already, then no need 1140686505 M * vrwttnmtu I haven't tried it too many times, but it seems OK 1140686523 M * Bertl okay, next step is to upload the following information somewhere (e.g. pastebin) 1140686534 M * Bertl - output of the most recent testme.sh script 1140686547 M * Bertl - output of 'vserver-info - SYSINFO' 1140686560 M * arnaud so... modifying the prefix doesn't change anything 1140686565 M * arnaud source ip is still the host ip 1140686569 M * Bertl - output of 'ls -la /path/to/guest/dev' 1140686586 M * Bertl arnaud: does it show up as primary in 'ip a ls' ? 1140686634 M * arnaud it shows up as the only eth0 ip 1140686649 M * Bertl hmm? 1140686664 M * Bertl I meant on the host :) 1140686665 M * vrwttnmtu Bertl, just doing that, and I notice that it's got the whole /dev full - lots of entries 1140686678 M * bonbons Bertl: one thing that causes X troubles (in relation to keyboard) are partial console-switches, I had this issue on my laptop when starting guests at system startup, immediately followed by X startup 1140686706 M * Bertl vrwttnmtu: see that's probably the issue here ... 1140686718 M * vrwttnmtu SYSINFO: http://rafb.net/paste/results/IO8HV687.html 1140686732 M * vrwttnmtu Shall I nuke it? Or would you like a copy of it first in case? 1140686754 M * Bertl vrwttnmtu: best to build a skeleton (with the -m skeleton build method) and use the /dev from there 1140686778 M * Bertl well, if it contains more than 10 entries, it's very likely to be wrong 1140686784 M * bonbons on my new install I can start guests & X together (XOrg 7.0), previous system was XOrg6.8 1140686814 M * Bertl bonbons: guests should not be able to do console switches 1140686923 M * bonbons Bertl: util-vserver may have had it's part in the trouble (no trouble on my vserver-utils + XOrg7.0 setup) 1140686967 M * Bertl bonbons: well, no idea how that would/could conflict with X 1140686985 M * arnaud Bertl, the ip appears after the host ip (which has a /24 mask), with a /32 mask : "inet 193.189.143.100/32 scope global eth0" 1140687015 M * Bertl ah, good, so both have primary status 1140687034 M * Bertl now make sure that the 193.189.143.100 is the _first_ ip assigned to your guest 1140687036 M * bonbons but I cannot test this anymore, the HDD in that laptop got unreliable since then. I guess it's because of open /dev/console where init's output goes 1140687053 M * arnaud Bertl, you mean the first in the interfaces/ folder? 1140687055 M * vrwttnmtu Bertl, Sorted out my /dev a little, and been in and out of the vserver a few times, and it seems OK 1140687076 M * Bertl bonbons: well, there is no /dev/console (except for the current tty) 1140687101 M * Bertl arnaud: yes, the one with the lowest lexical order 1140687153 M * Bertl okay, have to reboot my workstation (after 45 days uptime) because I need to swap the disk :) ... will be back shortly 1140687160 N * Bertl Bertl_oO 1140687199 M * cehteh .. no hotswap? :) 1140687220 M * bonbons don't know what exactly went on, the guests generated some output to display, X starting somewhere at same time, starting X after guest were up and running was not an issue 1140687255 Q * fwl Quit: This computer has gone to sleep 1140687585 J * Mr_Smoke ~smokey@marla.spinning-plates.eu.org 1140687614 M * Mr_Smoke good morning here 1140687625 M * Mr_Smoke bragon, booh ;) 1140687762 N * Bertl_oO Bertl 1140687764 M * Bertl back now ... 1140687769 M * Mr_Smoke I just bumped into "chbind: vc_net_create(): Invalid argument 1140687769 M * Mr_Smoke " 1140687773 M * Mr_Smoke hello Bertl 1140687784 M * Bertl hey Mr_Smoke! 1140687784 M * Mr_Smoke google tells me it has to do with dynamic context ... 1140687794 M * Mr_Smoke Bertl, same ppl different places eh ;) 1140687800 M * Bertl what kernel patch/tools do you use? 1140687813 M * arnaud Bertl, so... setting eth0 as the primary interface doesn't change the source ip 1140687814 M * Mr_Smoke 2.6.14.3 / vs 2.1.0 / utils -210 1140687849 M * Bertl Mr_Smoke: okay, any particular reason for the 2.6.14 version? 1140687874 M * Mr_Smoke Bertl, yes, because that was the closest one to also have a grsec patch available 1140687886 M * Bertl ah, i.c. :) 1140687887 M * Mr_Smoke for the moment grsec is disabled though 1140687894 M * Mr_Smoke one problem at a time :) 1140687909 J * fwl ~fwl@83.215.237.2 1140687912 M * Bertl could you give the testme.sh a spin and upload the output? 1140687918 M * Mr_Smoke thing is, in spite of all my good will, I couldn't find much info about dyamic context vs static etc 1140687925 M * Mr_Smoke all green for testme.sh 1140687939 M * Bertl I'm more interested in the 4 lines of info on the top 1140687968 M * Mr_Smoke may I paste here ? 1140687988 M * Bertl if just the info lines (i.e. the non green ones) 1140687995 M * Mr_Smoke sure 1140688001 M * Mr_Smoke Linux 2.6.14.3 #8 Wed Feb 22 13:08:09 CET 2006 i686 1140688001 M * Mr_Smoke Ea 0.30.210 273/glibc (DSa) 1140688007 M * Mr_Smoke VCI: (Tn) 1140688007 M * Mr_Smoke (root@ns33750.ovh.net) 1140688007 M * Mr_Smoke (gcc version 3.3.5 (Debian 1:3.3.5-13)) 1140688026 M * Bertl hmm, seems like you booted the wrong kernel :) 1140688030 M * Mr_Smoke erm ? 1140688058 M * Bertl well, it has no visible? vserver support, funny the script succeeds?! 1140688090 M * Mr_Smoke there isn't any vserver extraversion string indeed, but i can assure you it IS patched 1140688106 M * Bertl I'm more concerned about the VCI info 1140688108 M * Mr_Smoke oh 1140688112 M * Mr_Smoke what is that then ? 1140688126 M * Bertl it tell what options were enabled at compile time 1140688134 M * Mr_Smoke hm 1140688142 M * Bertl (kernel compile time that is) 1140688168 M * Bertl and you're sure the testme.sh succeeds with all tests? 1140688179 M * Mr_Smoke absolutely 1140688190 M * Bertl I assume the grsec/vserver fusion broke something then 1140688221 M * Bertl anyway, most likely you just need to reconfigure the tools with 1140688240 M * Mr_Smoke weird, i got the grsec/vserver merged patch from linux-vserver 1140688264 M * Bertl --enable-apis=NOLEGACY 1140688287 M * Mr_Smoke m'okay ... what about that dynamic/static context ? 1140688299 M * Bertl Mr_Smoke: well, there is no official one ATM, but which one did you get actually? 1140688326 M * Bertl it's quite simple, in the beginning there were dynamic and static context ids 1140688361 M * Mr_Smoke i can't remember which one :( *shame* 1140688362 M * Bertl as the dynamic ones are kind of problematic, they are being removed 1140688367 M * Mr_Smoke ok 1140688383 M * Bertl so it is advised to use static context IDs for every new installation 1140688423 M * Bertl I would suggest you start with a vanilla kernel + vserver patch, get that running and figure the grsec stuff later on 1140688455 J * shedi ~siggi@tolvudeild-204.lhi.is 1140688489 M * Mr_Smoke yeah, that would make sense 1140688526 M * Bertl somebody (IIRC harry) is working on a grsec/vserver fusion, so you might team up with him 1140688568 M * Mr_Smoke vserver.limehous.org => that's the one I got I believe 1140688582 M * Bertl never heard of it ... 1140688582 M * Mr_Smoke DaCa's 1140688592 M * Bertl ah, DaCa, well ... 1140688607 M * Mr_Smoke hm no it can't be 1140688624 M * Bertl seems 2.6.11.9 is latest there 1140688643 M * Mr_Smoke yep 1140688668 M * Mr_Smoke I'm suddenly wondering If I didn't cook that one on my own, but that'd be weird 1140688686 M * Mr_Smoke i wouldn't expect grsec to patch nicely over a vserver-patched kernel 1140688697 M * Wonka me neither 1140688707 M * Mr_Smoke and I remember making only a dozen manual hunks (with stuff like adding an include here and there) 1140688727 M * Bertl there are certain collisions, but it's nothing tragical ... it was done by several folks in the past 1140688756 J * ComplexMind ~ComplexHo@cpc1-brig3-6-0-cust194.brig.cable.ntl.com 1140688773 M * Bertl welcome ComplexMind! 1140688782 M * ComplexMind hey Bertl :) 1140688827 M * Mr_Smoke right ... will compile a vanilla+vs and see what happens then 1140688845 M * Mr_Smoke Bertl, but you reckon --enable-apis=NOLEGACY alone might help though ? 1140688891 M * Bertl Mr_Smoke: it might, but as your kernel doesn't show proper version information, it's hard to tell :) 1140688909 M * Bertl Mr_Smoke: the tools might have a hard time to figure it too ... 1140688919 M * Mr_Smoke right ... where is it supposed to get that information from ? 1140688939 M * Bertl either via the vserver syscall or from /proc 1140688968 A * Mr_Smoke searches /proc 1140688978 M * Bertl cat /proc/virtual/info 1140688993 M * Mr_Smoke no /proc/virtual :/ 1140689003 M * Bertl see, that's what I meant ... 1140689015 M * Mr_Smoke how strange 1140689147 M * Mr_Smoke looks like the proc entry is there though 1140689151 M * Mr_Smoke in the sources i mean 1140689177 M * Bertl as I said, probably the best thing is to start with a clean patch 1140689223 M * Hmmmm Bertl, can i bug once more? 1140689225 M * arnaud Bertl, on a "normal" configuration, does "ip route" gives the same result for guests and the host? 1140689232 M * Hmmmm i did the whole process from scratch 1140689242 M * Hmmmm i created a template 1140689248 M * Hmmmm o creatde a skeleton vserver 1140689271 M * Hmmmm now im getting a error without any explanation 1140689313 M * Hmmmm [root@asterisk vservers]# vserver boing start 1140689313 M * Hmmmm No command given; use '--help' for more information. 1140689420 M * SiD3WiNDR boing! :D 1140689472 M * Bertl arnaud: you mean inside the guest and outside on the host? 1140689486 M * arnaud yes 1140689515 M * Bertl Hmmmm: hmmm ... that is somewhat strange indeed, try with 'vserver --debug boing start' 1140689542 M * Bertl arnaud: no, it will only show ips assigned to the guest 1140689557 M * Bertl (i.e. routes based on ips assigned to the guest) 1140689606 M * Hmmmm Bertl, i get a long output, anything in particular that i should look for? 1140689624 M * Hmmmm An error occured while executing the vserver startup sequence; when 1140689625 M * Hmmmm there are no other messages, it is very likely that the init-script 1140689625 M * Hmmmm () failed. 1140689625 M * Hmmmm Common causes are: 1140689625 M * Hmmmm * /etc/rc.d/rc on Fedora Core 1 and RH9 fails always; the '\''apt-rpm'\'' build 1140689625 M * Hmmmm method knows how to deal with this, but on existing installations, 1140689627 M * Hmmmm appending '\''true'\'' to this file will help.' 1140689632 M * Hollow vrwttnmtu: FYI, you can use /sbin/MAKEDEV generic-vserver on gentoo 1140689655 M * Hollow (in guests) 1140689782 M * vrwttnmtu Hollow, that's handy to know 1140689819 M * vrwttnmtu Debian, as long as it has null, zero, and the two randoms seems to sort itself out when you boot it 1140689828 M * vrwttnmtu But I still remember that about Gentoo 1140689832 M * Bertl Hmmmm: upload the output somewhere 1140689849 M * bonbons SiD3WiND: did you already write something for IPv6 in guests? I've currently done the add/remove address (pending testing) 1140689861 M * SiD3WiNDR haven't had the time, sorry :) 1140689899 M * SiD3WiNDR can you send me a diff of what you did? 1140689910 M * SiD3WiNDR or let's put it in another way, was it hard? 1140689920 M * SiD3WiNDR cause actually I can only "hack" c, not write it ;) 1140689927 M * SiD3WiNDR so I need examples :> 1140689951 M * Hmmmm http://pastebin.com/568282 1140689971 M * bonbons was not that hard, mainly a matter of getting types right :) 1140690065 M * Hmmmm Bertl, here you go: http://pastebin.com/568283 1140690115 M * harry Bertl: question... 1140690128 M * harry i'm patching ./arch/ppc64/kernel/syscalls.c 1140690135 M * bonbons SiD3WiNDR: you can read my current patch here: http://homepage.internet.lu/brunop/net_ipv6.diff 1140690137 M * Bertl Hmmmm: is this gentoo? 1140690139 M * harry there used to be a sys_olduname 1140690145 M * Mr_Smoke Bertl, I think I've got a clue as to why proc/virtual isn't buikd ... i'll pastebin in a minute 1140690148 M * Hmmmm Bertl, Centos4 1140690154 M * harry it's there in 2.6.14.4, not in 2.6.14.7, but there again in 2.5.15 1140690156 M * harry 2.6.15 1140690162 M * harry any ideas what happened? 1140690169 M * harry i cant find anything in the changelogs 1140690170 M * Bertl Hmmmm: it seems your guest has no init scripts?! 1140690189 M * Hmmmm did i do something wrong again? 1140690193 M * Bertl harry: probably got removed 1140690203 M * harry and added again... 1140690204 M * Mr_Smoke there are a bunch of if'es before that call to proc_vx_init that relate to grsec... will try to place vx_init *before* that, see if it helps 1140690206 M * harry but... why? 1140690222 M * Bertl Hmmmm: no idea, but try to change the init style 1140690280 M * Bertl Mr_Smoke: I doubt that ... 1140690322 M * Hmmmm Bertl, how do i change that? 1140690337 M * Mr_Smoke Bertl, hm, lemme show you just in case 1140690364 M * Bertl Hmmmm: * /etc/vservers//apps/init/style 1140690368 M * Mr_Smoke Bertl, http://pastebin.com/568287 1140690376 M * Bertl Hmmmm: you find that stuff on the Flower Page 1140690389 M * Hmmmm Bertl, okay let me try it 1140690433 M * Bertl Mr_Smoke: hmm, well, yes, the patch is definitely wrong 1140690437 M * Mr_Smoke Bertl, i precisely have grsec PROC_USER and USERGRP enabled ... 1140690476 M * Bertl Mr_Smoke: if it is already wrong there, I'd really suggest to avoid it, as it is unlikely that it was tested 1140690499 M * Mr_Smoke :( 1140690695 M * SiD3WiNDR bonbons: why do you put NB_IPV6ROOT to 16 and the multiply it by 4 later? 1140690697 M * SiD3WiNDR hm 1140690698 M * harry no trace_syscall in ppc in version 2.6.14.7 :s 1140690707 M * SiD3WiNDR nm, I don't even know what NB_IPV6ROOT is I think :p 1140690722 M * Mr_Smoke right, i'll be off ... DSL debugging ... seeya 1140690725 Q * Mr_Smoke Quit: - Ecky ecky ecky ecky pikaaang zoop boing goodem zoo owli zhiv'. - Ni ! - Shhh... 1140690748 M * bonbons Because the data is of int32 size, but one IPv6 address is 128 bits (NB_IPV6ROOT is the max count of IPv6 addresses, bot the count of int32) 1140690811 M * bonbons if there was a uint128_t type that would be cool (but certainly totally unportable) 1140690837 M * SiD3WiNDR ;) 1140690845 M * SiD3WiNDR yea I'm starting to understand a bit :) 1140690858 M * SiD3WiNDR do the tools need to be patched as well? 1140690881 M * bonbons I'm now playing with userspace to have add/remove supportr from there (e.g. testing what I've already done on kernel side) 1140690927 M * bonbons sure tools have to be patched, but hopefully not too much (just add features) 1140691919 Q * fwl Quit: This computer has gone to sleep 1140692067 J * lilalinux ~plasma@80.69.35.186 1140692174 M * vrwttnmtu Bertl, One administrative question: What do you consider the "unique ID" of a vserver? The name, or the context? 1140692209 M * vrwttnmtu I'm assuming there can't be duplicates of either 1140692240 M * vrwttnmtu But if you had a spreadsheet of vservers, and needed some ID for each one, what would it be? 1140692274 M * Bertl definitely the context 1140692298 M * vrwttnmtu OK. I don't pretend to understand why, but if you say so, it's good enough for me 1140692301 M * vrwttnmtu :) 1140692318 M * Bertl well, let's think about it ... 1140692342 M * Bertl what happens when name != context 1140692366 M * Bertl i.e. what if the same name or context can exist twice 1140692399 M * vrwttnmtu I don't follow - you can't have 2 same contexts, can you? 1140692415 M * Bertl for the name, it would mean that the tools have screwed up somewhere 1140692435 M * vrwttnmtu So you can have the same name twice? 1140692450 M * Hmmmm Bertl, SUCCESS... turns out i hadn't copied the /etc 1140692451 M * Bertl as the 'name' is used for the config, which in turn should be identical if the name is the same 1140692465 M * Bertl Hmmmm: ah, congrats! 1140692503 M * Hmmmm Bertl, thanks a ton. for your patienece especially 1140692504 M * Bertl vrwttnmtu: so, having the same name twice is not really possible from the tool perspective, do you agree? 1140692522 M * Bertl Hmmmm: you're welcome! 1140692529 M * vrwttnmtu Bertl, I usually cd /etc ; cp oldvserver newvserver -Rfv 1140692534 M * vrwttnmtu And do the changes manually 1140692546 M * vrwttnmtu As I can never remember the long command to create :) 1140692559 M * vrwttnmtu So occasionally, I mess up and forget to change a context 1140692561 M * vrwttnmtu or something 1140692575 M * Hmmmm Bertl, now i'll create a do of this process. I would like to have it on the vserver site or wiki so that other ignoramus like myself can use it 1140692584 M * Bertl vrwttnmtu: yup, but you never messed uo the name, right? 1140692588 M * Bertl *up 1140692601 M * vrwttnmtu I'm sure I have at some stage or another 1140692601 M * Bertl Hmmmm: go ahead ... 1140692619 M * Bertl vrwttnmtu: how? the filesystem should make it impossible :) 1140692646 M * vrwttnmtu cat /etc/vservers/anonet-debian/name 1140692647 M * vrwttnmtu anonet 1140692651 M * vrwttnmtu Isn't that the name? 1140692736 M * Hmmmm Bertl, looks like the /root hasnt been created. how do i do that? should i just copy the /root for the base machine? 1140692772 M * Bertl Hmmmm: yes, you can 1140692845 M * Bertl Hmmmm: regarding wiki, keep a few things in mind though: first, search if the process isn't already described somewhere. second, make sure to use the preview, as every page change gets mailed to the wiki mailing list. third, make sure to use similar style than on existing pages 1140692869 M * Hmmmm Bertl, okay will keep all that in mind 1140692871 M * Bertl vrwttnmtu: well, I'd say anonet-debian is the name 1140692884 M * Bertl vrwttnmtu: or how do you start it? 1140692893 M * vrwttnmtu vserver-stat disagrees 1140692895 M * vrwttnmtu :) 1140692901 M * vrwttnmtu # vserver-stat 1140692901 M * vrwttnmtu CTX PROC VSZ RSS userTIME sysTIME UPTIME NAME 1140692901 M * vrwttnmtu 0 120 700.4M 912.6M 1h08m29 8m41s18 2h37m25 root server 1140692901 M * vrwttnmtu 8482 6 24.5M 2.3M 0m00s11 0m00s17 1h36m28 anonet 1140692915 M * Bertl so how do you start it? 1140692928 M * vrwttnmtu vserver anonet-debian start 1140692932 M * Bertl see? 1140692936 M * vrwttnmtu Yep 1140692944 M * Bertl so we actually have three identifiers here 1140692970 M * Bertl the first one is the guest name as used by the tools 1140692981 M * Bertl the second one is the xid (context id) 1140693006 M * Bertl and the third one is an entry which gets displayed in vserver-stat (but has no meaning whatsoever) 1140693012 M * vrwttnmtu Aaah :) 1140693028 M * vrwttnmtu Perhaps it would be nice to make the 3rd match the 1st 1140693029 M * vrwttnmtu ? :) 1140693036 M * Bertl now, certainly it would be much better to have just one of them 1140693069 M * Bertl I think, the third one was just a work-around as many folks complained that vserver-stat only shows the xid 1140693090 M * Bertl and I'm pretty sure if you _use the tools_ to create guests, they are in perfect sync :) 1140693098 M * vrwttnmtu So the /etc/vservers/one-name/name file is only used by vserver-stat? 1140693126 M * Bertl I don't know where it is used, but it certainly look like 1140693140 A * Bertl didn't even know it exists :) 1140693140 M * vrwttnmtu It is 1140693143 M * vrwttnmtu How weird 1140693149 Q * ComplexMind Ping timeout: 480 seconds 1140693159 M * vrwttnmtu I've just stopped, deleted that file, started, and the name is now the "right" name 1140693165 M * Bertl so, back to the original question ... 1140693188 M * Bertl we already agreed that the 'tool' name is unique in the filesystem 1140693192 M * vrwttnmtu Yep 1140693195 M * vrwttnmtu I concur 1140693205 M * Bertl (as you can hardly have two entries in /etc/vservers) 1140693245 M * Bertl now, for the kernel, a string is not very handy to reference a context 1140693256 M * Bertl and because of that (and for legacy reasons) the context ID is used 1140693272 M * Bertl the context ID is unique in the kernel space, i.e. the very same numer identifies the very same context 1140693301 M * vrwttnmtu OK. Context Id it is 1140693309 M * vrwttnmtu Can't argue with the kernel 1140693310 M * vrwttnmtu :) 1140693334 M * vrwttnmtu Is it possible to remove the reading of /etc/vservers/*/name 1140693336 M * vrwttnmtu ? 1140693337 Q * Aiken_ Quit: Leaving 1140693357 M * Bertl vrwttnmtu: probably, but I guess Enrico considers it a feature ... 1140693377 M * vrwttnmtu OK - I'll remove it from my vservers then 1140693757 J * ComplexMind ~ComplexHo@cpc1-brig3-6-0-cust194.brig.cable.ntl.com 1140693847 M * waldi Bertl: which versions should currently work on ppc? 1140693899 M * Bertl *rc9 1140693909 M * waldi okay 1140693911 M * Bertl ppc and ppc64 (untested) 1140694124 M * phreak`` Bertl: will soon test that (got a pSeries here, where I'm gonna install linux in the afternoon) 1140694151 M * Bertl good, please send the testme and testfs results to the list 1140694158 M * phreak`` ok 1140694322 M * waldi Bertl: 2.0.2-rc9 lacks the KTHREAD fix? 1140694341 M * Bertl there should be no KTHREAD in 2.0.2? 1140694346 M * waldi hmm 1140694355 M * Bertl ah, 1140694366 M * Bertl well, it has the fix, but that's wrong :) 1140694379 M * waldi hihi 1140694381 M * Bertl just revert it there, I figured this today, so it's not updated yet 1140694395 M * Bertl (showed up on the plm radar) 1140694395 Q * Hmmmm Quit: Leaving 1140695110 J * philipp ~philipp@84-73-110-118.dclient.hispeed.ch 1140695136 M * Bertl welcome philipp! 1140695142 M * philipp hi there 1140695326 J * Mr_Smoke ~smokey@marla.spinning-plates.eu.org 1140695329 M * Mr_Smoke re 1140695356 M * Mr_Smoke Bertl, that patch I pasted earlier ... it exists in DaCa's patch as well, it's identical 1140695358 M * Bertl wb Mr_Smoke! 1140695361 M * Mr_Smoke (thanx) 1140695379 M * Bertl so it's a bug which was copied without much thinking :) 1140695414 M * Mr_Smoke hum, you mean DaCa's 2.6.11 is bugged then ? 1140695415 M * Bertl it's definitely no good idea to disable proc initialization :) 1140695421 M * Mr_Smoke it's not disabled 1140695429 M * Mr_Smoke i've taken another look at it 1140695443 M * Bertl # 1140695444 M * Bertl #ifdef CONFIG_GRKERNSEC_PROC_ADD 1140695451 M * Mr_Smoke if the grsec configs are enable (and they are), it should initialize 1140695451 M * Bertl if this is not configured 1140695457 M * Mr_Smoke it is 1140695473 M * Mr_Smoke daria:/usr/src/linux# grep CONFIG_GRKERNSEC_PROC_ADD .config 1140695473 M * Mr_Smoke CONFIG_GRKERNSEC_PROC_ADD=y 1140695478 M * Bertl well, it doesn't make sense to disable the initialization at all 1140695489 M * Mr_Smoke agreed 1140695492 M * Bertl yeah, that's what I mean 1140695496 M * Bertl #ifdef CONFIG_GRKERNSEC_PROC_ADD 1140695504 M * Bertl (which is true in your case) 1140695510 M * Bertl then the # 1140695511 M * Bertl proc_vx_init(); 1140695514 M * Bertl is not called 1140695523 M * Mr_Smoke um 1140695547 M * Bertl only the #else branch calls it 1140695570 M * Mr_Smoke damn 1140695590 M * Mr_Smoke off to see what exactly grkernsec_proc_add does then 1140695619 M * Bertl most likely the # 1140695620 M * Bertl proc_vx_init(); 1140695630 Q * RoadRunnR Read error: Connection reset by peer 1140695631 M * Bertl should be _outside_ the #if 1140695657 M * Bertl moving it from line 12 to 13 (in your paste) should fix this 1140695688 M * Mr_Smoke that's my thinking too 1140695689 M * bonbons Bertl: do you know what may cause one of the nids to appear twice in /proc/virtnet/? (in my case always 12) 1140695698 M * Mr_Smoke but i want to be sure of what that option does 1140695715 M * Bertl bonbons: yes, it's a bug which was fixed in recent versions 1140695729 M * Bertl bonbons: it's harmless though 1140695756 M * bonbons what do you mean by recent? post 2.1.1rc7? 1140695783 M * Bertl http://vserver.13thfloor.at/Devel/PAT-2.1.1/delta-vnet-fix01.diff 1140695825 M * Bertl rc7 should have it 1140695832 M * Mr_Smoke right, i'll try moving proc_vx outside the if then 1140695900 M * Mr_Smoke umpteenth compilation time :) 1140695915 M * Bertl guess it should only take a few seconds 1140695929 M * Mr_Smoke yeah just a proc.o 1140695935 M * Mr_Smoke built-in stuff 1140695970 M * Bertl bonbons: but I see it is missing in 2.6.16-rc4-vs2.1.1-rc9, so obviously it got messed up 1140696012 M * Bertl bonbons: will be in the next release 1140696025 M * Mr_Smoke rebooting ... 1140696035 M * bonbons ok :) will continue with my IPv6 hacks 1140696042 M * Mr_Smoke bonbons, hmmm 1140696057 M * Mr_Smoke bonbons, hacking ipv6 support ? 1140696077 M * Mr_Smoke you've got a client here :) 1140696116 M * Bertl bonbons: if you want me to have a look at the kernel side, just let me know ... 1140696126 M * Mr_Smoke Bertl, got VCI stuff now 1140696148 M * Bertl good start, does it fix your guest issues? 1140696166 M * Mr_Smoke the chbind yeah 1140696180 M * Mr_Smoke now got trouble with proc/uptime, but at least there's a link to some web page with help :) 1140696206 M * bonbons Bertl: currently I have just add/remove of the IP addresses (completed removed of IPv4 and prevented duplicates) 1140696221 M * Bertl Mr_Smoke: http://linux-vserver.org/Flags+and+Caps 1140696239 M * Bertl sorry, http://linux-vserver.org/Caps+and+Flags 1140696256 M * Mr_Smoke cheers 1140696262 M * Mr_Smoke can't find vprocunhide ... duh 1140696269 A * Mr_Smoke searches 1140696282 M * Bertl the tools would have installed that in /etc/init.d/ or so 1140696309 M * Mr_Smoke ah right, /usr/local/etc then 1140696332 M * Mr_Smoke chbind: vc_net_create(): Invalid argument 1140696333 M * Mr_Smoke ..... AUUUGH 1140696373 M * Bertl please update the testme.sh output (v0.15) 1140696389 M * Bertl (now that we have a chance to get some info) 1140696396 M * Mr_Smoke VCI: 0002:0001 273 03000116 (TbLgnP) 1140696420 J * fwl ~fwl@83.215.237.2 1140696447 M * Bertl what was the tools line again? 1140696469 M * Mr_Smoke Ea 0.30.210 273/glibc (DSa) 1140696469 M * Mr_Smoke <= that one ? 1140696494 M * Bertl yup 1140696616 M * Mr_Smoke i think i'll rebuild a vs from scratch, just to make sure 1140696621 M * Bertl your config has dynamic contexts disabled, which should be fine for the tools 1140696623 J * RoadRunnR ~MrRoadRun@213.187.82.17 1140696646 M * Bertl Mr_Smoke: could it be that you actually have two different sets of utils installed? 1140696650 M * Bertl welcome RoadRunnR! 1140696676 M * Bertl Mr_Smoke: i.e. maybe one set from a package or a previous build, and the 0.30.210 in a different location? 1140696679 M * Mr_Smoke Bertl, well, i don't think so, I specifically removed/purged the default debian tools and compiled new ones 1140696799 M * Bertl ah, regarding your guest config, what xid/nid has it set? 1140696997 M * Mr_Smoke i'm rebuilding it atl 1140697000 M * Mr_Smoke atm 1140697014 Q * fwl Quit: This computer has gone to sleep 1140697030 M * daniel_hozac Bertl: pre-0.30.210 doesn't use vc_net_create though. 1140697034 A * Mr_Smoke works on a PocketPC ROM in the meantime 1140697083 M * daniel_hozac bonbons: any reason you're not using the same array for IPv6 addresses as for IPv4 addresses? 1140697131 Q * balbir Quit: Leaving 1140697140 M * daniel_hozac IIRC that was the idea back when the new networking API got created... 1140697141 M * bonbons Yep, how would I make difference between IPv4 addresses and IPv6 addresses? 1140697205 M * Mr_Smoke ah, regarding your guest config, what xid/nid has it set? <= erm ... where is that piece of info stored ? 1140697219 M * daniel_hozac Mr_Smoke: /etc/vservers//context 1140697227 M * Mr_Smoke i'm beginning to think i'm missing default configuration stuff 1140697232 M * bonbons the size of v4 and v6 addresses is different as well, and they would be used separately anyhow. IPv4 choosen to connect to IPv4, IPv6 to connect to IPv6. Only a full bind-any would use both 1140697246 M * Mr_Smoke daniel_hozac, no such directory (context) .. and uts is empty 1140697265 M * daniel_hozac bonbons: yes, but the size of the array is meant to fit 4 IPv6 addresses and 16 IPv6... 1140697278 M * daniel_hozac s/IPv6/IPV4/2 1140697284 J * ebacve icechat5@dynamic-oit-vapornet-a-1767.Princeton.EDU 1140697363 M * bonbons and? look more like XOR... or something inbetween (where it gets hard to say where the IPv6 are and where the IPv4 are) 1140697421 M * bonbons daniel_hozac: assume add IPv4, add IPv6, add IPv6, add IPv4, on next guest you add 3 IPv4 and one IPv6... 1140697477 M * daniel_hozac struct { int type; union { __u32 ipv4[4]; __u32 ipv6[4]; } a; } addrs[NB_IPV4ROOT >> 2]; 1140697500 M * Bertl welcome ebacve! 1140697502 J * tudenbart ~willi@xdsl-213-196-240-40.netcologne.de 1140697517 M * Mr_Smoke Bertl, is it normal that I don't have a context subdir ? or should it be uts ? 1140697528 M * daniel_hozac Mr_Smoke: it's not a directory. it's a file. 1140697531 M * ebacve thanks. im a complete newbie. 1140697532 M * Mr_Smoke oh 1140697539 M * daniel_hozac Mr_Smoke: and it'll only get created if you pass --context to vserver ... build. 1140697540 M * bonbons ok, that way would by fine 1140697549 M * Mr_Smoke daniel_hozac, ah, that would explain it 1140697554 M * Bertl ebacve: well, that's usually not a problem :) 1140697556 M * Mr_Smoke so far the tutorials didn't make use of that line 1140697572 M * daniel_hozac bonbons: i guess even better would be a bit field for the types... 1140697572 Q * mkhl Quit: 1140697592 M * Bertl Mr_Smoke: that basically legacy config 1140697611 M * Bertl Mr_Smoke: and you are using a non-legacy kernel (i.e. dynamic contexts and such disabled) 1140697634 M * Mr_Smoke Bertl, right ... so the context info is ...where ? 1140697647 M * Bertl it should be in /etc/vservers//context 1140697655 M * Bertl you can add the missing info with e.g. 1140697662 M * Bertl echo "1001" >/etc/vservers//context 1140697671 M * Bertl (which will use xid=1001 for your guest) 1140697671 M * Mr_Smoke right 1140697684 M * bonbons daniel_hozac: yep, for now I will try to get my version working, optimization will come afterwards :) 1140697691 M * Bertl make sure that it is unique, otherwise the guests' contexts will overlap 1140697703 M * Mr_Smoke Bertl, YAY. vserver booted now :) 1140697703 M * daniel_hozac bonbons: i just want to avoid adding 324 bytes to each nx_info :) 1140697735 M * Mr_Smoke Bertl, i was trying to stick to the tutorials, none of them mentions that one has to define the xid manually ... Cheers :) 1140697748 M * daniel_hozac Mr_Smoke: please fix that 1140697758 J * fwl ~fwl@83.215.237.2 1140697760 M * daniel_hozac whereever you see the docs suggesting to use dynamic xids, fix it. 1140697771 M * Bertl daniel_hozac, bonbons: when we really add ipv6 to the 'old style' networking, we have to make the entries dynamic 1140697785 M * daniel_hozac Bertl: what do you mean? 1140697791 M * Mr_Smoke daniel_hozac, well it doesn't suggest anything about dynamic xids actually ... and it's not a wiki, it's the Debian grimoire 1140697792 M * bonbons Bertl: that was also my thought, get the list dynamic 1140697810 M * bonbons allocate the memory only for the addresses used 1140697816 M * daniel_hozac Mr_Smoke: well........ all the official linux-vserver docs are on linux-vserver.org which is a wiki. 1140697816 M * Bertl bonbons: yup 1140697827 M * daniel_hozac oh, you meant like that, heh. 1140697868 M * Bertl well, the debian docu is _official_ for debian I guess, but they do not disable dynamic contexts either, I'd assume 1140697901 M * daniel_hozac well, Debian does still use legacy configurations, so they can't disable dynamic contexts. 1140697902 M * Mr_Smoke hm 1140697915 M * Mr_Smoke how come the interface isn't up then... 1140697929 Q * dothebart Ping timeout: 480 seconds 1140697955 M * Bertl Mr_Smoke: chances are good that it _is_ up, and you jsut don't see it :) 1140697963 M * Bertl Mr_Smoke: are you using ifconfig to check? 1140697967 M * ebacve Bertl, i should introduce myself. i'm doing a project with Marc F. (mef), trying to get planetlab to work on windows using colinux 1140697985 M * Bertl ebacve: ah, yeah, I heard from you then :) 1140697988 M * ebacve i wanted to ask if anyone ever tried to get vserver to work with colinux. 1140697991 M * Mr_Smoke Bertl, sorry, I didn't express myself correctly ... it is up, but hasn't taken the ip parameters I had passed when building 1140698011 M * Bertl Mr_Smoke: the question remains, ifconfig? 1140698022 M * Mr_Smoke yes 1140698037 M * Bertl Mr_Smoke: won't see it, too old, use 'ip' from iproute2 1140698044 M * Mr_Smoke ah, apt-get time then 1140698075 M * Bertl alternatively you can make it an old-fashioned alias 1140698075 M * Bertl which will be visible to ifconfig too 1140698085 M * Bertl ebacve: not that I know of ... have you tried yet? 1140698091 M * Mr_Smoke nah, getting iproute2 into the vserver is a good idea 1140698119 M * ebacve bertl, no, i'm working on the basics still, but I will in the near future. 1140698145 M * Bertl bonbons, daniel_hozac: so if you really want to progress there, we should first make it dynamic (i.e. remove NB_IPV4ROOT) 1140698151 M * ebacve first i'll try to get colinux from 2.6.10 (or 2.6.11) to 2.6.12 1140698174 M * Bertl ebacve: lol, would be good to get it to 2.6.16 :) 1140698184 M * daniel_hozac finally a patch that is less updated than grsec! 1140698186 M * daniel_hozac :) 1140698207 M * Bertl nonono! I have a netconsole patch here which wasn't updated since 2.6.8 :) 1140698213 M * daniel_hozac lol 1140698231 M * bonbons Bertl: no problem for having it dynamic (though I currently don't get new IPv6 addresses in, bug somewhere kernel/userspace?-> adding printk()s) 1140698251 M * Bertl bonbons: show me the code! :) 1140698295 M * ebacve i'll try to get the version to match that of planetlab first (2.6.12). 1140698305 M * ebacve then find a vserver patch for that version 1140698324 M * Bertl ebacve: planet lab will soon update, but I guess you know that 1140698338 M * ebacve i feel uninformed:) 1140698346 M * daniel_hozac bonbons: memcpy instead of the 4 assignments? 1140698359 M * ebacve Bertl, to 2.6.16? 1140698384 M * Bertl well, I guess it will be 2.6.15 first, but once it's out/stable probably 2.6.16 too 1140698390 M * ebacve Hmm. 1140698406 M * bonbons Bertl: http://homepage.internet.lu/brunop/net_ipv6.diff (not the very very last version) 1140698429 M * Bertl ebacve: at least last time I talked with mef, he said he will start to migrate to a new devel model based on current releases 1140698442 M * Bertl s/will/plans to/has already/ :) 1140698473 M * ebacve bertl, I'll ask mef what versions he had in mind for me to use. 1140698475 M * bonbons daniel_hozac: is an option, is memcpy inlined? 1140698475 M * Bertl ebacve: but I have no idea on the timeline though ... 1140698482 M * ebacve thanks 1140698576 M * daniel_hozac bonbons: no idea. 1140698601 M * Bertl ebacve: you're welcome! hope I did not cause confusion :) 1140698619 M * ebacve not at all 1140698669 M * Bertl ebacve: could you give me a short statment how colinux works? 1140698844 M * Bertl wow, the interesting detail is, they released a version on Feb 6th, 2006 (that was the one for 2.6.11 :) 1140698872 M * daniel_hozac lol 1140698873 M * waldi mkfs.ext2 /dev/zero 1140698877 M * waldi hihi 1140698883 M * ebacve bertl, colinux is a modified 2.6.10 linux kernel together with a special windows driver. the colinux kernel intercepts function calls from linux apps and executes them by interacting with the hardware through the windows driver. hence, linux apps can run on windows 1140698887 M * Bertl waldi: ah, you are testing? 1140698901 M * waldi yes 1140698903 M * Bertl waldi: requires a loopback or real device 1140698940 M * ebacve (bertl) I see there was a 2.6.11 development (not stable) versyion of colinux. 1140698946 M * Bertl ebacve: what I would like to know is: is it a special arch, like UM or XEN, or some other hack? 1140698959 M * ebacve bertl, a special arch mostly. 1140698980 M * ebacve in addition to some extra linux drivers 1140698981 M * Bertl ebacve: ah, then should combining with Linux-VServer only be a matter of minutes 1140698981 M * waldi Bertl: can't you add an error? 1140699000 M * Bertl waldi: yeah, probably will 1140699007 M * waldi Bertl: 2 fails 1140699015 M * Bertl waldi: ah? where? 1140699025 M * waldi but this is the old kernel with 2.1.1-rc6 or so 1140699026 M * bonbons Bertl: how do the defines in the kernel work? I test #ifdef CONFIG_IPV6 and it doesn't seem to get compiled in though it's in .config (CONFIG_IPV6=m) 1140699030 M * waldi [101]# failed. 1140699034 M * waldi for ext2/ext3 1140699038 M * Bertl not good 1140699041 M * ebacve bertl: that's what I understood from mef. I am unfamiliar to linux kernels though, so everything takes a bit more time with me. 1140699060 M * ebacve but I will speed up as the semester goes by, hopefully 1140699063 M * waldi Bertl: i'll redo the test the 2.0.2-rc9 1140699072 M * daniel_hozac bonbons: i thought that would define CONFIG_IPV6_MODULE or something like that. 1140699072 M * Bertl ebacve: that will come naturally ... 1140699112 M * Bertl bonbons: do you include linux/config.h ? 1140699144 M * bonbons is already included 1140699146 M * Bertl bonbons: and yes, there are two different configs depending on module or monolithic 1140699165 M * Bertl i.e. =y defines CONFIG_IPV6 1140699167 M * bonbons hm, how should I handle those? 1140699175 M * Bertl while =m defines CONFIG_IPV6_MODULE 1140699195 M * Bertl bonbons: accordingly and properly :) 1140699239 M * ebacve bertl, off I go. thanks for your help. I will be back frequently. 1140699248 M * Bertl ebacve: okay, cya! 1140699254 M * waldi Bertl: 2.0.2-rc9 dies before printk setup 1140699255 M * bonbons is #ifdef CONFIG_IPV6 || CONFIG_IPV6_MODULE fine? (don't know how module things are handled "behind the scenes") 1140699275 M * Bertl waldi: not good either ... this is on ppc? 1140699278 M * waldi ppc64 1140699287 M * Mr_Smoke hm 1140699293 M * Mr_Smoke i'm too used to seeing ppc = pocketpc these days 1140699298 M * Mr_Smoke freaky moment 1140699302 M * waldi okay. i go back to 2.1.1 1140699353 M * daniel_hozac bonbons: #if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE) i guess. 1140699363 M * Bertl waldi: hmm hmm ... 1140699397 M * Bertl waldi: could you enable the debug stuff including vserver debug and history tracing? 1140699410 M * bonbons daniel_hozac: Don't know, never used #ifdefs with more than one arg 1140699424 M * Bertl waldi: I mean, stack, spinlock, and such 1140699458 M * waldi Bertl: not yet. but this will not help anything without a working debugger 1140699478 M * Bertl waldi: debugger? console! 1140699537 M * waldi Bertl: how will the console work without working printk? 1140699563 M * Bertl it might spit out more info ... 1140699572 M * Bertl I doubt printk itself fails 1140699576 M * waldi it does 1140699590 M * waldi the first printk does is to output the kernel header 1140699662 M * waldi so it never comes near to the init function 1140699760 M * daniel_hozac umm, does vanilla work? 1140699772 M * waldi yes 1140699787 M * Bertl that's quite interesting ... 1140699791 M * waldi also the same with 2.1.1-rc7 patch worked 1140699797 M * waldi + the powerpc fi 1140699801 M * Bertl waldi: what compiler? 1140699812 M * waldi 4.0 head 1140699827 M * Bertl hmm, maybe that's the rpoblem= 1140699835 M * Bertl eek! 1140699838 M * Bertl *problem 1140699843 M * waldi no 1140699938 M * Bertl I'm not aware that Linux-VServer changes something in the early printk 1140700053 M * daniel_hozac my combination of i686 SMP HIGHMEM64G Fedora Linux-VServer failed in a similar way. 1140700097 M * daniel_hozac i doubt it is in any way vserver's fault though... 1140700107 M * Bertl but vanilla works? 1140700116 M * daniel_hozac if i removed any of the said parts, it worked. 1140700135 M * Bertl hmm, can you test this easily? 1140700158 M * daniel_hozac not anymore... i changed to HIGHMEM4G a couple of months ago. 1140700179 M * Bertl (which works fine) 1140700183 M * daniel_hozac right. 1140700250 M * waldi Bertl: it changes things in the core memory management 1140700255 M * daniel_hozac this is also with a gcc 4.0 compiler. 1140700276 M * Bertl waldi: well, yes, but just accounting, nothing more 1140700299 M * Bertl waldi: and this is not done for xid=0 tasks 1140700423 M * waldi hmm 1140700545 M * Bertl so whatever causes this must either be a very strange side-effect or just a catalyst to some 'other' hidden issue 1140700825 Q * philipp Ping timeout: 480 seconds 1140700993 M * bonbons Bertl: vxdprintk(VXD_CBIT(nid, )) 1140701016 M * bonbons Bertl: vxdprintk(VXD_CBIT(nid, 3)), ...) what's the meaning of VXD_CBIT() part? 1140701101 M * daniel_hozac vserver debug check bit. 1140701123 M * daniel_hozac i.e. if debug bit 3 in nid is set, print the message. 1140701189 J * philipp ~philipp@84-73-110-118.dclient.hispeed.ch 1140701194 Q * ebacve Ping timeout: 480 seconds 1140701349 M * bonbons hmm... and where are those debug bits located? 1140701373 M * Bertl wb philipp! 1140701395 M * Bertl bonbons: they are stored in a global variable 1140701409 M * Bertl you can set them via kernel command line or sysctl 1140701441 M * bonbons ok, some suggestion what values I should use for vxdprintk() in add/remove IP addresses? 1140701466 M * Bertl first, have a look at Documentation/vserver/debug.txt 1140701542 M * Bertl I'd suggest to use bit 6 1140701562 M * bonbons hmm... not so clear to mee... column 3 is format, but column 1 and 2... 1140701578 M * Bertl col 1 = bit #, col 2 = hex value 1140701604 M * Bertl 2^5 = 20h 1140701630 N * ebiederm_oO ebiederm 1140701639 M * Bertl wb ebiederm! 1140701660 M * bonbons so this is listing of what debug output is generatedwith what value? 1140701667 M * Bertl yup 1140701724 M * bonbons well, will look and try after meal :) 1140702019 M * waldi yeah 1140702026 M * waldi 2.1.1-rc9 breaks the same way 1140702035 Q * fwl Ping timeout: 480 seconds 1140702129 M * daniel_hozac rc7->rc9 is pretty much just the vchild_reaper changes that should influence anything. 1140702164 M * Wonka narf. 1140702164 M * Wonka kernel/power/ui.c: In function 'suspend_cleanup_console': 1140702165 M * Wonka kernel/power/ui.c:563: warning: implicit declaration of function 'vx_rmap_pid' 1140702170 M * Wonka WHY? 1140702188 M * daniel_hozac Wonka: #include or #include (depending on devel or stable series) 1140702199 M * daniel_hozac Wonka: it most likely uses find_task_by_pid. 1140702204 M * Wonka it errors in find_task_by_pid() 1140702207 M * Wonka yes 1140702213 M * Wonka which is defined in sched.h 1140702222 M * Wonka which includes vs_base.h 1140702252 M * waldi daniel_hozac: i revert the vchild_reaper change and check again 1140702301 J * fwl ~fwl@83.215.237.1 1140702380 M * Wonka daniel_hozac: thanks. i included in kernel/power/ui.c and it compiles 1140702392 M * Wonka next: 1140702393 M * Wonka include/asm/suspend2.h: In function 'fix_processor_context': 1140702393 M * Wonka include/asm/suspend2.h:116: error: 'per_cpu__cpu_gdt_table' undeclared (first use in this function) 1140702465 M * Bertl Wonka: this is with additional patches, right? 1140702469 M * Wonka yes 1140702475 M * Bertl k, just checking 1140702477 M * Wonka linux-2.6.16_rc4_git5-suspend2_2.2.0.1_for_2.6.16_rc3-vs2.1.1_rc9 1140702499 M * Wonka hmm 1140702504 M * Roey whendja suppose 2.6.16 will release 1140702504 M * Roey ? 1140702515 J * Smutje_ ~Smutje@xdsl-87-78-85-35.netcologne.de 1140702528 M * Wonka lemme check patch-2.6.16-rc3-rc4... 1140702624 Q * Smutje Ping timeout: 480 seconds 1140702624 N * Smutje_ Smutje 1140703310 Q * cehteh Remote host closed the connection 1140703714 M * Wonka *sigh* 1140703761 M * Bertl doesn't work as expected? 1140703917 J * cehteh foobar@cehteh.homeunix.org 1140703946 M * Wonka nope 1140703955 M * Wonka how did suspend2 compile with that? 1140703963 M * Wonka suspend2_2.2.0.1_for_2.6.16_rc3, i mean 1140703964 M * Mr_Smoke brb 1140703982 M * Bertl Wonka: don't know, never tried 1140703986 M * Mr_Smoke bbl 1140703989 Q * Mr_Smoke Quit: - Ecky ecky ecky ecky pikaaang zoop boing goodem zoo owli zhiv'. - Ni ! - Shhh... 1140704045 M * Wonka in 2.6.16-rc3 has »DEFINE_PER_CPU(struct desc_struct, cpu_gdt_table[GDT_ENTRIES]);« and »EXPORT_PER_CPU_SYMBOL(cpu_gdt_table);« removed... 1140704050 M * Wonka -in 1140704059 M * Wonka so it cannot have compiled. 1140704082 M * Wonka now i am wondering whether to join #suspend2 and complain 1140704110 M * Wonka or just stuff it and retry with the next version 1140704191 M * Bertl did you test the susp2 patches on vanilla first? 1140704851 M * harry config:/usr/local/config/kernel/linux-2.6.14.7# find . -name "*.rej"|wc -l 1140704852 M * harry 21 1140704866 M * harry only the fs's to go :) 1140704867 M * Bertl hmm .. not too bad :) 1140704895 M * harry lets hope not :s 1140704912 M * Bertl yeah, well, if you did it properly ... 1140704926 J * prae ~prae@ezoffice.mandriva.com 1140704955 M * harry Bertl: i THINK i did... 1140704960 M * harry last time it worked pretty well ;) 1140705366 M * Wonka Bertl: was a suspend2 problem, got it solved with help from freenode/#suspend2 1140705560 M * Bertl good 1140705590 M * arnaud so... i have a vserver with two ip : primary = lo, secondary = eth0. with 2.6.14 (http://linux-vserver.derjohn.de/), i get "invalid argument" every time i want to connect to an external server. i exchange the interfaces and everything works fine. with 2.6.15 it seems to work fine with lo as primary interface and eth0 as secondary interface BUT it connects to external servers with the host ip 1140705617 M * arnaud (french school => bad english.) 1140705623 M * Bertl np 1140705682 M * Bertl arnaud: looks like something changed in the network stack then 1140705725 M * Bertl arnaud: will try to investigate it when I find some time 1140705758 M * arnaud ok! 1140706165 Q * philipp Ping timeout: 480 seconds 1140706217 M * arnaud oh, the 2.6.14 is patched with 2.1.0 1140706224 M * arnaud i think i'm tired erm 1140706247 M * Bertl ah, interesting ... could you try with 2.6.16 and 2.1.x maybe? 1140706251 M * arnaud (my 2.6.15 is patched with stable vserver patch) 1140706280 M * Bertl if that fixes the issue for you, then we have to move a few changes into stable too ... 1140706447 M * Bertl have to get something to eat, back shortly ... 1140706461 N * Bertl bertl_oO 1140706509 J * matta ~matta@c-68-32-239-173.hsd1.pa.comcast.net 1140706795 M * harry 15 to go... :s 1140706814 M * harry quota is a bitch to work on :s 1140706987 J * gerrit ~gerrit@129.33.1.37 1140707296 Q * arnaud Quit: Leaving 1140707314 J * arnaud ~arnaud@d80-170-59-45.cust.tele2.fr 1140707394 M * arnaud bertl_oO, http://matvey.org.ru/pub/debian/dists/stable/main/binary-i386/kernel-image-2.6.12lvs_10.00.Custom_i386.deb works ok (source ip is set to context ip etc) 1140707805 J * cthompson ~ct@66.98.136.16 1140707877 N * bertl_oO Bertl 1140707882 M * Bertl back now 1140707897 M * bonbons daniel_hozac: What do I need to feed sysctl with to enable the debugging printk's? 1140707920 M * Bertl arnaud: but I have no idea what that contains :) 1140707944 M * cthompson sorry if this is a FAQ I was too blind to find, but how do I set a vserver up to use dhclient? 1140707956 M * daniel_hozac cthompson: you don't :) 1140707976 M * cthompson that's not EXACTLY the answer I was looking for :) 1140707982 M * daniel_hozac if you for some reason need DHCP, use NAT. 1140708007 M * cthompson so, dhcp an ipalias on the host and NAT it to a private net for the vservers? 1140708019 M * daniel_hozac right. 1140708110 M * arnaud Bertl, well i don't know too ;) 1140708127 M * arnaud Bertl, how can i know if a kernel is patched with stable or dev vserver patch ? 1140708144 M * Bertl arnaud: hard to tell ... 1140708161 M * cthompson what's the technical block on dhcp? I found a few references to setting up old style with IP_ROOT=dhcp 1140708168 M * Bertl arnaud: most features move from devel to stable after some time 1140708186 M * Bertl arnaud: well, mainline use a significant extension to the name 1140708209 M * Bertl arnaud: but I guess you do not have that :) 1140708279 M * daniel_hozac cthompson: that's with jacques' utils. 1140708281 M * arnaud i don't 1140708322 M * daniel_hozac cthompson: that still sets it up from the host, except you let the utils handle it. 1140708330 M * cthompson ah, OK 1140708335 M * daniel_hozac cthompson: the guests themselves still cannot do DHCP. 1140708336 M * arnaud Bertl, do you think it would be useful to compile 2.6.14-2.01 and 2.6.15-2.0.2-rc9 and compare their behaviour? 1140708341 M * cthompson makes sense 1140708347 M * cthompson this is all new to me 1140708357 M * cthompson and, might I add, slick as hell 1140708370 M * bonbons Bertl: how to I unmask debug messages for vserver? 1140708386 M * Bertl arnaud: I'd suggest to compare 2.6.16-rc4-vs2.0.2 with 2.6.16-rc4-vs2.1.1 if you suspect stable vs. devel 1140708407 M * Bertl arnaud: but all input on mainline patches can help 1140708455 M * arnaud Bertl, well i'm using 2.6.15-2.0.1.2 (that's the kernel with the weird source ip problem) 1140708465 M * arnaud hmm 1140708475 M * arnaud i don't really want to test devel :) 1140708499 M * Bertl hmm, I thought your kernel was devel? 1140708505 M * Bertl (the working one) 1140708591 M * arnaud i have tested on another computer two other kernels i found precompiled : a 2.6.14-devel (works) and a 2.6.12-unknown-maybe-stable (works) 1140708617 M * daniel_hozac wow, you're brave. 1140708624 M * daniel_hozac testing unknown precompiled kernels... 1140708712 M * arnaud daniel_hozac, they were tested on a tests-only computer 1140708725 M * arnaud not on a production server 1140708729 M * Bertl arnaud: maybe your setup is different, so it works there? 1140708740 M * arnaud maybe.. 1140708751 M * Bertl so you should try your kernel there too? 1140708772 M * daniel_hozac bonbons: echo x > /proc/sys/vserver/debug_nid ? 1140708823 M * arnaud Bertl, thats what i wanted to try: maybe-buggy2.6.15-2.0.1.2 vs. 2.6.14-2.01 1140708824 M * bonbons daniel_hozac: tanks, should be that 1140708847 M * daniel_hozac arnaud: why such an old 2.6.15 version? 1140708866 M * daniel_hozac 2.6.15.4-vs2.0.2-rc9 is certainly better. 1140708942 M * arnaud what is the "really stable" version? 1140708946 M * arnaud 2.6.14-2.01 ? 1140708989 M * daniel_hozac 2.6.14.3, IIRC. 1140709013 M * arnaud ok 1140709841 Q * romke Quit: leaving 1140710248 Q * gerrit Ping timeout: 480 seconds 1140710364 J * gerrit ~gerrit@129.33.1.37 1140710372 M * Bertl wb gerrit! 1140710656 M * waldi Starting vservers of type 'default'.../usr/lib/util-vserver/start-vservers: line 199: make: command not found 1140710659 M * waldi yeah 1140710672 M * Bertl apt-get install make ? 1140710675 M * Bertl :) 1140710690 M * waldi no, bugreport 1140710712 M * Bertl to whom? 1140710722 M * waldi the debian maintainer 1140710726 M * Bertl of? 1140710752 M * Bertl I don't think 0.30.204 has parallel start at all ... 1140710913 M * waldi bah 1140710941 M * waldi my last snapshot from today works, the kernel i just built does not 1140710947 M * bonbons Update to my ipv6 patches: http://homepage.internet.lu/brunop/vserver/net_ipv6.diff 1140710974 Q * gerrit Ping timeout: 481 seconds 1140710981 M * bonbons http://homepage.internet.lu/brunop/vserver/vserver-utils-1.0.3.1.diff completes add/remove support for vserver-utils (svn revision 82) and fixes some other issues 1140711027 M * Bertl bonbons: want comments? 1140711038 M * bonbons Note, the kernel patch still only manipulates the list of ip associated, no support in IP stack yet 1140711052 M * bonbons Bertl: comments are welcome :) 1140711286 M * Bertl - the kernel uses in6_addr for ipv6 ips 1140711305 M * Bertl (using __u32 is not a proper way) 1140711344 M * waldi Bertl: i found the problem 1140711347 M * waldi fuck 1140711354 M * Bertl bonbons: - your patch contains the inet fix (which is not part) 1140711373 M * Bertl +int vc_net_find_v4(struct nx_info *nxi, uint32_t addr, uint32_t mask); 1140711374 M * Bertl +int vc_net_find_v4(struct nx_info *nxi, uint32_t addr, uint32_t mask) 1140711374 M * waldi i installed a ppc32 image 1140711384 M * Bertl waldi: congrats! 1140711410 M * waldi this can't work 1140711421 M * Bertl bonbons: why do you 'declare' each function before defining it? 1140711475 M * bonbons that's because compiler sometimes generates warnings 1140711503 M * Bertl bonbons: no, if there are warnings, then they are rightfully, otherwise this is just wrong 1140711532 M * bonbons for the inet fix, it's because I applied it only on one side 1140711562 M * Bertl atoquad() is already there it's called NIPQUAD 1140711583 M * Bertl - you do not #define/#undef stuff in functions :) 1140711617 M * Bertl + nxi->ipv6[pos++] = vc_data.ip[p2++]; 1140711618 M * Bertl + nxi->ipv6[pos++] = vc_data.ip[p2++]; 1140711630 M * Bertl very ugly, make it a memcpy or assignment 1140711647 M * waldi uh 1140711662 M * waldi two index variables? 1140711673 M * bonbons with conversion to in6_addr that should go away 1140711693 M * Bertl waldi: well, it copies 16 bytes, in 4 byte steps :) 1140711744 M * waldi Bertl: i hope it does not play cache pingpong 1140711747 M * Bertl bonbons: - debug format should match the exitsting formats 1140711773 M * Bertl waldi: even if it does, it doesn't really matter but it's damn ugly :) 1140711814 M * waldi yep 1140711884 M * harry config:/usr/local/config/kernel/linux-2.6.14.7# find . -name "*.rej"|wc -l 1140711885 M * harry 0 1140711887 M * harry PING!!!!!!!!!!!!!!! 1140711900 M * Bertl congrats! 1140711905 M * bonbons Bertl: so for the debug, include method name and remove nice text? 1140711911 M * Bertl harry: now let's see if it compiles 1140711928 M * Bertl bonbons: something like that 1140711968 M * harry Bertl: first... copy the tree 1140711969 M * harry :) 1140711977 M * harry then, extend fs 1140711980 M * harry then... 1140711982 M * harry ;) 1140711990 M * Bertl bonbons: you do not use the function name anyway :) 1140712039 M * Bertl bonbons: also, try to make your patches with -NurpP (this will show the context name) 1140712045 M * harry /dev/rootvg/usrlv 5242716 4810516 432200 92% /usr 1140712050 M * harry /dev/rootvg/usrlv 10485436 4810516 5674920 46% /usr 1140712054 M * harry lvm rules ;) 1140712070 M * waldi now i wait for ext3 online resize 1140712093 M * Bertl hmm, ext2 does online resizing, IIRC? 1140712118 M * waldi in ext3 it is currently disabled 1140712131 M * Bertl probably for good reasons :) 1140712149 M * bonbons ok, next patch will be with -NurpP, with fixes for comments 1140712181 M * waldi hmm, my diff don't know -P 1140712194 M * harry Bertl: /me reiserfs'es 1140712208 M * waldi oh, oh, rasierfs 1140712216 M * waldi *hide* 1140712216 M * harry -rw-r--r-- 1 root staff 7746436 Feb 23 17:29 patch-2.6.14.7-vs2.1.1-rc8-grsec-2.1.9.diff 1140712222 M * Bertl harry: then quick, make a copy before it's gone 1140712224 M * harry hmm... this is quite large... :S 1140712237 M * waldi this is too large 1140712238 M * Bertl looks wrong to me 1140712243 M * harry uhu 1140712252 M * harry config:/usr/local/config/kernel# diff -NurpP --minimal linux-2.6.14.7 linux-2.6.14.7-g-v > patch-2.6.14.7-vs2.1.1-rc8-grsec-2.1.9.diff 1140712276 M * Bertl use lsdiff or diffstat on that 1140712290 M * Bertl with a little luck you have just the '*.orig' files there 1140712291 M * waldi and don't work as root 1140712298 M * harry aaaaaah... true :) 1140712339 M * harry config:/usr/local/config/kernel# find linux-2.6.14.7-g-v/ | grep ".orig" | xargs rm 1140712351 M * waldi diff -X .orig 1140712361 M * waldi s/X/x/ 1140712370 M * harry -rw-r--r-- 1 root staff 1546898 Feb 23 17:32 patch-2.6.14.7-vs2.1.1-rc8-grsec-2.1.9.diff 1140712374 M * harry totally good now! 1140712449 J * romke ~romke@procyon.romke.net 1140712455 M * Bertl bonbons: the vnet fix is also not part of your changes :) 1140712510 M * bonbons Bertl: that's already fixed for upcoming diff 1140712548 M * harry running make -j16 1140712562 M * Bertl harry: on your 32way? 1140712575 M * waldi Bertl: testfs fails to test for the mount point 1140712580 M * harry hmm... stopped it 1140712597 M * harry Bertl: i thought it was the quad dualcore... but it's only a dualcore xeon HT 1140712600 M * Bertl waldi: you have to provide that too, default is /test 1140712614 M * harry include/linux/vs_memory.h:14:16: warning: "CONFIG_SPLIT_PTLOCK_CPUS" is not defined 1140712623 M * waldi Bertl: it reports that as a normal test failure which is annoying 1140712636 M * Bertl waldi: patches are welcome! 1140712705 M * bonbons bonbons: - your patch contains the inet fix (which is not part) <-- which one is this one? maybe I mix something 1140712817 Q * prae Read error: Connection reset by peer 1140712819 M * waldi fuck 1140712826 M * waldi my host machine crashed 1140712845 M * Bertl hmm? 1140712860 M * Bertl bonbons: http://vserver.13thfloor.at/Experimental/delta-inet-fix02.diff 1140712905 Q * shedi Quit: Leaving 1140712995 J * prae ~prae@ezoffice.mandriva.com 1140713007 J * stefani ~stefani@superquan.apl.washington.edu 1140713031 M * Bertl welcome stefani! 1140713168 M * bonbons Bertl: Oh, how did that one come in, I saw that diff via split rcX but don't remember having changed anything while looking there?? 1140713216 M * Bertl bonbons: same for that one: http://vserver.13thfloor.at/Devel/PAT-2.1.1/delta-vnet-fix01.diff 1140713239 M * stefani hola 1140713284 M * bonbons The vnet fix I did apply it on my test-box, but forgot to apply it to the reference tree before doing the diff 1140713460 M * ebiederm Ok my /proc changes have been sent off to Andrew.... 1140713481 M * Bertl ebiederm: only the proc stuff, right? 1140713497 M * ebiederm Bertl: Yes only the proc stuff. 1140713509 M * Bertl ebiederm: did it change compared to the version mugwump uploaded yesterday? 1140713542 M * ebiederm I cleaned up the patches a little more. 1140713568 M * Bertl okay, could you provide a monster patch with those changes included for me? 1140713568 M * ebiederm The diff between the two trees would probably be less than a hundred lines... 1140713582 J * shuri ~boafroid@64.235.209.226 1140713588 M * ebiederm In a bit. 1140713604 M * ebiederm I need to run off and take care of a few non-kernel related things. 1140713621 M * ebiederm Just figured I would give everyone the heads up. 1140713652 M * Bertl great! tx! 1140713676 M * ebiederm Later. 1140713682 N * ebiederm ebiederm_oO 1140713879 M * harry hmm... there is some work here... /me off to fitness first... 1140713887 M * Bertl k, cya 1140713904 M * waldi Bertl: http://137.250.31.225/linux/testfs.diff 1140713963 M * Bertl ahem, no, please keep the syntax, it is used in other scripts too 1140714079 M * Bertl (the change to "" as default is fine with me) 1140714167 M * waldi better? 1140714237 M * Bertl I take it except for hunk 3, but that is no problem I guess 1140714253 M * Bertl thanks a lot! 1140714843 M * Bertl waldi: okay, script is uploaded, you know that you should test with -tx, right? 1140714860 M * Bertl (well, should is relative :) 1140714919 M * Roey harry 1140714923 M * waldi Bertl: hmm? 1140714923 M * Roey harry: fitness first? 1140714924 M * Roey the gym?? 1140714937 M * Roey harry: where are you in the East Coast of the USA? 1140714940 M * waldi ah 1140714977 M * waldi Bertl: ext2/ext3/xfs/reiserfs succeed, jfs failes 1140714987 M * Bertl 2.0 or 2.1? 1140714987 M * waldi 2.0.2-rc9 1140714992 M * Bertl 2.0 is expected 1140714995 M * waldi okay 1140715000 M * Bertl I did add the jfs stuff recently 1140715016 M * Bertl (they didn't have an xattr implementation at all :) 1140715046 M * Bertl was already submitted (and accepted?) upstream, so I did put it into devel but not stable 1140715933 M * waldi now lets test s390 1140715949 M * Bertl cool! 1140716033 J * rs ~rs@office.dailymotion.com 1140716039 M * rs hi 1140716043 M * Bertl welcome rs! 1140716059 M * rs how are you ? 1140716103 M * Bertl fine, tx! and you? 1140716114 M * waldi Bertl: xen-vserver will be one of the next 1140716142 M * Bertl ah, but that will require arch aptches 1140716157 M * waldi yes 1140716183 M * Bertl k, let me know when you have a xen patch 1140716211 M * waldi i just try to find out which of different kernels i want to use 1140716248 J * philipp ~philipp@84-73-110-118.dclient.hispeed.ch 1140716266 M * waldi Bertl: did you see the smp alternatives patch in the akpm tree? 1140716281 M * Bertl nope 1140716286 M * waldi http://kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.16-rc4/2.6.16-rc4-mm1/broken-out/x86-smp-alternatives.patch 1140716289 M * Bertl wb philipp! 1140716300 M * waldi it patches lock away on i386 1140716351 Q * philipp Quit: 1140716396 J * id23 ~id@p54A04162.dip0.t-ipconnect.de 1140716423 M * Bertl welcome id23! 1140716427 M * id23 Hi Bertl 1140716433 M * id23 hello #vserver 1140716437 M * Bertl waldi: looks good :) 1140716482 M * waldi Bertl: we, the debian kernel team, just thinks about adopting the patch and than drop our up kernels 1140716626 M * Bertl hmm, yeah, might make sense for a distro 1140717116 M * waldi do you know if x86_64 also have such heavy lock overhead? 1140717173 M * Bertl I'd assume so, as it is x86 compatible ... 1140717192 M * Bertl (but maybe I'm wrong there) 1140717224 M * waldi seems so :/ 1140717533 Q * fwl Quit: This computer has gone to sleep 1140717781 Q * id23 Quit: Leaving 1140717825 M * waldi arch/s390/kernel/ptrace.c: In function ‘sys_ptrace’: 1140717825 M * waldi arch/s390/kernel/ptrace.c:733: warning: label ‘out_task’ defined but not used 1140717828 M * waldi arch/s390/kernel/ptrace.c:729: error: label ‘out_tsk’ used but not defined 1140717887 M * waldi out_tsk != out_task 1140718125 J * id23 ~id@p54A04162.dip0.t-ipconnect.de 1140718195 M * Bertl waldi: checking 1140718252 M * Bertl which patch is this now? 1140718283 M * waldi 2.0.2-rc9 1140718298 M * waldi mm/mmap.c:1847: warning: ‘rb_parent’ may be used uninitialized in this function 1140718301 M * waldi mm/mmap.c:1847: warning: ‘rb_link’ may be used uninitialized in this function 1140718304 M * waldi mm/mmap.c:1845: warning: ‘prev’ may be used uninitialized in this function 1140718306 M * waldi bah 1140718314 M * Bertl unrelated 1140718318 M * waldi i know 1140718355 A * phreak`` bows in front of that p650 :) 1140718544 M * phreak`` Bertl: could you please post the links to the testfs.sh/testme.sh ? (vpn sucks especially the cisco one) 1140718579 M * Bertl http://vserver.13thfloor.at/Stuff/SCRIPT/ 1140718591 M * waldi phreak``: use vpnc? 1140718609 M * phreak`` waldi: i'm using that already :) 1140718616 Q * prae Quit: Execute Order 69 ! 1140718643 M * waldi Bertl: if i fix the ptrace.c, it seems to build 1140718662 M * waldi Bertl: do you know why this one syscall entrance point is arch-specific? 1140718671 M * Bertl waldi: okay, but I don't see the ptrace issue here 1140718678 M * Bertl waldi: is that 2.6.15.4? 1140718685 M * waldi no, 2.6.16-rc4 1140718725 M * Bertl strange 1140718752 M * Bertl well, my devel system was just rebooted, so I will have to wait until it comes back 1140718785 M * waldi and my test kernel build is mostly finished 1140718900 M * bonbons Bertl: uploaded new cleaned-up patch, hope there are no more glitches left... 1140718928 M * Bertl bonbons: same url? 1140718932 M * bonbons yep 1140718955 M * Bertl ah, debug looks much better :) 1140719028 J * Viper0482 ~Viper0482@p54975B3D.dip.t-dialin.net 1140719045 M * Bertl bonbons: is there really a need for the arpa/inet.h case? 1140719093 P * Viper0482 1140719129 M * bonbons in network_cmd.h? I guess yes as it's copied/used by userspace (for in6_addr) 1140719130 M * Bertl bonbons: my headers here define the linux/in6.h quite well 1140719139 M * Bertl (for userspace) 1140719173 M * Bertl important detail I missed: 1140719186 M * Bertl - you cannot simply change the vcmd_net_addr_v0 1140719212 M * bonbons not even staying binary compatible? 1140719225 M * Bertl (you have to create a new one, and make a separate syscall command with a new revision, otherwise you will lose compatibility) 1140719252 M * Bertl bonbons: I doubt it _is_ guaranteed to be binary compatible 1140719260 J * Viper0482 ~Viper0482@p54975B3D.dip.t-dialin.net 1140719288 M * Bertl (but maybe it is, and I just missed that, nevertheless a new command would be required, as the old one does not support the new types) 1140719300 M * bonbons for me it worked fine with old version in userspace (no guaranty of course) 1140719329 M * bonbons in that case it might be useful to immediately think about more dynamic size <- defined by type & count 1140719426 M * Bertl have to look into that first ... 1140719487 M * bonbons I mean now for userspace <-> kernel communication. Kernel internal is a different task 1140719535 M * Bertl no, we do not introduce dynamic structures 1140719561 M * Bertl this is something I will avoid at all costs ... 1140719613 M * bonbons fine for me too. I'm away for 2h30-3h... 1140719625 M * Bertl k, last question 1140719641 M * Bertl do we really need the if defined() around the includes? I don't think so 1140719697 M * Bertl anyway, cya later 1140719715 M * bonbons don't know, guess bet one check to much than one too few 1140719925 J * ntrs ~ntrs@vs835.rosehosting.com 1140720028 Q * ntrs Quit: 1140720131 Q * id23 Quit: Leaving 1140720602 M * waldi Linux waldi02 2.6.16-rc4-vserver-s390x #2 SMP Thu Feb 23 19:24:43 CET 2006 s390x GNU/Linux 1140720745 J * shedi ~siggi@inferno.lhi.is 1140720784 M * waldi Bertl: testme succeds 1140720984 M * waldi EXT2-fs: loop0: tagxid not permitted on remount. 1140721001 Q * phreak`` Quit: leaving 1140721055 M * waldi Bertl: okay, testfs succeds 1140721066 M * Bertl waldi: you should get a bunch of those messages 1140721077 M * Bertl (they are actually part of the test run) 1140721086 M * waldi yes 1140721444 M * daniel_hozac Bertl: dynamic size for the user<->kernel space interaction? 1140721460 M * Bertl daniel_hozac: is a definitive no :) 1140721472 M * daniel_hozac right, i agree. was just checking. 1140721486 J * ntrs ~ntrs@vs835.rosehosting.com 1140721630 Q * ntrs Quit: 1140721951 J * bitriot ~spk@s142-179-181-190.ab.hsia.telus.net 1140721966 M * Bertl welcome bitriot! 1140721980 M * bitriot hi 1140723435 J * phreak`` ~phreak``@styx.xnull.de 1140723454 M * Bertl wb phreak``! 1140723590 M * phreak`` thanks Bertl :) 1140723707 J * mkhl mkhl@200-153-181-224.dsl.telesp.net.br 1140723839 Q * vrwttnmtu Quit: Leaving 1140723987 Q * phreak`` Quit: leaving 1140724023 J * phreak`` ~phreak``@styx.xnull.de 1140724853 P * stefani I'm Parting (the water) 1140725180 Q * mkhl Quit: 1140725254 M * mugwump wow, I'm really impressed by the endurance of Jörg Schilling 1140725344 J * mkhl mkhl@200-148-41-243.dsl.telesp.net.br 1140725433 M * Bertl mugwump: yeah, I had to tease him ... I'm already feeling bad about it :/ 1140725491 M * mugwump I took a shot at him too, ended up with a few private e-mails about it. he's just disappointed that people can't see past his arrogant approach and see his technical genius. 1140725520 J * trouti ~G_w@196.203.220.58 1140725615 J * comfrey ~comfrey@h-64-105-87-234.sttnwaho.covad.net 1140725628 M * mugwump I tried to tell him that it didn't matter how right you were,... 1140725652 M * trouti hi 1140725707 M * mugwump hello trouti, where are you from? 1140725730 M * trouti Tunisia 1140725732 M * trouti and u 1140725737 N * trouti george_wassouf 1140725747 M * mugwump Britain originally, but NZ now. 1140725755 M * george_wassouf nice 1140726197 Q * george_wassouf Quit: 1140726205 J * fwl ~fwl@83.215.237.1 1140726447 M * Bertl mugwump: yeah, it's a shame, we should put up a shrine ... 1140726531 M * Bertl mugwump: btw, you can get a 'proc free' version in a few minutes 1140726560 Q * Snow-Man Ping timeout: 480 seconds 1140726562 M * Bertl (i.e. devel - proc checks and pid virtualization) 1140726807 J * Snow-Man ~sfrost@kenobi.snowman.net 1140727029 M * bitriot Hmm, what's the difference between 2.0.1 and 2.0.1-r1? 2.0.1-r1 seems to be the closest thing to a stable version that I can use with 2.6.15 kernel on gentoo 1140727160 Q * sladen Ping timeout: 480 seconds 1140727188 J * gerrit ~gerrit@129.33.1.37 1140727322 M * Snow-Man bitriot: uhh 1140727344 M * Snow-Man bitriot: Try http://vserver.13thfloor.at/Experimental/patch-2.6.15.4-vs2.0.2-rc9.diff ? 1140727380 M * bitriot I was really hoping to avoid patching manually 1140727457 M * daniel_hozac 2.0.2-rc6 is masked. 1140727500 M * bitriot Yeah, everything is masked except 2.0.1 for 2.6.14 1140727508 M * daniel_hozac (no significant changes between -rc6 and -rc9) 1140727532 M * bitriot I don't know what the best plan is? 2.0.2-rc6 you think? 1140727538 M * daniel_hozac yes. 1140727543 M * bitriot ok 1140727587 J * sladen paul@starsky.19inch.net 1140727781 Q * gerrit Ping timeout: 480 seconds 1140727818 Q * lilalinux Remote host closed the connection 1140728019 M * bonbons I'm back 1140728155 M * Bertl wb bonbons! 1140728225 Q * fwl Quit: This computer has gone to sleep 1140728249 M * bonbons So, how about guaranty of sizeof(struct in6_addr) == sizeof(uint32_t[4])? 1140728288 M * Bertl not convinced yet, but in any case we should bump the version 1140728460 M * Bertl bonbons: the vc_net_find_* seems a little dangerous to me 1140728468 M * Bertl (i.e. it probably needs more thought) 1140728509 J * fwl ~fwl@83.215.237.1 1140728520 M * bonbons one more bound check? or thought in what respect? 1140728520 J * gerrit ~gerrit@129.33.1.37 1140728522 M * Bertl but it is fine for testing, and as I said, we have to remove the static entries anyway 1140728539 M * Bertl bonbons: what if you add the same IP with two different masks? 1140728557 M * Bertl bonbons: do we really want that? 1140728567 M * bonbons it's seen as two different ones by now (as by iproute2 tools) 1140728599 M * Bertl maybe we should also extend the check mechanism to subnets 1140728631 M * bonbons I know that case is somehow doubious, and depends very much how the mask information is used afterwards... 1140728727 M * bonbons check on subnets? what kind of check? not allow different addresses on colliding subnets? like 192.168.0.10/28 and 192.168.0.250/24? 1140728777 M * Bertl let's forget about that for now .. we can do a lot of cleanup later 1140728816 M * Bertl IMHO the important part is to get the actual networking running 1140728821 Q * Viper0482 Quit: bin raus, 1140728826 M * Bertl and I guess for that purpose it's fine as it is 1140728862 M * Bertl I will complain and bitch about the details later 1140728871 M * bonbons ok, so I will start getting IPv6 working inside the guests 1140728933 M * Bertl look for the ipv6 checks (in the patches) they should give you some hints where you need to add stuff 1140728937 M * bonbons better start with something ugly to improve and that works than with something good-looking that does not work :) 1140729005 M * bonbons I've seen some of them, but understanding the machinery around is I guess important to get it working without tapping in the dark :) 1140729091 M * daniel_hozac did you change to a dynamic in-kernel structure yet? 1140729124 M * bonbons the aim is not to just have IPv6, but more to learn while doing something useful -- practive what one doesn't learn at school/university 1140729143 M * bonbons daniel_hozac: no, nothing changed to dynamic yet 1140729445 M * phreak`` daniel_hozac, bitriot: >=vserver-sources-2.1.1-rc7 ain't masked :) just ~x86/~amd64 and not stable (x86/amd64) 1140729523 M * daniel_hozac phreak``: sorry, i thought ~x86 == masked. can you tell i've never used Gentoo? :) 1140729537 M * phreak`` daniel_hozac: heh ;) 1140729553 M * phreak`` daniel_hozac: ~x86 == testing :) 1140729565 M * daniel_hozac ok, thanks. 1140729573 M * phreak`` you're welcome :) 1140729853 M * bitriot I thought ~arch meant 'masked' too 1140729932 M * bitriot (masked by: ~x86 keyword) 1140729939 M * bitriot ^^^ that's probably why 1140730083 M * daniel_hozac phreak``: btw, any reason there's no 2.0.2-rc9? 1140730130 J * LiamH ~none@healy.washington.dc.us 1140730246 M * LiamH How do I halt my vserver? I thought 'halt' might do the trick, but I get an error. 1140730251 M * daniel_hozac halt -f 1140730278 M * LiamH Works, thanks. 1140730287 M * daniel_hozac halt by itself will only work if you're running an init inside the guest. 1140730300 M * LiamH And apparently I wasn't 1140730378 M * LiamH and how do I restart it? 1140730403 M * daniel_hozac from inside while it's running? reboot -f 1140730422 M * LiamH no, after it's been halted. 1140730432 M * daniel_hozac vserver ... start 1140730434 M * LiamH vserver start 1140730470 M * LiamH Thanks. 1140730501 M * bitriot Is vserver stop the same as halting it? 1140730507 M * daniel_hozac yes. 1140730517 M * daniel_hozac halt -f will call vserver stop. 1140730525 M * daniel_hozac reboot -f will call vserver restart. 1140730536 M * LiamH Ah, so it can be halted wihtout entering it. 1140730560 M * bitriot using vshelper? 1140730565 M * daniel_hozac exactly. 1140730573 M * cthompson any Gentoo vserver gurus about? 1140730578 M * bitriot ah, ok, so that's what that's for 1140730581 M * cthompson Selected arch not supported, or profile does not exist! 1140730599 M * cthompson I've tried athlon-xp, i686 and x86 stage3 tarballs 1140730625 M * daniel_hozac cthompson: try searching the IRC logs. i recognize the error but can't remember the solution, if any. 1140730636 M * cthompson will do 1140730637 M * bitriot cthompson: add the architucture to the vserver-new line 1140730637 M * daniel_hozac (or just wait for Hollow or phreak`` to read it :)) 1140730646 M * bitriot vserver-new ............. x86 1140730651 M * cthompson bitriot: oooooh 1140730682 M * bitriot I think that's exactly the same problem I had 1140730685 M * cthompson that did it 1140730755 M * cthompson cool 1140730760 M * cthompson thanks 1140731081 Q * gerrit Ping timeout: 480 seconds 1140731385 M * bitriot What's the standard way to automatically start vservers on boot in gentoo? 1140731429 M * daniel_hozac i believe it's the same as for other distros, i.e. echo default > /etc/vservers//apps/init/mark 1140731490 M * bitriot ok, I'll try that 1140731738 M * bitriot Yup, that works, thanks 1140731796 J * gerrit ~gerrit@129.33.1.37 1140731964 J * ebacve icechat5@dynamic-oit-vapornet-a-1767.Princeton.EDU 1140731981 M * Bertl wb ebacve! 1140731995 M * ebacve thanks. 1140732090 Q * rs Quit: rs 1140732213 J * brisho ~brians@c-69-254-229-215.hsd1.ks.comcast.net 1140732367 M * Bertl wb brisho! 1140732498 M * brisho Bertl: howdy 1140732516 Q * fwl Quit: This computer has gone to sleep 1140732538 M * brisho I've been trying to figure out why I can't seem to start more than one vserver 1140732549 M * Bertl and? 1140732567 M * brisho I can start up vserver1 (which has apache) and it starts up fine 1140732615 M * brisho but when I go to start up vserver2 (containing postfix) it seems to go through the init.d scripts 1140732629 M * Bertl and? 1140732637 M * brisho ok, but then when I do a vserver-stat it doesn't show vserver2 at all 1140732651 M * brisho and i can't enter vserver2 either 1140732654 M * Bertl and if you do it the other way round? 1140732663 M * brisho lemme try that... 1140732714 M * brisho nope, i stopped vserver 1 and then tried to start vserver2, but got the same results. 1140732730 M * Bertl so, you are using sysv init, I assume? 1140732746 M * Bertl i.e. you see the output from the startup 1140732747 M * brisho from Debian sarge 1140732762 J * Aiken ~james@tooax6-018.dialup.optusnet.com.au 1140732764 M * Bertl something like 'starting blabla' 1140732771 M * brisho yep 1140732792 M * Bertl well, I _assume_ that whatever you try to start, fails (but debian shows success) 1140732808 M * brisho that's what seems to be happening 1140732811 M * Bertl and because that fails, nothing is left running, and so the context/guest terminates 1140732836 M * Bertl try to activate sshd for the second guest (that should keep it running) 1140732836 M * bonbons Bertl: in include/linux/vs_network.h for addr_in_nx_info I see that 0.0.0.0 is only matched if it's the first addresse in the list... shouldn't the ANY address be matched no matter what's it's position? 1140732855 J * fwl ~fwl@83.215.237.1 1140732859 M * Bertl bonbons: nope, the 0.0.0.0 match is a special case 1140732875 M * Bertl bonbons: it's basically a hack to disable checking completely 1140732923 M * brisho ok will try that 1140733056 M * bonbons Bertl: don't really get it... the 0.0.0.0 means anyhow "I'm good for every target" 1140733116 M * Bertl bonbons: when you want to have a guest which allows to bind to all IPs, then you specify only a single IP 0.0.0.0 1140733116 M * Bertl s/allows/is allowed/ 1140733134 M * Bertl it doesn't make sense to check for this special case on any other position 1140733207 M * bonbons let's say it doesn't make much sense (but adding 0.0.0.0 once there is already at least one address existing must get special treatment then) 1140733232 M * Bertl just ignore this special case for now 1140733272 M * bonbons so either here or at address addition some more care must be taken of the 0 address. Will just let it there with a note 1140733286 M * brisho Bertl: no go. I cut the init.d scripts down to start only ssh, and also set sshd_config 1140733311 M * brisho Bertl: to listen only on the vserver2 IP. Same results. 1140733413 M * Doener brisho: you also adjusted sshd_config on the host, right? 1140733427 M * Doener adjusting in the vserver is not needed 1140733456 M * Doener ... unless you want your vserver's sshd to only listen on a limited subset of the vserver's ip addresses of course 1140733457 M * brisho Doener: hmmm, no, I adjusted it in the vserver 1140733472 M * Doener ok, i guess you have a mail server, but no http server on the host? 1140733491 M * Doener that'd explain why vserver1 runs, but vserver2 not (given the information I have) 1140733507 M * brisho what I'm trying to do is put a http server in vserver1 and a mail server in vserver2 1140733510 J * rs ~rs@vol75-7-82-229-177-124.fbx.proxad.net 1140733526 M * Doener you need to limit the host's processes yourself, the limiting of the vserver's processes is done automatically 1140733564 M * Doener no, if you have a mail server on the host that listens on 0.0.0.0:25, that will block any mail server in any vserver that tries to listen on port 25 on any address 1140733575 M * Doener s/^no/now/ 1140733575 J * matt1 ~matta@c-68-32-239-173.hsd1.pa.comcast.net 1140733588 Q * fwl Quit: This computer has gone to sleep 1140733625 M * Doener so adjust sshd_config on the host, and your mail server's configuration, also on the host, so that those processes only listen on a specific address 1140733657 Q * matta Ping timeout: 480 seconds 1140733687 M * Doener for postfix, that's, for example, inet_interfaces = localhost 1140733709 M * Bertl I'm off for today .. has been a long one for me ... 1140733717 M * Doener g'night Bertl! 1140733723 M * Bertl have a good whatever ... and cya tomorrow! 1140733723 M * brisho Bertl: thanks for your help...goodnite 1140733732 N * Bertl Bertl_zZ 1140733733 M * bonbons Good night Bertl! 1140733800 M * brisho Doener: I'm still trying to wrap my brain around this stuff, so I'm starting small here... 1140733840 M * brisho I set sshd on the host to listen on the host's IP 1140733854 M * brisho then set the vserver to listen on the vserver ip 1140733861 M * brisho but still get the same results... 1140733863 M * Doener no need for the latter 1140733869 M * Doener did you restart the host's sshd? 1140733907 M * brisho yes 1140733934 M * Doener ok, what does this output chcontext --ctx 1 netstat -tnlp | grep ssh 1140734012 M * brisho tcp 0 0 172.16.219.99:22 0.0.0.0:* LISTEN 5082/sshd 1140734037 M * Doener looks good... 1140734041 M * brisho 172.16.219.99 is the host's IP 1140734059 M * brisho 219.237 is vserver1, 219.238 is vserver2 1140734094 M * brisho I just noticed that when I go to start vserver2, the first line of output is 1140734097 M * brisho RTNETLINK answers: File exists 1140734103 Q * gerrit Read error: Operation timed out 1140734115 M * brisho that doesn't come up when I start vserver1 (which starts OK) 1140734125 M * Doener that's because the ip address already exists 1140734136 M * Doener in your case probably because the vserver was never stopped, but just died 1140734149 M * Doener nothing to worry about 1140734203 M * Doener hum hum... 1140734230 M * Doener did sshd log anything in the vserver? 1140734239 M * brisho on vserver2? 1140734248 M * Doener yep 1140734293 M * brisho nope, that vserver's /var/log/syslog is completely empty 1140734376 M * Doener and /var/log/messages and daemon.log? 1140734402 M * brisho there is no daemon.log 1140734427 M * Doener ok 1140734732 M * brisho Doener: Well, I've gotta go now, so will look at this some more later. Thanks for your help. 1140734739 M * Doener np 1140734995 J * gerrit ~gerrit@129.33.1.37 1140735131 J * Doener` doener@i5387C41D.versanet.de 1140735352 Q * LiamH Quit: bye 1140735380 Q * brisho Quit: using sirc version 2.211+KSIRC/1.3.11 1140735550 Q * Doener Ping timeout: 480 seconds 1140736234 J * anonc ~anonc@staffnet.internode.com.au 1140737022 P * anonc adios 1140737894 Q * gerrit Ping timeout: 480 seconds 1140738630 J * gerrit ~gerrit@129.33.1.37