1135901114 Q * Marinus Quit: 1135901183 Q * Doener Quit: Leaving 1135901240 M * tso bye 1135901241 Q * tso Quit: BitchX: the fresh-maker! 1135902748 J * Aiken ~james@tooax8-065.dialup.optusnet.com.au 1135903353 J * _are_ ~are@dslb-084-056-143-087.pools.arcor-ip.net 1135903361 M * _are_ hi 1135903474 M * _are_ when starting a vserver i get 'Too many IP numbers, max 16', checked a bit and found this to come from chbind. Is this a real limit or just a random restriction of chbind? 1135903640 M * derjohn _are_ please send complaints to bertl :) I had the same problem and no one agreed :) 1135903652 M * _are_ hmm, ok, as soon as you ask in irc, google gives a result, there is a define to fix this 1135903660 M * derjohn _are_, take linux-vserver.derjohn.de 1135903664 M * _are_ http://www.paul.sladen.org/vserver/archives/200304/0332.html 1135903696 M * mnemoc having 16 IPs con the same vserver is stupid 1135903697 M * derjohn there are .debs for util-vserver (prepatches) and a 2.6.14 kernel. all for SID 1135903712 M * derjohn mnemoc, hehe ;) I know 1135903752 M * _are_ mnemoc: not in the way I use vservers, I use them as containers on drbd disks to do failover setups. 1135903753 M * mnemoc but the limit is hardcoded on vserver patch ... and utils, so patch them 1135903814 M * derjohn _are_ are you on debian? then switch to my .deb or patch yourself. hints are at the bottom of this file: http://linux-vserver.org/some_hints_from_john 1135903832 M * _are_ yes, debian, sarge in this case 1135903866 M * derjohn _are_, we setup rsync servers on rsync into 192er vserver. If the real one fails we start the backup ones ... 1135903892 M * derjohn _are_, so we never know if the real has more than 16 (e.g. bouncers ...) 1135903954 M * derjohn _are_, may deb are build for sid. but you can take the src and modify the build-deps of dietlibc and dpkg-buildpackage 1135903997 M * derjohn but basically you have to change the define of "IPV4ROOT" from 16 to 64 1135904003 M * _are_ well, i am not very sure about usining yours or patching myself. both are no big deal for the moment, but i would have to do it for every new kernel/util-vserver. might turn into a hassle. 1135904040 Q * Johnnie Read error: Connection reset by peer 1135904050 M * _are_ what actually is IPV4ROOT meant to be for? 1135904061 M * derjohn kernel/network.h:#define NB_IPV4ROOT 64 1135904099 M * derjohn _are_, this is some kind of define for the internel management of the guests's ip in it's context. 1135904131 M * _are_ so only in the vserver patch? 1135904135 M * derjohn _are_, the lookup is in linear time "O(n) ?" and people here assume 256 to slow ... 1135904162 M * derjohn _are_, yes I think so. But I am in no way a kernel_hax0r ... :) 1135904180 M * _are_ well, can't find it in the native no-vserver source here. 1135904201 M * derjohn _are_, and dont forget to patch the utils 1135904239 M * _are_ have now the choice between patching kernel/utils eahc time there is a new version or mount --bind-ing the data to 2 vservers and use it that way 1135904245 M * derjohn BTW: the new networking code wich is called ngnet will be able to handle more IPs. This code is expected to be finished in 6 or 12 monaths 1135904309 M * _are_ yesyes, ngnet will fix all troubles i ever had with networking and when i first encountered vservers 1 year ago it had been in discussion / alpha test already 1135904312 M * derjohn _are_, yes. but: you have to compile and patch the kernel any. the change is only 1 line. 1135904348 M * _are_ that's definitly a point 1135904395 M * derjohn _are_, bertl went to usa and worked pretty hard on the new code. it run with 127.0.0.1 already ... completely separated. so the host doesnt even knoe about the guest ip when it comes to routing. 1135904466 M * derjohn _are_, yes. And I try to sweettalk all responsibles to make IPV4ROOT some kinde of compiletime/make option (to get at least rid of the patching stress) 1135904469 M * derjohn a 1135904471 M * _are_ hehe, well, uhm, sort of an emergency here as it has to be fixed today. I am afraid I can't wait for ngnet and have to patch it. 1135904508 M * derjohn _are_, no I say things like "_are and maaaany others need that, too and it wont cause much trouble. 1135904524 M * _are_ compile time option is only useful for the kernel, the tools would have to check out the kernel and use the appropriate limit 1135904541 M * derjohn (indeed I think it would be pretty easy for a coder to make the utils ask the kernel whats it's define ...) 1135904566 M * derjohn _are_, hr, yes, zwo dumme ein gedanke. 1135904571 M * _are_ if it is an exported symbol or in a .h that is always there: yes 1135904631 M * derjohn _are_, /me is not fluent in c .... I always wanted to file a wish on savannah .. i cannot remeber if i did.... 1135904679 M * _are_ well, not fluent either, but my passive c is present to some extend 1135904693 M * derjohn _are_, ist hier 2h. I'll have to do some energy saving in oder to get a spinup early tomorrow. 1135904725 M * _are_ exactly same time here :-> 1135904739 M * derjohn _are_, but im the meantime you could trust my debs... (at least they are signed be be and some trust of DDs) 1135904757 M * derjohn _are_, I saw it in your reverse lookup ... 1135904760 M * derjohn :) 1135904803 M * _are_ will check your debs and see if i get a clue on kernel config stuff to add this option as a config to the krnel 1135904844 M * derjohn _are_, so I really would like to discuss longer, but I have to walk home now. BTW: which city are you from? 1135904928 M * _are_ stuttgart 1135904946 M * derjohn _are_, i took original debian sources oh 2.6.14-4 from experimental 1135904952 M * derjohn (not vanilla) 1135904978 M * derjohn ah, greetz to the city of engineering ;) 1135904978 M * _are_ will have a look 1135905037 M * derjohn but: there are deps on libc of sid .... to you have to recompile. in case of the utils: debian/control -> change depency of dietlibc ... every else is said to be fine. 1135905042 M * derjohn off now 1135905049 M * derjohn by(t)e .... 1135905077 M * _are_ will do 1135905214 J * sebi ~sebi@Fcdbf.f.strato-dslnet.de 1135905316 Q * sebi_ Ping timeout: 480 seconds 1135905675 M * _are_ anyone with a little more plan on C than me in here who can tell me if in #define NB_IPV4ROOT 16 I replace the 16 by another define it works or is the preprocessor not reentrant? 1135905740 M * Adrinael It works. 1135905751 M * Adrinael Taken that the thing you put there is defined above it 1135905856 M * _are_ well, it will be in the kernels .config. 1135905894 M * _are_ i have not checked the make process, but seems autoconf.h is generated form .config and does a #define on the .config variables 1135905962 M * Adrinael You will surely get a compile error containing the thingie there if it doesn't work right 1135905987 M * _are_ sounds acceptable 1135906526 J * Johnnie ~jdlewis@jdlewis.whizwire.org 1135907585 N * nokoya nokoyaz 1135907611 N * nokoyaz nokoya 1135907647 J * Aiken_ ~james@tooax6-222.dialup.optusnet.com.au 1135907880 M * _are_ compile worked 1135908002 Q * Aiken Ping timeout: 481 seconds 1135911030 J * Guest329 JavaUser@ACC68273.ipt.aol.com 1135911037 M * Guest329 Hi 1135911052 N * Guest329 A_Train_Is_Exciting 1135911070 M * A_Train_Is_Exciting Hello 1135911071 M * A_Train_Is_Exciting everyone 1135911086 P * A_Train_Is_Exciting 1135912627 J * cehteh foobar@cehteh.homeunix.org 1135912719 M * cehteh hi 1135912724 M * cehteh # vserver test2 build -m debootstrap -- -d sarge --arch i386 1135912724 M * cehteh /usr/local/lib/util-vserver/vserver-build: unrecognized option `--arch' 1135912730 M * cehteh .. is that a bug? 1135912761 M * cehteh (I want to bootstrap a i386 userland on x86_64) 1135915642 Q * lilalinux Remote host closed the connection 1135916327 Q * ryker Quit: Leaving 1135916648 J * balbir ~balbir@59.92.138.85 1135918503 Q * balbir Quit: Leaving 1135918994 Q * jayeola Quit: quit 1135919539 J * mordur ~mordur@dsl-201-4.hive.is 1135922072 J * Smutje_ ~Smutje@xdsl-87-78-61-86.netcologne.de 1135922180 Q * Smutje Ping timeout: 480 seconds 1135923845 Q * mordur Quit: Leaving 1135924730 J * stefani ~stefani@c-24-19-46-211.hsd1.wa.comcast.net 1135924735 P * stefani 1135926011 J * Aiken__ ~james@tooax8-008.dialup.optusnet.com.au 1135926030 Q * Aiken__ Quit: 1135926327 Q * Aiken_ Ping timeout: 480 seconds 1135926855 M * eyck did anyone tried running vmware as a vserver guest? 1135927652 M * eyck it seems that non-graphics things work fien (NetBSD install for example), but anything more colorfull results in black screen 1135927778 J * ichigo ~m@203.81.233.61 1135927814 M * ichigo hello everybody. can somebody tell me how i use my printer in the vserver. 1135927957 M * eyck what do you mean? 1135927972 M * eyck you've got local usb printer? network printer? what? 1135929975 M * ichigo eyck : usb hp laser jet 1010. setup on the host 1135930826 M * eyck can you see /dev/usb on guest? 1135931019 M * ichigo nope. not /dev/usb or even /proc/bus/usb 1135931134 M * _are_ vprocunhide you already used? 1135931246 J * Milf ~Miranda@ipsio271.ipsi.fraunhofer.de 1135931420 M * Milf morning folks 1135931426 M * ichigo yes i used vproccpuunhid. the vserver refuses to start without it. 1135931547 J * comfrey ~comfrey@cpe-69-203-195-46.nyc.res.rr.com 1135932383 J * tso ~tso@rev.193.226.232.31.euroweb.hu 1135932390 M * tso hi all 1135934298 Q * _are_ Quit: bbl 1135934698 M * ichigo soory had to leave for a while...i did do /etc/init.d/vprocunhide start. before starting the vserver. 1135935197 M * Milf Can anyone give me a hint on how to find which of my host servers is hosting a specific IP? 1135935228 M * Milf The IP is pingable, I want to use it for a test server. But I can't find it on any of my hosts using 'ip addr' 1135935601 N * Bertl_zZ Bertl 1135935606 M * Bertl morning folks! 1135935609 M * Milf Good morning Bertl 1135935754 M * Bertl Milf: if 'ip addr ls' doesn't show it, then it's not used/assigned ... in general ping is not _such_ a good test, as a switch might answer that one .. try tracepath instead 1135935811 M * Bertl ichigo: you have to _copy_ the required devices (from the host's /dev) into the guest 1135935843 M * Milf tracepath? What's tracepath? If you mean traceroute, that doesn't tell me anything I don't already know :) 1135935878 M * Bertl tracepath is part of iputils 1135935975 M * Bertl eyck: where would/does 'more colorful' stuff output? 1135936033 M * Milf Hmmm, found it, but that still doesn't tell me anything new, as I'm not being told which real interface the virtual one resides on. 1135936055 M * Bertl cehteh: try -d sarge ARCH=i386 1135936090 M * Bertl Milf: ahem, you use it from where? 1135936128 M * Bertl Milf: (maybe I got your question wrong) 1135936154 M * Milf I used tracepath from a server in a different network. 1135936161 M * Milf I'll try it from the same network. 1135936191 M * Milf Zip. Nothing. 1135936250 M * Milf Meaning: Yes I get some output, but nothing that will help me find the machine that the address is being used on. 1135936286 M * Bertl okay, what does 'server' and 'same network' mean to you? 1135936306 M * Bertl and could you /msg me the tracepath output? 1135936313 M * Milf 'server' means which machine, which NIC the IP address is being used on 1135936355 M * Milf 'same network' means same network segment, same netmask, same broadcast address. 1135936373 M * Bertl ah, the ip can only be reached on some host, yes? 1135936391 M * Bertl and there you reach it via lo (in your trace) 1135936404 M * Milf It can be reached from anywhere, so I suspect I've already got a guest server somewhere using it. 1135936416 M * Milf So I want to shut down that guest server to free up my test-IP 1135936458 M * Bertl okay, what does 'arp' show on the host, right after you 'pinged' the IP? 1135936472 M * Bertl (/msg me again) 1135936517 M * Milf Yeah, that's something new, it gives me a MAC address. 1135936565 M * Bertl :) now go and look with 'ip link ls' 1135936573 M * Milf That one I might find. Thanks Bertl. 1135936585 M * Bertl you're welcome! 1135936610 M * eyck Bertl: uh? what do you mean? vmplayer is an X app 1135936632 M * Bertl eyck: so it does use X forwarding in your setup? 1135936644 M * eyck actually, I've got X in guest, 1135936660 M * Bertl try with x forwarding, and see if it stays black 1135936666 M * eyck ok, thnx, 1135936683 M * Bertl if so, then it tries to access xv or sdl or whatever 1135936701 M * Bertl (and probably requires some -x11 switch or so) 1135936722 M * eyck I've got xv working OK 1135936756 M * Milf Bingo, found the culprit. 1135936888 M * Milf Hmmm, a vserver seems to have stopped uncleanly. Now the address is still in use, although the vserver is stopped. On startup, I get RTNETLINK answers: File exists 1135936925 M * Bertl usually the result of changes to the config files (e.g. prefix/netmask) while the guest is running 1135936940 M * Bertl (folks always do that :) 1135937001 M * Milf Hmmm, I'll have a look into that. I think it might be my ghastly hack-fix to the problem of vservers not stopping: inserting a reboot -f into the last init phase. 1135937030 M * Bertl inside the guest, I hope? 1135937042 M * Milf inside the guest, yes 1135937062 M * Bertl well, that should not stop the tools from working properly (at least not vanilla 0.30.209) 1135937078 M * Milf yep, that may be it. on stopping I get RTNETLINK answers: Cannot assign requested address 1135937093 M * Milf Hmmm, this server has got vanilla 0.30.208 I think 1135937096 M * Bertl that just means that the prefix/netmask is wrong 1135937103 J * prae ~prae@ezoffice.mandriva.com 1135937130 M * Bertl (or maybe the interface for the ip) 1135937142 M * Milf Yeah but the server keeps the IP occupied. 1135937166 M * Milf Ok, solved on problem, now I go create another :) 1135937174 M * Bertl and, what was it? 1135937193 M * Milf The problem of finding which of my hosts was using the IP 1135937213 M * Bertl ah, okay, please /msg me the 'ip addr ls' output of the host 1135937237 M * Milf Ahem I found the server. Thank you. 1135937257 M * Milf As to the new problem to create: I don't know what that is, I haven't created it yet :) 1135937304 M * Bertl I was talking about the RTNETLINK issue 1135937946 M * Bertl Milf: okay, so it was an incorrectly assigned prefix :) 1135937978 M * Milf we can only hope :) 1135937994 M * Milf As it was a testing server, I will not overrate the problem. 1135938038 M * Bertl if you get 'Cannot assign requested address' this means that the address is already taken (or, on shutdown, that the address could not be found) 1135938085 M * Bertl as the entire address consists of the ip _and_ the netmask, different prefixes/netmasks give different addresses, if if the ip matches 1135938116 M * Bertl (which is often confusing for the administrator) 1135938279 M * Milf I solved the problem by deleting the test server :) 1135938346 Q * jkl Ping timeout: 480 seconds 1135939161 M * yang I am trying to setup many IP addresses for vserver, but this command doesnt work - for e in `seq 1 49`; do mkdir $i && pushd $i && echo eth0 >dev && echo 86.110.64.$(($i+150)) >ip && popd;done 1135939183 M * yang Try `mkdir --help' for more information. 1135939183 M * yang mkdir: too few arguments 1135939199 M * Bertl $i vs 'e' ? 1135939264 M * Bertl btw, you also want to write a prefix or netmask there 1135939287 M * yang could you tell me the right line? 1135939303 M * Bertl change the 'e' in the 'for' to 'i' 1135939323 M * Bertl and add something like echo 24 >prefix 1135939325 M * yang ok 1135939333 M * yang to i or to $i 1135939355 M * Bertl ahem, you got that line from somewhere, I suppose? 1135939365 M * yang yes, someone on this channel wrote me this line 1135939380 M * Bertl ah, i.c. well, you should look into bash a little ... 1135939384 M * yang its a lot of work to setup 50 ips by hand 1135939392 M * Bertl yes, it would be i not $i 1135939423 M * Bertl btw, I'd do something like: 1135939448 M * yang it created well now...but as you said, its missing the prefix line 1135939491 M * Bertl for d in `seq 1 49`; do mkdir $d && echo eth0 >$d/ip && .... ; done 1135939510 M * Bertl IMHO the pushd/popd is fun but just confusing if something goes wrong 1135939564 M * yang so i can make it like this for d in `seq 1 49`; do mkdir $d && echo eth0 >$d/ip && .... && echo 24 >prefix ; done 1135939615 M * Bertl echo 24 >$d/prefix but yes 1135939641 M * Bertl the pushd basically puts the current directory on a stack and changes the directory 1135939665 M * Bertl the problem is, that commands connected with && will only be executed, if the previous command succeeded 1135939703 M * Bertl which in turn means, that if any of those commands in the && sequence fails, the last popd (which changes back to the push-ed dir) will not get executed 1135939768 M * yang well assuming that i have allready created previously the files...with for e...now trying this for d in `seq 1 49`; do mkdir $d && echo eth0 >$d/ip && .... && echo 24 >$d/prefix ; done it doesn't add the prefix file 1135939835 M * Bertl because the mkdir already fails ... 1135939846 M * Bertl first remove the old dirs 1135939862 M * aba oder make mkdir -p if you don't care for the old dirs contents 1135939876 M * yang ok 1135939932 M * yang bash: ....: command not found 1135939959 M * Bertl *sigh* :) the .... was a placeholder for your 'other' commands :) 1135939968 M * Bertl yang: try: 1135940020 M * Bertl for d in `seq 1 49`; do mkdir $d && echo eth0 >$d/ip && echo 86.110.64.$d >$d/ip && echo 24 >$d/prefix ; done 1135940034 M * Bertl make the 1 and 49 match your start and stop digit 1135940048 M * Bertl i.e. if your IPs start at 86.110.64.150 1135940070 M * Bertl use `seq 150 200` instead 1135940119 M * Bertl still I don't see why somebody would need 50 ips inside a guest 1135940209 M * yang it's used for IRC vhosts 1135940211 M * Bertl aside from the fact that there is a good reason why the current limit is 16 ips per guest 1135940222 M * yang why a good reason? 1135940244 M * Bertl because it slows down performance if you set it higher 1135940286 M * yang I have promised to my friend that he will be able to use 50 ips 1135940298 M * Milf what for? 1135940312 M * yang he wants to do irc vhosts 1135940383 M * Milf Waht are irc vhosts? 1135940397 M * Bertl why not have 4 guests with 15 ips each? 1135940405 M * yang somehow i thought, before i was installing the vserver, that it can grab the interfaces from the main server 1135940452 M * yang Bertl: would 4 guests with 15 IPs work faster than 1 with 60 ? 1135940458 M * Bertl definitely 1135940501 M * Bertl but you can allow your guest to access all host IPs too 1135940527 M * yang yes 1135940535 M * yang that would be the best 1135940548 M * Bertl (at least it should work) by using just an entry like this: 1135940567 M * Bertl '0/ip' contains 0.0.0.0 1135940583 M * Bertl '0/prefix' contains 0 1135940589 M * Bertl '0/nodev' 1135940611 M * Bertl (no other dirs or entries) 1135940633 M * Bertl btw, this is unverified, so it might just _not_ work 1135940680 M * yang this is an excellent solution, i will try 1135940701 M * Bertl if that doesn't work, let me know, we can work around that too 1135940748 M * yang Milf: irc vhosts like "domain.tld" in your host 1135940840 M * Bertl Milf: just check http://www.sh3lls.net or similar ... 1135941117 M * yang well now 1135941120 M * yang I have on main server 1135941122 M * yang eth0:3 Link encap:Ethernet HWaddr 00:10:DC:29:8B:61 1135941122 M * yang inet addr:86.110.64.4 Bcast:86.110.64.255 Mask:255.255.255.0 1135941130 M * yang but i cannot use this IP in my vserver 1135941145 M * yang looks like that hack doesnt work 1135941177 M * Bertl enter the vserver and do 'cat /proc/self/ninfo' 1135941212 M * yang vserver1:/# cat /proc/self/ninfo 1135941212 M * yang NID: 49174 1135941212 M * yang V4Root[0]: 0.0.0.0/0.0.0.0 1135941212 M * yang V4Root[bcast]: 255.255.255.255 1135941246 M * Bertl and how did you try to 'use' it? 1135941260 M * yang i added it as hostname in my irc client 1135941271 M * yang usually should work, then ip is assigned 1135941279 M * yang when 1135941296 M * Bertl i.e. the guest is running some ircd? 1135941304 M * yang no 1135941328 M * Bertl so how is the guest IP and your irc client related? 1135941333 M * yang i did restart on vserver after cahnging config 1135941346 M * Bertl yeah, that's fine ... 1135941369 M * yang also ifconfig doesnt appear 1135941375 M * Bertl that's fine too 1135941385 M * Bertl (i.e. needs another small change) 1135941403 M * yang previously when i added 0 and 1 in interface, it used to grab the ip 1135941428 M * Bertl try to start a daemon, e.g. sshd on the guest and bind to ip 0 1135941460 M * Bertl for example, you could do sshd -d -p 2222 1135941467 M * Bertl (inside the guest) 1135941479 M * Bertl and then try to conenct to port 2222 on 86.110.64.4 1135941488 M * yang 1:17 -!- Irssi: Unable to connect server efnet.xs4all.nl port 6667 [Cannot assign requested address: 86.110.64.150] 1135941520 M * yang ok i ll try 1135941540 M * Bertl I'd assume the 86.110.64.150 is not assigned anywhere on the host, right? 1135941588 M * yang yes i added it with command ifconfig eth0 add 86.110.64.150 1135941615 M * yang but on the root server 1135941696 M * yang debug1: rexec_argv[0]='/usr/sbin/sshd' 1135941696 M * yang debug1: rexec_argv[1]='-d' 1135941696 M * yang debug1: rexec_argv[2]='-p' 1135941696 M * yang debug1: rexec_argv[3]='2222' 1135941696 M * yang debug1: Bind to port 2222 on 86.110.64.150. 1135941699 M * yang Bind to port 2222 on 86.110.64.150 failed: Cannot assign requested address. 1135941701 M * yang Cannot bind any address. 1135941804 M * Bertl why does it try to bind 86.110.64.150? 1135941819 M * yang becouse its the last interface added i think 1135941820 M * Bertl it should instead try to bind 0.0.0.0 1135941856 M * yang irssi always connects on the ip which was added the last 1135941863 M * yang on default 1135941877 M * Bertl how is irssi related to the sshd? 1135941884 M * yang well its not 1135941902 M * Bertl so, why does the sshd try to bind to 86.110.64.150? 1135942015 M * yang hehe, its becosue 1135942023 M * yang my etc/ssh/sshd config on host is 1135942025 M * yang ListenAddress 86.110.64.150 1135942036 M * Bertl okay, change that to 0.0.0.0 or comment it out 1135942187 M * yang there is another thing 1135942193 M * yang i have on my root server also this line 1135942200 M * yang ListenAddress 86.110.64.2 1135942207 M * yang in sshd config 1135942215 M * yang now i get this 1135942217 M * yang debug1: private host key: #1 type 2 DSA 1135942217 M * yang debug1: rexec_argv[0]='/usr/sbin/sshd' 1135942217 M * yang debug1: rexec_argv[1]='-d' 1135942217 M * yang debug1: rexec_argv[2]='-p' 1135942219 M * yang debug1: rexec_argv[3]='2222' 1135942222 M * yang debug1: Bind to port 2222 on 0.0.0.0. 1135942224 M * yang Server listening on 0.0.0.0 port 2222. 1135942227 M * yang socket: Address family not supported by protocol 1135942258 M * Bertl looks fine, try with ssh -p 2222 86.110.64.4 1135942264 M * Bertl or 86.110.64.2 1135942283 M * yang yang@zeus:~$ ssh -p 2222 86.110.64.4 1135942283 M * yang ssh: connect to host 86.110.64.4 port 2222: Connection refused 1135942283 M * yang yang@zeus:~$ ssh -p 2222 86.110.64.150 1135942283 M * yang ssh: connect to host 86.110.64.150 port 2222: Connection refused 1135942283 M * yang yang@zeus:~$ ssh -p 2222 86.110.64.2 1135942285 M * yang ssh: connect to host 86.110.64.2 port 2222: Connection refused 1135942306 M * sladen telnet 86.110.64.2 22 1135942308 M * sladen telnet 86.110.64.2 2222 1135942327 M * yang not even that 1135942515 M * Bertl /etc/hosts.allow? 1135942539 M * Bertl does the guest's sshd log something? 1135942550 M * Bertl or did it already terminate? 1135942564 M * yang its empty 1135942574 M * yang no it logs 1135942577 M * yang but no other 1135942591 M * Bertl okay, what did the sshd say? 1135942624 M * yang nothing else 1135942642 M * Bertl you pasted the ssh (not the sshd) messages 1135942666 M * Bertl ah, I get it, you use rexec orsoemthing like that 1135942674 M * yang vserver1:/# /usr/sbin/sshd -d -p 2222 1135942674 M * yang debug1: sshd version OpenSSH_4.2p1 Debian-5 1135942674 M * yang debug1: read PEM private key done: type RSA 1135942674 M * yang debug1: private host key: #0 type 1 RSA 1135942674 M * yang debug1: read PEM private key done: type DSA 1135942677 M * yang debug1: private host key: #1 type 2 DSA 1135942679 M * yang debug1: rexec_argv[0]='/usr/sbin/sshd' 1135942682 M * yang debug1: rexec_argv[1]='-d' 1135942684 M * yang debug1: rexec_argv[2]='-p' 1135942687 M * yang debug1: rexec_argv[3]='2222' 1135942689 M * Bertl and the -d was interpreted by your remote execution thingy 1135942689 M * yang debug1: Bind to port 2222 on 0.0.0.0. 1135942692 M * yang Server listening on 0.0.0.0 port 2222. 1135942694 M * yang socket: Address family not supported by protocol 1135942731 M * Bertl connect to the host (via ssh), then do 'vserver enter' 1135942739 M * Bertl (replace by your guest name) 1135942763 M * yang yes i am connected 1135942773 M * Bertl then execute 'sshd -d -p 2223' 1135942789 M * Bertl should not return, but instead keep running 1135942824 M * yang vserver1:/# sshd -d -p 2222 1135942824 M * yang sshd re-exec requires execution with an absolute path 1135942838 M * Bertl ts,ts must be debian :) 1135942852 M * Bertl okay, try again with /sbin/sshd or wherever your sshd is 1135942883 M * yang i get hte same message as i pasted those 10 lines 1135942903 M * Bertl the debug1 lines? 1135942909 M * yang right 1135942926 M * Bertl okay, and it is still 'running? 1135942930 J * shedi ~siggi@213-140-22-77.fastres.net 1135942951 M * Bertl welcome shedi! 1135942963 M * yang yes, root 11031 0.0 0.6 4600 1732 pts/6 S+ 11:41 0:00 /usr/sbin/sshd -d -p 2222 1135942984 M * Bertl okay, except that we are looking for -p 2223 now :) 1135943007 M * Bertl but so be it .. try now to connect via ssh -p 2223 86.110.64.4 1135943007 M * yang so i kill this and start for 2223 1135943047 M * yang i am doing all this on host 1135943071 M * Bertl (well, after the 'enter' you are on the guest) 1135943075 J * jso ~jso@ip-62-235-233-120.dial.scarlet.be 1135943082 M * Bertl welcome jso! 1135943109 M * yang i mean guest yes 1135943115 M * jso Bertl, hello ;-) 1135943137 M * Bertl yang: okay, when you 'ssh' to it, the sshd should log something 1135943162 M * yang its not logging 1135943172 M * Bertl (if it doesn't say anything, you might have a firewall running?) 1135943176 M * yang no 1135943221 M * jso yang, Bertl, did you already tried to launch sshd in debug mode? 1135943221 M * Bertl I'll test it .. give me a few minutes 1135943233 M * Bertl jso: it's running in debug mode 1135943294 M * jso mmm, so what client said in verbose mode? 1135943352 M * yang hmm 1135943356 M * yang this is strange 1135943367 M * yang yang@jason:~$ ssh 86.110.64.2 2223 1135943367 M * yang Password: 1135943461 M * yang http://pastebin.com/484014 1135943467 M * yang its not letting me in with password 1135943477 M * yang but sshd doesnt log a thing 1135943504 M * yang ahhh 1135943508 M * yang i didnt do -p 1135943512 M * harry hehe 1135943522 M * harry it's not telnet, yhou know :p 1135943544 M * yang debug1: Connecting to 86.110.64.2 [86.110.64.2] port 2223. 1135943544 M * yang debug1: connect to address 86.110.64.2 port 2223: Connection refused 1135943544 M * yang ssh: connect to host 86.110.64.2 port 2223: Connection refused 1135943579 M * harry nc 86.110.64.2 2223 1135943583 M * harry what does that tell you? 1135943613 M * harry telnet 1135943613 M * harry sry 1135943640 M * yang yang@jason:~$ telnet 86.110.64.2 2223 1135943640 M * yang Trying 86.110.64.2... 1135943640 M * yang telnet: Unable to connect to remote host: Connection refused 1135943648 M * harry no sshd running there 1135943674 M * harry so it's quite normal that you can't login on that port 1135943675 M * harry :) 1135943685 M * jso yang, in sshd_config: Port 2223 ? 1135943697 M * harry jso: not necessarity 1135943699 M * Bertl jso: the -p 2223 should override that 1135943700 M * yang no becouse sshd runs on guest and i am ssh-ing from root 1135943716 M * harry iptables -t nat -A PREROUTING -i $EXTERNAL_IF -p tcp --destination-port 2222 -j REDIRECT --to-ports 22 1135943720 M * harry that's possible too :) 1135943736 M * harry Bertl: no 1135943773 M * harry Bertl: if you ssh daemon doesn't run on port 2223, it makes no sense that your client tries to connect to it 1135943842 M * jso harry, what I mean ;-) 1135943880 M * Bertl harry: what does sshd -d -p 2223 do in your opinion? 1135943901 M * harry Bertl: 12:49 < yang> yang@jason:~$ ssh 86.110.64.2 2223 1135943902 M * harry 12:51 < yang> i didnt do -p 1135943913 M * harry that... connects to a ssh server running at port 2223 1135943916 M * harry 12:54 < jso> yang, in sshd_config: Port 2223 ? 1135943923 M * Bertl spot the difference ssh vs sshd 1135943924 M * harry that is needed to start the sshd on port 2223 :) 1135943940 M * Bertl harry: no, that's bullshit! 1135943960 M * harry you can also start it with /usr/sbin/sshd -p 2223 1135943967 M * harry true... but that wasn't the problem afaik 1135943973 M * Bertl what he did, pls read the lines 1135943998 M * yang is this value correct V4Root[bcast]: 255.255.255.255 1135943999 M * Bertl yang: I can confirm that it doesn't work this way, will investigate 1135944021 M * harry Bertl: i never seen him say sshd -p 1135944025 M * Bertl yang: what kernel/patches do you use right now? 1135944039 M * harry what he tried in the patch looked like the sshd was running on port 22 1135944044 M * yang Bertl: the ones from debian etch 1135944045 M * Bertl harry: 1135944046 M * Bertl 12:37 < yang> debug1: rexec_argv[0]='/usr/sbin/sshd' 1135944046 M * Bertl 12:38 < yang> debug1: rexec_argv[1]='-d' 1135944046 M * Bertl 12:38 < yang> debug1: rexec_argv[2]='-p' 1135944046 M * Bertl 12:38 < yang> debug1: rexec_argv[3]='2222' 1135944053 M * harry true... 1135944056 M * harry 2222 1135944059 M * harry mind the last 2 1135944063 M * harry that's not 2223 1135944068 M * Bertl he redid that with 2223 1135944083 M * Bertl maybe you want to read the irc logs _again_? 1135944089 M * harry ah, whatever... we all know what the problem is :) 1135944093 M * harry we know the sollution :) 1135944098 A * harry continues his beer 1135944098 M * harry :) 1135944123 A * tso needs coffee ... or beer too 1135944139 M * harry tso: get your ass over at 22c3 1135944142 M * harry i still have some 1135944143 M * harry :) 1135944168 M * tso harry too far away from here ;-) 1135944256 M * harry pfoeh! :) 1135944380 M * Bertl yang: try to make yourself comfortable with compiling a mainline kernel (and installing booting it on your machine) 1135944492 M * harry and try to masturbate with your hand after you've been sitting on it for 30 minutes (that's called: the stranger) 1135944500 M * harry ;) 1135944516 M * tso harry is that help? did you try it? :) 1135944569 M * harry tso: no idea... 1135944581 M * harry as long as i don't have to do it myself, i don't really care 1135944591 M * harry i have a girl that does that kind of stuff for me :) 1135944626 F * ChanServ +o Bertl 1135944639 M * yang Bertl: i like to keep debian organised kernels 1135944642 K harry Bertl come back when you're sober ... 1135944642 M * tso harry just one (girl)? ;-) 1135944652 J * harry ~harry@d515321D1.access.telenet.be 1135944654 M * harry i'm sober! 1135944665 M * Bertl then act like that ... 1135944674 M * harry naaaah... not today :) 1135944684 A * harry doing some xbox stuff now 1135944687 M * harry so cya'll 1135944740 M * Bertl yang: well, that's your decision, but if you want this to work, you'll have to add patches ... 1135944743 M * yang Bertl: i have this installed if it helps you with version revision - ii kernel-patch-vserver 2.01 1135944766 M * Bertl yang: and the patches will be against mainline+vserver ... 1135944773 M * FaUl Bertl: i had some issues with debian-installation on an ultra-1 workstation with sparc64 - i will try this tomorrow again, and then i'll test the kernel again 1135944786 M * Bertl FaUl: okay, tx! 1135944803 M * yang FaUl: managed to boot your ultra-1 ? 1135944834 M * yang nice 1135944963 M * FaUl yang: yes, but the serial terminal don't like that ncurses-interfaceu 1135944970 M * FaUl i'll try this tomorrow or so 1135944973 M * yang Bertl: so if i had the newest 2.6.14 kernel with new patches you think it would work? 1135944998 A * jso have some lunch, bbl ;-) 1135945003 M * yang FaUl: i had problems on sparc4 with menu installation, didnt want to show me colors 1135945053 M * Bertl yang: no, it won't, but I can tell you how to fix that (by changing two lines in the code) 1135945084 F * Bertl -o Bertl 1135945096 M * Bertl yang: testing that now ... 1135945156 M * yang well i hope we have the interface added correctly 1135945157 M * yang -rw-r--r-- 1 root root 8 2005-12-30 12:08 ip 1135945157 M * yang -rw-r--r-- 1 root root 0 2005-12-30 12:08 nodev 1135945157 M * yang -rw-r--r-- 1 root root 2 2005-12-30 12:08 prefix 1135945247 M * Bertl yes, I guess you have, as I said, I can reproduce it here 1135945412 J * menomc ~amery@200.75.27.106 1135945520 Q * mnemoc Ping timeout: 480 seconds 1135945520 N * menomc mnemoc 1135945538 M * yang does it work for you with such config which i have? 1135945557 M * yang you can read network from root server? 1135945650 M * Bertl yang: you need this patch: http://vserver.13thfloor.at/Devel/PAT-2.1.1/delta-anyip-feat01.diff 1135945710 M * Bertl yang: you can apply it to almost any vserver kernel, should work fine for recent debian kernels too 1135945749 M * yang aha, does it require a new kernel-compilation? 1135945755 M * Bertl yes 1135946028 M * yang patch -p1 -i delta-anyip-feat01.diff --dry-run 1135946028 M * yang can't find file to patch at input line 3 1135946028 M * yang Perhaps you used the wrong -p or --strip option? 1135946028 M * yang The text leading up to this was: 1135946028 M * yang -------------------------- 1135946030 M * yang |--- ./include/linux/vs_network.h.orig 2005-12-21 19:04:06 +0100 1135946033 M * yang |+++ ./include/linux/vs_network.h 2005-12-30 13:24:56 +0100 1135946035 M * yang -------------------------- 1135946038 M * yang File to patch: 1135946068 M * Bertl what is your current dir? 1135946083 M * yang jason:/usr/src/linux-source-2.6.12# 1135946106 M * Bertl vanilla kernel? 1135946109 M * yang yes 1135946119 M * Bertl okay, you have to apply the vserver patch first 1135946128 M * yang no, i allready have the 1135946136 M * yang kernel-patch-vserver from debian 1135946143 M * yang i got debian-source kernel 1135946155 M * Bertl okay, then check for the include/linux/vs_network.h file 1135946168 M * Bertl if it isn't there, then the patch (vserver) was not applied yet 1135946206 M * yang hmmm 1135946225 M * yang strange 1135946234 M * Bertl IIRC, then debian puts the patches in a separate dir 1135946247 M * Bertl (and does only apply them on build at some point) 1135946249 M * yang yes in /usr/src/kernel-patches 1135946269 M * Bertl you could 'simply' add the patch there (after the vserver patch) 1135946280 M * Bertl but don't ask me for details, never built a debian kernel 1135946480 M * yang i think it wont work 1135946496 M * Bertl what? 1135946500 M * yang becouse as you said, the vserver-patch comes when compileing kernel 1135946509 M * yang so i cannot add this diff before? 1135946526 M * Bertl no, you can't but you can 'append' it to the vserver patch 1135946557 M * yang shall i move it inside /usr/src/kernel-patches/diffs 1135946563 M * yang or? 1135946564 M * Bertl or just get a vanilla kernel, patch it with the vserver patch and then use the delta 1135946592 M * Bertl yang: as I said, you have to ask aba, micah or some debian folks regarding debian kernels 1135946627 M * aba yang: you can apply all patches by hand IIRC. But kernel building is still some black magic to me. 1135946629 M * yang ok, i will ask in the evening 1135946654 M * yang well i never added any patches before in my life, so i really dont know the procedure 1135946675 M * Bertl your patch line is quite fine for _adding_ patches 1135946695 M * Bertl (well, make that _applying_ patches) 1135946724 M * Bertl i.e. if you apply the vserver patch before you try with the delta, that will work 1135946746 M * yang what if i get this file vs_network.h from the CVS kernel tree, add it there and apply the diff? 1135946769 M * Bertl debian vserver cvs kernel tree? 1135946804 M * yang no, the kernel.org CVS ? 1135946825 M * yang well i am just guessing 1135946827 M * Bertl ah, well, then you can get a vanilla kernel and patch that with vserver and then the delta 1135946840 M * Bertl works quite fine (that's what I do) 1135946865 M * Bertl http://www.kernel.org/ 1135946886 M * Bertl get the 2.6.14.5 kernel from there (probably the 'F'ull version) 1135946889 M * yang so, you tested this vserver which we worked on, and it works on you without problem? reading interfaces from main server? 1135946907 M * Bertl yes, after applying the delta 1135946920 M * yang ok then i ll do that 1135946955 M * Bertl http://www.13thfloor.at/vserver/s_rel26/v2.01/ 1135946969 M * Bertl here are the stable patches (they will work fine for 2.6.14.5) 1135947009 M * Bertl but you should first try with a vanilla kernel (i.e. no patches at all) as debian might need some modifications to boot a vanilla kernel 1135947022 M * Bertl some things come to my mind here: 1135947025 M * Bertl - udev 1135947033 M * Bertl - no cramfs 1135947050 M * yang i will just use the debian kernel config, without modifications... 1135947068 M * Bertl which might give you issues and is not the best choice 1135947097 M * ichigo yang: do you want to compile kernel in deiban..or for any other distro even you can use the "make oldconfig" method after applying the patch. 1135947103 M * Bertl yang: instead you should try to figure what hardware your host really has, and configure a minimal monolithic kernel 1135947105 M * yang well if the kernel doesnt boot, i am in trouble since the server is in colocation 1135947139 M * Bertl yang: if you do not have (some kind of remote) console access, I would be _very_ careful 1135947152 M * yang that is what i am saying 1135947160 M * ichigo Bertl: a small kernel is usefull but few know what to include and what to discard. best to use the orignal kernel as a guideline and work from there. 1135947163 M * yang debian default kernels boot fine... 1135947203 M * Bertl yang: then wait for micah, and/or find some debian folks who can tell you how to _add_ the delta to the debian vserver-kernel source package 1135947215 M * yang yes, better 1135947273 M * Bertl ichigo: well, the 'original' (debian) kernel also includes the kitchen-sink, which is seldom present on servers :) 1135947288 M * tso yang yes, but debian def. kernel loads most of device drivers from module, and has an initrd...so if you don't know exactly what drivers are needed, you should compile with debian kernel config and make initrd before booting it up 1135947293 M * ichigo yang. then all you need to do is patch a vannilla kernel. copy the configuration into .config and do make oldconfig. then answer some questions. and you'll have a kernel that work well on your distro and computer. 1135947354 M * Bertl ichigo: unlikely 1135947386 M * ichigo bertl: hahahahah very funny. but seropisly. few ppl can claim to master kernel compiling ..... even i who has compiled about 50 kernels cannot. and its always better to use oldconfig cuz then you end up with a kernel that is well supported on your distro 1135947429 M * Bertl ichigo: I would be _very_ surprised if using the debian default config on a recent 2.6.14.5 kernel would give anything bootable (on debian) 1135947443 M * ichigo i did. but it was 2.6.14.4 1135947458 M * Bertl really? scsi hardware? 1135947464 M * ichigo see the config it does not find in the default. it just asks?? 1135947476 M * ichigo no no scsi hardware?? 1135947503 M * ichigo ofcoase mkinitramfs is also needed. 1135947514 M * Bertl from what 'kernel' was the old .config? (debian) 1135947527 M * ichigo i think it wa 2.6.8. 1135947560 M * ichigo i also did the same for ubuntu..that was 2.6.12 1135947580 M * Bertl well, I have to trust you on that, but IIRC the IDE config changed significantly and a 2.6.8 .config would not lead to a proper kernel when used on 2.6.14 1135947601 M * Bertl (instead it would miss the IDE disk support) 1135947606 M * yang i am scared on booting a vanilla kernel, if the machine would fail... 1135947609 M * tso and for sata too 1135947652 M * tso yang do you use grub? 1135947668 M * yang lilo 1135947670 M * Bertl yang: well, that is caused by your provider/colocator who does not provide some remote access 1135947671 M * ichigo i agree with what you say too. BUT for ppl who don't know what the 100000s of options mean it better to use make oldconfig. if you wat some chages then after oldconfig do make menuconfig 1135947719 M * Bertl ichigo: usually the 'make defconfig' gives a sane default 1135947726 M * tso ichigo yes of course, except if he uses special sata or scsi controllers, and it depends on old kernel version as Bertl said before 1135947763 M * Bertl you then walk the options (with make menuconfig) and read the help ... hardware you don't have get deselected, hardware you know you have can be added/configured 1135947824 M * ichigo yang lang. i just thought up a good solution for you. do uname -r. you'll get the kernel you use. get the vannilla source for that kernel. try compiling that a few time to get experinance then move on to the latest kernels... 1135947848 M * Bertl ichigo: I agree with you that 'make oldconfig' is the way to go, but not with a kernel with everything selected :) 1135947882 M * tso i didn't use lilo for a long time, is there any feature like grub-reboot? 1135947888 M * Bertl ichigo: how would that help? 1135947899 M * ichigo Bertl: to select everything i think you use make allconfig. no that scary but even that has few disadvatages except one mothership of a kernel. 1135947905 M * Bertl tso: yes, actually it came from there, it's the lilo -R option 1135947928 M * Bertl ichigo: the debian kernels are like 'allmodconfig' 1135947940 M * ichigo Bertl. it would help cuz he could easiily use make oldconfig without anyworries. 1135947968 M * ichigo Bertl:yeah i am not sure about the exact command i just red about it in the doc once 1135947971 M * Bertl ichigo: wrong again, because even if the kernel says 2.6.8-something, it's really more a 2.6.10 or 11 1135947976 M * tso Bertl thx, then yang can reboot the machine with the new kernel and ask for colocation provider to reboot the machine if something goes wrong.. 1135948023 M * ichigo Bertl. now that not my fault is it.. i tend to belive the output of uname -r 1135948050 M * Bertl ichigo: well, but debian does not update the release numbers, only the -XX 1135948072 M * Bertl so 2.6.8-1 was a 2.6.8 kernel (plus some changes)= 1135948097 M * Bertl but 2.6.8-42 might be a 2.6.14 kernel (minus some changes) 1135948119 M * ichigo really?? ohh who'd have thought. but i suppose there must be some way to know the exact kernel number. maybe ask on the debian irc 1135948164 M * Bertl the thing is, they do not update to the newer kernels, they just add patches (between the kernels) which they consider 'useful' or 'necessary' 1135948186 M * Bertl only after some time, they do update ... 1135948221 M * Bertl but, nevertheless there might be a proper config for 2.6.14 too, because IIRC, micah uploaded some vserver kernel with 2.6.14 to unstable or so 1135948243 M * Bertl that might even have the proper defaults for linux-vserver 1135948308 M * ichigo why not try to install the 2.6.14 kernel form the sid repos. than see the config for that and use that. there is a 2.6.14 kernel in the sid repos. i know since i run sid inside my vserver and alteast know whats in the repos. 1135948341 M * Bertl well, that should do too 1135948352 M * ichigo he where did yang go???? my entire page is without a single post from him.... 1135948657 M * ichigo 1135948679 M * ichigo ......where is everybody. 1135948706 A * tso drinking a coffee 1135948715 N * ichigo ichigo___food 1135948791 N * ichigo___food hollow_ichigo 1135948840 A * jso bb ;-) 1135949152 M * jso Bertl, btw iirc, by default, debian disable ssh as root ( PermitRootLogin no), though? 1135949167 M * tso jso yes 1135949223 M * Bertl yes, might be so, nevertheless if you are referring to yang's issues, this does not apply 1135949275 M * tso jso sry, newly installed ssh not disable it, i was wrong 1135949352 M * jso tso, no pb (here my unstable i386 said no ;-) but should be a custum of mine, i don;t remember) 1135949400 M * tso jso some minutes earlier i debootstrap a new sid and sarge with ssh and permitrootlogin yes in both of them 1135949422 M * hollow_ichigo yes sid atleast has it on by default. 1135949470 M * jso Bertl, here with kernel 2.6.14.4+ vs-2.1.0 (hppa box), I don't reach to reproduce ? 1135949489 M * jso tso, cool (definitely of mine ;-) ) 1135949523 Q * shedi Quit: Leaving 1135949852 M * jso Bertl, mmm I have got this: 1135949854 M * jso # ssh -vvv -p2223 hplvsrvr 1135949854 M * jso OpenSSH_4.2p1 Debian-5, OpenSSL 0.9.8a 11 Oct 2005 1135949854 M * jso debug1: Reading configuration data /root/.ssh/config 1135949854 M * jso debug1: Reading configuration data /etc/ssh/ssh_config 1135949854 M * jso debug1: Applying options for * 1135949855 M * jso debug2: ssh_connect: needpriv 0 1135949857 M * jso debug1: Connecting to hplvsrvr [192.168.248.145] port 2223. 1135949859 M * jso debug1: connect to address 192.168.248.145 port 2223: Connection refused 1135949863 M * jso ssh: connect to host hplvsrvr port 2223: Connection refused 1135950067 M * jso Bertl, just because I launch sshd in my guest with -d (in this case only one attempt is possible and have to re-launch sshd, so all came back ok ;-) 1135950203 J * jrso ~Joel_Soet@ip-62-235-233-120.dial.scarlet.be 1135950242 M * Bertl welcome jrso! 1135950346 M * jrso Bertl, apologies for noise (I just made a mistake by looking for this irc log) (jrso == jso ;-) ) 1135950392 M * Bertl np :) 1135950411 M * Bertl just haven't realized that jso is you :) 1135950468 M * jso Bertl, obviously ;-) 1135950472 M * eyck yeah, that was sneaky.. 1135950495 Q * jrso Quit: ChatZilla 0.9.61 [Mozilla rv:1.7.12/20051007] 1135950602 M * hollow_ichigo hey bertl... earlier you mentioned that to get my usb devices to work under the vserver i will have to copy the nods. how exactly. i assume its not simply a cp 1135950785 M * eyck cp -a 1135950856 M * eyck pokurcz:~# cp -a /dev/ram1 RAM1 1135950856 M * eyck pokurcz:~# ls -l RAM1 1135950856 M * eyck brw-rw---- 1 root disk 1, 1 Mar 25 2005 RAM1 1135951211 J * bubulak_ ~bubulak@cicka.wnet.sk 1135951385 Q * bubulak Ping timeout: 480 seconds 1135951415 N * bubulak_ bubulak 1135952275 J * dothebart titan@p5086644C.dip.t-dialin.net 1135952365 Q * bubulak Ping timeout: 480 seconds 1135952404 M * jso Bertl, I just read back what I missed on yang issue; what is missing me is a 'netstat -an' from the guest? 1135952415 Q * dothebar1 Ping timeout: 480 seconds 1135952797 J * MostlyHa1mless ~mh@melbourne.mostly-harmless.ca 1135952808 Q * MostlyHarmless Read error: Connection reset by peer 1135953092 P * undefined 1135953138 J * robig ~robig_m@envoppp47.envia-tel.de 1135953143 M * robig hi 1135953176 M * robig is there a patch for 2.6.14.5 kernel? 1135953186 M * robig + available 1135953214 M * Bertl the one for 2.6.14.3/4 should work quite fine 1135953231 M * Bertl if you encounter the slightest issues (when applying) let me know 1135953291 M * robig you mean http://www.13thfloor.at/vserver/s_rel26/v2.01/patch-2.6.14.3-vs2.01.diff ? 1135953360 M * Bertl for example 1135953371 M * hollow_ichigo thanks eyck. i will try that. 1135953420 M * hollow_ichigo but my scanner nodes are created under /proc/bus/isb and that is reset every time a start the comp. how can i gt that. 1135953435 M * hollow_ichigo */proc/bus/usb 1135953471 Q * MostlyHa1mless Ping timeout: 480 seconds 1135953493 M * Bertl hollow_ichigo: mayb do some rm and cp -va on every startup? 1135953518 J * bubulak ~bubulak@cicka.wnet.sk 1135953568 M * jso Bertl, I was on going to forget to put my question: what's about ipv6 support (in develop 2.1.0)? 1135953585 M * Bertl host yes, guest still no, will have to wait for ngnet 1135953621 M * jso Bertl, ah thanks ;-) 1135953768 M * Bertl jso: but ngnet is the next thing on my todo list, a new scheduler is just finished ... 1135953873 M * FireEgl =D 1135953905 M * jso Bertl, cool (mmm, I will certainly try next ... year ;-) well I mean next week) 1135954384 J * Doener doener@i5387DA36.versanet.de 1135954446 M * Bertl welcome Doener! 1135954455 M * Doener hi Bertl! 1135954482 M * Bertl you just missed a very interesting discussion on the OVZ channel 1135954512 M * Doener are there logs? 1135954524 M * Bertl well, I can give you the short version: 1135954557 M * Bertl except for the networ virtualization, OVZ has one important feature which is missing in LVS, the limit on lowmem 1135954701 M * hollow_ichigo whats OVZ??? 1135954710 N * hollow_ichigo ichigo 1135954711 M * yang Is there support for IPv6 interface in vserver? 1135954729 M * Adrinael yang, coming up. Don't hold your breath. 1135954734 M * Bertl yang: on the host yes, inside the guest not yet 1135954804 M * Doener ichigo: openvz, FLOSS spin-off of Virtuozzo 1135954814 M * Bertl ichigo: the open soruce releases from a company called SWSoft which usually sells Virtuozzo (which is some kind of 'competing product') to linux-vserver 1135954845 M * Bertl Vortuozzo(tm)(R) that is :) 1135955433 M * ichigo ok then. 1135955474 M * ichigo thanks fo the nice explanation. 1135955486 M * Roey hello 1135955487 M * Roey Bertie 1135955492 M * Roey Adrenal 1135955493 M * Roey etc 1135955494 M * Roey whatup 1135955497 M * Roey *Adrinael, sorry 1135955509 M * Roey Adrinael: why is your nick like a gland 1135955532 M * Roey How can I get vservers to start upon bootup? 1135955541 M * Roey I finally decided to just... use a 100 MB / 1135955545 M * Roey and have everything mounted unert here. 1135955548 M * Roey under there. 1135955560 M * Adrinael Roey, I don't know. I blame my left cortex. 1135955568 M * Roey =) 1135955830 M * Roey ok 1135955849 M * Roey Bertl: is there some way I can do a netstat -anp on the host and see all of the vserver clients' open ports? 1135955877 M * Roey Bertl: also, how can I get the clients to start automatically on bootup? 1135956086 M * Doener Roey: there comes a init script with the tools that is nicely commented 1135956105 M * Roey Doener: I know 1135956113 M * Roey Doener: I looked at it. There are two. legacy and default. 1135956117 M * Roey Doener: I don't know which does which. 1135956139 M * Doener IIRC you create a file called 'mark', put the 'group' name in, e.g. 'default' and then add the script to your runlevel 1135956173 M * Doener the name of the script determines which vservers are started 1135956189 M * Doener (some part of the script name has to equal the group name in the 'mark' file) 1135956213 M * Roey hmm 1135956214 M * Doener so the default vserver-default script starts all vservers in group 'default' 1135956225 M * Roey there is a group name? 1135956227 M * Roey what group name? 1135956229 M * Roey UNIX group? 1135956232 M * Doener for further details check the comments in that script 1135956236 M * Roey where do you put this mark file? 1135956237 M * Doener 16:22:19 Doener IIRC you create a file called 'mark', put the 'group' name in, e.g. 'default' and then add the script to your runlevel 1135956246 M * Roey ok 1135956253 M * Roey I don't know what the script is though. 1135956259 M * Roey just a file called 'mark' ?? 1135956265 M * Doener 16:23:34 Doener so the default vserver-default script starts all vservers in group 'default' 1135956300 M * Doener vservers-default actually... with an 's' 1135956391 Q * ichigo Ping timeout: 480 seconds 1135956476 M * Roey Doener: ok.. 1135956503 M * Roey export MARK NUMPARALLEL LOCKFILE 1135956503 M * Roey exec /usr/lib/util-vserver/vserver-wrapper "$@" 1135956508 M * Roey that's what S90vservers-default says. 1135956533 M * cehteh t# vserver test2 build -m debootstrap -- -d sarge ARCH=i386 1135956533 M * cehteh /usr/local/etc/vservers/.defaults/vdirbase/test2/debootstrap 1135956533 M * cehteh /usr/local/lib/util-vserver/functions: line 206: -n: command not found 1135956534 M * cehteh :( 1135956594 M * Adrinael cehteh, test2 already exists 1135956598 M * Adrinael Put --force in there 1135956630 M * cehteh emp: No such script: /usr/lib/debootstrap/scripts/ARCH=i386 1135956640 M * cehteh .. try'n'error day :P 1135956672 M * Adrinael -d sarge --arch=i386 ain't it? 1135956698 M * cehteh /usr/local/lib/util-vserver/vserver-build: unrecognized option `--arch=i386' 1135956717 M * Roey doh. 1135956724 M * cehteh seems to be an flaw in vservers opt parser 1135956806 M * Roey Doener, Bertl: I still do not understand how to get vserver clients started automatically upon system bootup. I have three servers: mail, web and dns. 1135956820 M * Adrinael cehteh, did you put that after the -- ? 1135956826 M * cehteh Adrinael: yes 1135956836 M * Adrinael Hm, then there's a bug 1135956846 M * cehteh yes 1135956864 M * cehteh thats what i wanted to say :P 1135956955 M * Doener Roey: did you read the comment right above the #MARK=default line in the initscript? 1135956994 M * Doener that tells you where to put the mark file and what to put in there... 1135956998 M * cehteh exec $_VNAMESPACE --new -- $_VSERVER ----nonamespace "${OPTIONS_ORIG[@]}" .. eh looking at the source, is the ----nonamespace correct anyways? 1135957007 M * Doener for the default script it is 'default'... 1135957044 M * Doener by uncommenting that line and changing the value you can use other 'groups'/'flavours'... 1135957047 M * Roey Doener: I didn't know what a "flavor" is. 1135957074 M * Roey Doener: and really it says nothing about 'startup' 1135957078 M * Roey it just says 'selected' 1135957083 M * Roey which I have no idea what it means in this context. 1135957085 M * Roey :( 1135957092 M * Roey can't it just say 1135957112 M * Roey "put the startup order you want in THIS FILE" or something like that ? 1135957116 M * Doener "the 'default' vservers will be started very late ... other vservers can be selected" 1135957119 Q * yang Quit: reboot 1135957127 M * Doener that'd not be what it does... 1135957131 M * Roey or "servers will be automatically started according to the order specified in ..." 1135957145 M * Roey Doener: I have no clue what that line says! "the 'default' vservers will be started very late ... other vservers can be selected" 1135957149 M * Roey Doener: maybe I am just not smart. 1135957161 M * Roey I read it a buncha times before I even asked this question fifteen minutes ago 1135957167 M * Roey and I stillc annot understand what that line is saying 1135957182 M * Roey Doener: (I don't mean to be disrespectful in any way btw) 1135957187 M * Roey Doener: (I am just frustrated) 1135957202 M * Roey http://dictionary.reference.com/wordoftheday/archive/2005/12/29.html 1135957277 M * Doener the script is about starting vservers... then it says 'ok, the default ones get started quite late, if you need some to start earlier, copy this script adjust the MARK value and put according marks into your vserver configuration'... so that copy will then start a different 'flavor'/'group' of vservers, marked with instead of 1135957296 M * Roey ok 1135957296 M * Roey but 1135957300 M * Roey I don't know what 'quite late' means 1135957309 M * Roey does 'quite late' mean "not started automatically" 1135957310 M * Roey ?? 1135957320 M * Doener 16:28:28 Roey that's what S90vservers-default says. 1135957343 M * Doener S90 <-- the 90 means the script is executed almost at the end of the boot process 1135957343 M * Roey I know, I think the wording is convoluted to the point that it is hard to understand. 1135957347 M * Roey right. 1135957350 Q * virtuoso Ping timeout: 480 seconds 1135957367 M * Roey Doener: then say "started late due to the low init priority (90) of this script" 1135957383 M * Roey Doener: but that doesn't say anything about auto-starting the clients. 1135957389 M * Roey Doener: my clients don't start at all autoamtically 1135957394 M * Roey Doener: I'm trying to figure out why. 1135957411 M * Doener are they marked as belonging to the default 'group'/'flavor'? 1135957479 M * Doener echo default > /etc/vservers//apps/init/mark 1135957507 M * Doener (might need to create the last one or two directories first) 1135957672 J * yang ~yang@cpe-212-18-59-124.dynamic.amis.net 1135957685 M * Bertl wb yang! 1135957708 M * yang damn 1135957723 M * yang I booted the server, and it came up well 1135957723 J * virtuoso ~s0t0na@shisha.spb.ru 1135957750 M * yang but after vserver1:/# /usr/sbin/sshd -p 2222 -d it has frozen 1135957759 M * yang i dont receive any response 1135957889 M * yang i have applied that network patch to it 1135957895 M * yang and recompiled 1135957907 M * Bertl and booted the 'patched' kernel, yes? 1135957926 M * yang yes 1135957961 M * Bertl and you can not reach the server anymore? 1135957967 M * yang no 1135957969 M * cehteh mhm where is the vserver-build scriot normally installed? 1135957983 M * cehteh .. or better i am to lazy, where can i post bugs :P 1135958000 M * Bertl cehteh: best on the ML 1135958036 M * cehteh Bertl: you dont have any other idea how to pass the --arch opt to debootstrap? 1135958050 J * lilalinux ~plasma@h1-gw.of.net-lab.net 1135958057 M * Bertl cehteh: why, doesn't ARCH= work? 1135958075 M * cehteh emp: No such script: /usr/lib/debootstrap/scripts/ARCH=i386 1135958077 M * Bertl cehteh: it worked for me on x86_64 to get an x86 guest installed 1135958099 M * cehteh vserver test2 build --force -m debootstrap -- -d sarge ARCH=i386 1135958119 M * Bertl try ' ARCH=i386 vserver test2 build --force -m debootstrap -- -d sarge 1135958126 M * Bertl welcome lilalinux! 1135958142 M * cehteh ah that worked 1135958150 M * cehteh at least it seems so 1135958153 J * yang2 ~yang@cpe-212-18-59-124.dynamic.amis.net 1135958173 M * cehteh but that --arch doesnt work is still a bug i think .. 1135958176 Q * yang Quit: leaving 1135958177 M * Bertl yang2: sorry, the patch has a bug 1135958189 M * yang2 ... 1135958197 M * Bertl just saw it 1135958270 M * Bertl yang2: fixed the issue right now, so please reload the patch 1135958296 M * Bertl yang2: host should work fine as long as you do not start/run guests inside 1135958354 M * lilalinux hi Bertl 1135958426 M * yang2 what do you mean "not start guests inside" , how else should i start vserver? 1135958482 M * Bertl yang2: no, what I mean is, the patch you applied was broken (again I'm terribly sorry for that), but if you boot with the patched kernel, you should be able to fix it (i.e. recompile a new kernel) 1135958498 Q * prae Quit: Execute Order 69 ! 1135958505 M * Bertl yang2: as long as you do not start guests (which do networking) 1135958537 M * yang2 i gotta wait for the NOC to come back from the new years partys... 1135958562 M * Bertl they are already celebrating? 1135958593 M * Bertl btw, a hint (for next time) add panic=60 or so to your kernel boot options 1135958612 M * Bertl that will automatically reboot your server if something bad happens 1135958852 M * yang2 ok 1135958878 M * Bertl I know, doesn't help you right now ... 1135958899 M * yang2 so i gotta apply a fixed patch 1135958906 M * yang2 and then we hope it works 1135958920 M * Bertl yes, did you apply it yourself or build some debian kernel with it? 1135958928 M * yang2 apply myself 1135958936 M * yang2 and make-kpkg debian kernel 1135958940 M * Bertl okay, keep the old one (once your server is back) 1135958970 M * Bertl the old patch I mean, and download the new one (http://vserver.13thfloor.at/Devel/PAT-2.1.1/delta-anyip-feat01.diff) somewhere else 1135958995 M * Bertl then remove the old one with 'patch -p1 -R' 1135959021 M * Bertl (either patch -p1 -R -i old.patch or patch -p1 -R /etc/vservers/web/apps/init/mark' 1135963226 M * Roey for 'web', 'mail', and 'dns' clients. 1135963232 M * Roey let's see if they start on system bootup now. 1135963276 M * jso Roey, it will only start 'web' vserver ? 1135963333 M * Roey jso: well I did that statement three times. One for 'web', one for 'dns' and one for 'mail'. 1135963353 M * jso Roey, ok so it will do the drill ;-) 1135963366 M * cehteh Bertl: if i make a vinit wraper for putting the the root server into its own context are you interested in adding it to the distribution (if you say yes, then i'll likely put bit more generic efforts into the script rather than just a dirty hack) 1135963384 M * Roey Bertl, Doener, jso: hey it worked!!!! 1135963433 M * Roey Bertl, Doener, jso: ok, is it standard practice to make changes to the clients by chrooting into them, or rather to SSH into them (since they have different IPs) 1135963434 M * Roey ? 1135963446 M * Doener ssh 1135963448 M * Bertl cehteh: that is userspace, so you have to ask enrico ... 1135963489 M * Roey Bertl, Doener, jso: and if I chroot or do 'vserver mail enter' can I make root un-loginable by putting a * for its password in the client's /etc/passwd? 1135963555 Q * Milf Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org 1135963562 M * cehteh mhm is enrico here on irc? 1135963577 M * Bertl cehteh: unfortunately no ... 1135963622 M * cehteh well .. maybe i'll just do it an then send a patch, i need it anyways and it seems not big 1135963640 M * jso Roey, is ssh is your only login entry? (if yes there is the option 'PermitRootLogin no' in sshd_config to do so ;-) 1135963657 Q * soltesz Remote host closed the connection 1135963662 M * cehteh i actually wonder that noone asked for such .. or maybe that because i dont read the ML often 1135963736 M * Roey jso: I think I'd rather make it so that the only way to administer changes to clients is to do it via 'vserver enter' 1135963736 M * Roey jso: and that avoids 'login' completely. 1135963740 M * Roey jso: so I can set root's password to '*' 1135963836 M * Bertl Roey: you can do that, but be aware that using 'enter' is less secure than using 'ssh' 1135963846 M * Roey Bertl: oh??? why?? 1135963850 M * jso Roey, just stop login services so (sshd, telnetd and so on) 1135963859 M * Roey jso: ok. 1135963863 M * Roey Bertl: how is it less secure? 1135963895 M * Bertl Roey: because the 'enter' comes from the host system and 'possibly' brings connections from there into the guest 1135963920 M * Roey connections?? 1135963921 M * Roey like what? 1135963929 M * Roey you mean that it inherits the environment of the host? 1135963939 M * Bertl like ther terminal or open filehandles or such 1135963945 M * Roey ok. 1135963951 M * Roey I don't know what could possibly be wrong with that though 1135963955 M * Roey (what could go bad?) 1135963959 M * Roey s/bad/wrong 1135963964 M * Roey s/wrong/wrongly 1135963967 M * Roey it's an adverb. 1135963998 M * Bertl well, in the worst case scenario, a malicious guest binary could take over your host ... 1135964044 M * Roey ok. 1135964053 M * Roey I thought that the host can mess with the guest anyway though. 1135964068 M * Roey because host's 'root' user can modify the guests' file tree. 1135964081 M * Bertl yeah, but not the other way round :) 1135964088 M * Roey how can I tell which debian stream I installed to the guest? 1135964108 M * Roey you're saying that the guest may mess the root's file tree??? 1135964109 M * Bertl IIRC, there is something like /etc/*release* 1135964109 M * Roey hmm. 1135964112 M * Roey ok 1135964130 M * Roey no...no there si not. 1135964143 M * Adrinael It's /etc/debian_version 1135964155 M * Bertl yep, right ... 1135964176 M * Roey nice 1135964176 M * Roey thanks 1135964202 M * Roey it just says '3.1' 1135964206 M * Roey is that Sarge then? 1135964208 M * cehteh chbind: vc_set_ipv4root(): Function not implemented 1135964209 M * cehteh ... bah 1135964231 M * Roey I am asking because they didn't have an /etc/apt/sources.list file defined, and I wanted to know which source lines to feed them. 1135964246 M * cehteh i guess i should turn on the legacy net interface in my kernel? 1135964282 M * Roey what's that? 1135964296 M * cehteh oh it is on 1135964303 M * cehteh err .. diabled 1135964323 M * cehteh Bertl: negated flags in the kernel config are evil :P 1135964332 M * cehteh [*] Disable Legacy Networking Kernel API 1135964347 M * Bertl cehteh: defaults are good :) 1135964371 M * cehteh who wants a legacy system nowadays!? :) 1135964389 M * cehteh [ ] Enable Legacy Kernel API shall i turn that on too? 1135964411 M * Bertl what's the default? 1135964415 A * cehteh thought he could turn that off because he used the most recent utils 1135964438 M * Bertl cehteh: well, you could get hollows tools, they support the non-legacy networking api 1135964548 M * cehteh ... /me rebuild the kernel 1135964708 M * Roey anyone here using apt-proxy? 1135964866 M * jso Bertl, cehteh, afaik we still need 'Enable Legacy Kernel API' ( but may be do I wrongly unsderstand your followup,though) 1135964937 M * jso Roey, no sorry (still at the bottum of my todo list ;<) ) 1135964953 M * Bertl jso: hmm, no, how do you conclude that? 1135965025 M * Roey jso: ok 1135965031 M * Roey jso: I mean I mostly got it working on another servr. 1135965036 M * Roey I don't know what I'm doing wrong 1135965037 M * Roey but anyway. 1135965046 T * Bertl http://linux-vserver.org/ | latest stable 2.01, 1.2.10, 1.2.11-rc1, devel 2.1.0, exp 2.1.0.1 | util-vserver-0.30.209 | He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the wiki, and we'll forget about the minute ;) 1135965067 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.15-rc7-vs2.0.1.diff 1135965073 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.15-rc7-vs2.1.0.diff 1135965078 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.15-rc7-vs2.1.0.1.diff 1135965115 M * jso Bertl, Cool 2.6.15 stuff . 1135965118 M * Bertl I would be especially interested in feedback regarding 2.1.0.1 as it has a completely new hard scheduler 1135965284 M * jso Bertl, back to legacy: if I disable CONFIG_VSERVER_LEGACY it solved the test pb but I could start any more vservers because it disable dynamic context support. 1135965480 M * Bertl hmm, dynamic contexts should not be required for enricos tools, no? 1135965642 M * jso Bertl, oops I missed some link, so sorry, I will try first to find enrico's tools first. 1135965965 M * Roey YEHORAM GAON 1135966037 M * Bertl hmm, musician? 1135966087 A * jso just grab patch-2.6.15-rc7-vs2.1.0.1.diff, just have lunch and come back soon ;-) 1135966391 T * * http://linux-vserver.org/ | latest stable 2.01, 1.2.10, 1.2.11-rc1, devel 2.1.0, exp 2.1.0.1 | util-vserver-0.30.209 | He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the wiki, and we'll forget about the minute ;) 1135966391 T * Bertl - 1135966437 J * SNy 22b7ab81d9@bmx-chemnitz.de 1135966441 J * Vudumen vudumen@perverz.hu 1135966455 J * soltesz ~soltesz@targe.CS.Princeton.EDU 1135966487 J * mnemoc ~amery@200.75.27.106 1135966491 M * Bertl welcome soltesz! 1135966511 M * soltesz hello :-) 1135966643 M * eyck Bertl: woaa, you were right,, via ssh -X it works... 1135966650 M * Roey Bertl: ok :) 1135966651 J * emp ~emp@70.57.239.35 1135966652 M * Roey Bertl: fine 1135966695 M * eyck Bertl: have you got any idea how can I check what's missing from guest's X? 1135966754 J * meandtheshell ~markus@85-124-14-196.dynamic.xdsl-line.inode.at 1135966951 J * pusling pusling@195.215.29.124 1135967056 M * Bertl welcome meandtheshell! 1135967062 M * Bertl welcome pusling! LTNS! 1135967084 M * meandtheshell Bertl: evening bertl ;-) 1135967086 M * pusling Bertl: shouldn't it just be welcome back ;) 1135967159 J * stefani ~stefani@superquan.apl.washington.edu 1135967221 M * Bertl wb stefani! :) 1135967225 M * Roey hi 1135967226 M * Roey Bertl 1135967234 M * Roey I'm hyaving trouble apt-get updating from within the client. 1135967258 M * Roey I have my name resolver set up OK though. 1135967263 M * Roey in /etc/resolv.conf. 1135967277 J * Psy0rz ~psy0rz@lounge.datux.nl 1135967281 M * Bertl welcome Psy0rz! 1135967304 M * Bertl Roey: do you use externalized or internalized package management? 1135967326 M * stefani hola Bertl 1135967339 J * dothebart ~willi@xdsl-87-78-46-87.netcologne.de 1135967453 J * sebi ~sebi@Fcdbf.f.strato-dslnet.de 1135967517 M * Roey Bertl: well the clients have their own little world 1135967521 M * Roey Bertl: I'm not using unify 1135967581 M * Bertl Roey: that's fine, but IIRC, debootstrap defaults to internalized anyway ... 1135967630 Q * comfrey Ping timeout: 480 seconds 1135967803 M * Bertl Roey: so what do you get when you do 'apt-get update'? 1135967870 M * Roey root@web:/# apt-get update 1135967870 M * Roey Err http://debian.csail.mit.edu testing/main Packages 1135967870 M * Roey 404 Not Found 1135967872 M * Roey things like that. 1135967879 M * Roey it does pick up one source that it can find though 1135967888 M * Roey Get:1 http://ftp.us.debian.org stable/main Packages [3349kB] 1135967890 M * Roey but... 1135967896 M * Roey I don't know if it is doing amd64 or ia32 1135967899 M * Roey the kernel is amd64 1135967909 M * Roey the host's userland uses amd64 libs. 1135967912 M * Roey libc rather. 1135967930 M * Bertl well, the guest will run with ia32 personality I assume 1135967947 M * Bertl nevertheless you might want to make that explicit 1135968013 M * Roey ok. 1135968094 M * Roey root@web:/# apt-get update 1135968094 M * Roey Err http://ftp.de.debian.org testing/main Packages 1135968094 M * Roey 404 Not Found 1135968101 M * Roey Bertl: I have this source which works fine in the host: 1135968113 M * Roey deb http://ftp.de.debian.org/debian-amd64/debian/ testing main contrib non-free 1135968118 M * Roey but it gives me the 404 in the client. 1135968120 M * Roey what gives? 1135968288 M * Bertl check the resolv.conf entries, amybe they differ? 1135968307 M * jso Roey, what your dns answers (nslookup ftp.de.debian.org?) 1135968335 M * Roey lemme check. 1135968351 M * Bertl but it could easily be an x86_64 vs x86 issue ... 1135968371 M * Roey root@web:/# ping ftp.de.debian.org 1135968371 M * Roey PING ftp.de.debian.org (141.76.2.4) 56(84) bytes of data. 1135968371 M * Roey 64 bytes from ftp.de.debian.org (141.76.2.4): icmp_seq=1 ttl=47 time=132 ms 1135968376 M * Bertl (if apt-get is trying to be smart and deduces the arch from the uname) 1135968388 M * Roey Bertl: I don't know. 1135968395 M * Roey I only know that it's not apt-get updating :) 1135968402 M * Bertl try to make some adjustments to the config 1135968409 M * Roey how so? 1135968410 M * Bertl the guest config 1135968415 M * Roey Bertl: ok.. where in it? 1135968418 M * Bertl first, add the proper architecture ... 1135968503 M * Roey ok. 1135968548 M * Roey deb http://ftp.de.debian.org/debian-amd64/debian/ testing main contrib non-free 1135968557 M * Roey this is what the guest's /etc/apt/sources.list says. 1135968616 M * Roey btw the guest's /etc/debian_version says 3.1 1135968619 M * Roey so that is Stable. 1135968637 M * Roey and since there is no x86_64 version of Stable, then the stream must be Testing. 1135968641 M * Roey (i.e. Etch). 1135968659 M * Bertl # /etc/vservers//uts/machine 1135968664 M * Bertl add i386 there 1135968666 M * Roey what's that? 1135968706 M * Roey Bertl: there is only 'nodename' in that directory. 1135968744 M * cehteh Roey: http://ftp.de.debian.org/debian-amd64/debian-pure64/dists/stable/ 1135968760 M * Bertl Roey: yeah, add a file called 'machine' which contains 'i386' 1135968776 M * Bertl then restart your guest and check (inside the guest) with uname -a 1135968890 M * Roey ok one sec. 1135968902 M * Roey I mean eventually I just want to make it localhost:9999/ 1135968908 M * Roey and use the host's apt-proxy 1135968917 M * Roey er, not localhost, but the apt-proxy server's ip, rather. 1135968930 M * Bertl you could do that, but the host is x86_64 no? 1135968951 M * Roey yeah. 1135968954 M * Roey the kernel is. 1135968961 M * Bertl and the guest is i386 1135968970 M * Bertl so they won't share too many packages I guess 1135969013 M * Bertl btw, also add # /etc/vservers//personality 1135969064 M * Bertl with linux_32bit 1135969066 M * Roey what shoud I put in there? 1135969074 M * Roey this seems immature btw :) 1135969089 M * Roey this is not done automatically when the clients are generated?? 1135969107 M * Bertl no, because the tool can not possibly know about your intentions 1135969136 M * Bertl you might as well plan to isntall x86_64 stuff or have a 32bit env on x64_64 guests 1135969149 M * Bertl *x86_64 that is ... 1135969214 M * Roey I have an x86_64 host and want x86_64 guests 1135969216 M * Roey if that makes sense. 1135969228 M * Roey rightnow I can't seem to be getting x86_64 deb lines working. 1135969234 M * Roey history | grep /etc 1135969235 M * Roey er 1135969242 M * Bertl ah, well, then you should install a 64bit guest 1135969250 M * Roey how? 1135969252 M * Roey I thoguht I did that. 1135969260 M * Bertl IIRC, you did install a 32bit one, no? maybe I'm just confused 1135969261 M * Roey or that it copies the host's environment by deault. 1135969270 M * Roey I don't know. I did whatever the default text says to do. 1135969275 M * Bertl what command did you use to create the guest? 1135969370 M * Roey for the firstone, 1135969372 M * Roey vserver. 1135969381 M * Roey for the second & third, I just copied over stuff 1135969387 M * Roey and chnaged the properties in /etc/vserver/... 1135969401 M * Roey btw so I understand that one needs the followign commands on every guest: 1135969403 M * Roey sudo bash -c 'echo default > /etc/vservers/web/apps/init/mark' 1135969403 M * Roey sudo bash -c "echo i386 > /etc/vservers/web/uts/machine" 1135969403 M * Roey sudo bash -c 'echo linux_32bit > /etc/vservers/web/personality' 1135969411 M * Roey yes? 1135969442 M * Bertl that is only an _improvement_ for 32bit guests 1135969458 M * Bertl if you are aiming for 64bit guests, then you should forget that 1135969543 M * Roey ok 1135969549 M * Roey then what should I do for 64-bit guests? 1135969562 M * Roey I don't knwo which version I'm running for the guests, actually 1135969564 M * Roey how can I check 1135969565 M * Roey ? 1135969590 M * Bertl first, I asked how you did create the 'first' guest, so could you paste that here? 1135969612 M * cehteh can i undo chbind's with certain capabilities? 1135969640 M * cehteh (NET_ADMIN or so) 1135969684 M * Roey sudo vserver web build --rootdir /vservers/web/ -n web --hostname web.senet-int.com --interface bond0:192.168.5.145/24 -m debootstrap -- -d sarge 1135969794 M * Roey Bertl: like that. 1135969856 M * Bertl looks reasonable to me, let's do a tiny test here: 1135969893 M * Bertl what does 'file /vservers/bin/ls' tell you (adjust /vservers to your dir) 1135969949 M * Roey katzr@sink2:/vservers/web/etc$ file /vservers/web/bin/ls 1135969952 M * Roey doh 1135969961 M * Roey /vservers/web/bin/ls: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), for GNU/Linux 2.2.0, dynamically linked (uses shared libs), stripped 1135969996 M * cehteh http://www.pipapo.org/pipawiki/Vserver/vinit 1135970035 M * Roey cehteh: ok 1135970073 M * Bertl Roey: and 'file /bin/ls' ? 1135970135 M * Roey /vservers/web/bin/ls: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), for GNU/Linux 2.2.0, dynamically linked (uses shared libs), stripped 1135970144 M * Roey Bertl: so it is settled. 1135970152 M * Roey I put in the normal Sarge 1135970182 M * Bertl okay, so no 64bit userspace in sight :) 1135970322 J * _are_ ~are@dslb-084-056-143-087.pools.arcor-ip.net 1135970328 M * _are_ hi 1135970336 M * Bertl welcome _are_! 1135970346 M * _are_ Hi Bertl 1135970420 M * Roey Bertl: that's fine. 1135970421 M * derjohn ehlo, all 1135970423 M * _are_ Bertl: noticed yesterday the number of interfaces a vserver may have is limited and seen some posts on how to extend it and some others it might be slower. So I enhanced kernel config to make it configurable. Would it be ... 1135970423 M * _are_ possible to include this in the standard patch? 1135970425 M * Roey derjohn: HI! 1135970435 M * Roey derjohn: listen I was thinking.... 1135970444 M * Roey derjohn: I finally did settle on having a 100 MB / 1135970445 M * _are_ ofc the utils need to be patched, too, or even better get the right numbers on runtime 1135970449 M * Bertl _are_: no, because it would horribly break userspace tools 1135970451 M * derjohn Roey, you know, horses have a larger brain? 1135970454 M * derjohn Roey, :) 1135970455 M * Roey derjohn: but I suppose you know how to prepare an initrd :) 1135970470 M * Roey derjohn: ah but elephants' brains are larger than those of horses! 1135970486 M * derjohn Roey, yes . i did it manuall, too (please dont take that sentence out of conext!:)) 1135970505 M * Bertl _are_: if you extend util-vserver to use the newer kernel interfaces for networking, I'll spend some time on making it variable (without performance loss) inside the kernel 1135970509 M * _are_ Bertl: well, which userspace tools? util-vserver? only chbind uses this number 1135970557 M * _are_ what exactly is the 'newer' kernel interface for networking? I am not exactly a kernel coder and my C-knowledge is limited 1135970564 M * cehteh looooong time ago i wrote a patch that adds immutable and mutable flags to the ips one can bind 1135970574 M * derjohn Roey, you need to build an initrd by hand? 1135970584 M * cehteh that had almost no performance penalty 1135970591 M * Bertl _are_: well, you do not need to do kernel coding at all 1135970607 M * Bertl cehteh: what's that for? 1135970636 M * Bertl _are_: let me show you the 'new' interfaces (just a second) 1135970637 M * _are_ I know I don't need kernel coding, patch is adding a Kconfig and replacing the 16 by the config define 1135970700 M * Bertl forget about the number, I'm talking about the interface 1135970724 M * cehteh Bertl: chbind --ip 10.20.0.0 --immutable 255.255.0.0 --mutable 0.0.0.255 then the vserver admin gets 10.20.0.0 as default for 0.0.0.0 and can choose to bind 10.20.0.0 to 10.20.0.255 when configuring servers 1135970778 M * _are_ Bertl: sure, give me a hint what you talk about an di will take a look if it is anything within my capabilities 1135970786 M * cehteh idea was that when we get nested vservers this could even be used to constrain further means one can add bits to --immutable but not remove them 1135970840 M * Bertl _are_: http://www.13thfloor.at/vserver/d_rel26/v2.1.0/split-2.6.14.4-vs2.1.0/31_2.6.14.4_cmdef.diff.hl 1135970852 M * Bertl search for network_cmd.h 1135970853 M * cehteh besides it is really lightweight compared to other solutions .. and can be combined with other ideas to support more than one of such masks 1135970878 M * _are_ ok 1135970887 M * Roey Bertl, derjohn, cehteh: why do I get this in the clients after I issue some commands liek apt-get: http://rafb.net/paste/results/Vasc3w74.html 1135970896 M * Bertl cehteh: and that worked? 1135970902 M * Roey derjohn: well I wanted to use yaird. 1135970906 M * Roey derjohn: yaird had some issues though. 1135970914 M * cehteh Bertl: sure looooong ago 1135970916 M * Roey derjohn: so I gave up (for this machine) and will use it for the next server. 1135970922 M * Bertl cehteh: patches? 1135970939 M * cehteh 2000 or 2001 just some time after the project was introduced 1135970942 M * derjohn Bertl, I vote for a nin linear lookup (hash?) and a variable # of IP per guest !! 1135970965 M * Bertl derjohn: same rules apply as for _are_ :) 1135970971 M * cehteh Bertl: if you are interested i can do it again i think even if i find that old source they are horribly out of date 1135970983 M * derjohn Roey, does you machine boot with non modular kernel and 100 MB rootfs now? 1135970995 M * Bertl cehteh: i'd like to have a look at it, definitely 1135971024 M * derjohn Bertl, rules? The only rules I know is debian/rules :) 1135971024 M * Bertl cehteh: IMHO the 'old' patch should be quite fine (as the network system did not really change) 1135971055 M * cehteh hehe i proposed it serveral times .. but someone always argued thats not the way he wants it and i am programmer not politican who likes endless debates 1135971091 M * Bertl cehteh: well, I can't guarantee that I'll like it, but I'll definitely have a look 1135971116 M * Bertl (and for sure you will get some feedback/reasoning from me) 1135971154 M * cehteh just adding the 2 masks to the context and checking them on some socket calls .. it was really simple .. and fixing the userspace tools to handle that 2 masks 1135971159 M * Bertl derjohn: s/rules/prerequisites/ 1135971166 M * cehteh lemme see if i find it on the ML archive 1135971200 M * Roey derjohn: yeah 1135971204 M * Roey derjohn: that's how it is now. 1135971212 M * Roey derjohn: the machine boots with no initrd. 1135971220 M * Roey derjohn: to a / that is on /dev/sda3, ext3, native. 1135971234 M * Roey derjohn: and all the other mount points are on the LVM volume group 'main' 1135971263 M * derjohn Roey, I suppose you did a fresh vserver build mit debian (debootstrap) and did not login ... you have not done base-config 1135971329 M * derjohn Roey, do a apt-get install locales or dpkg-reconfigure locales I HTH :) 1135971366 M * derjohn Roey, fine that the machine can go into production and you have a nice HNY. 1135971496 M * Roey ok 1135971498 M * Roey got it. 1135971499 M * Roey EEEEEEEEK 1135971506 M * Roey why are all three clients sharing the same filesystem?!?!?!?! 1135971518 M * Roey the mountpoints are all different, how it is supposed to be. 1135971541 M * Roey but for some reason all three IPs that I assigned in interfaces/0/ip go to the same client! 1135971542 M * Roey why? 1135971575 M * _are_ Bertl: my knowledge is very limited, the code you gave me replaces/adds some new interface for network interface ading? 1135971603 M * Bertl Roey: maybe because you did some 'evil' copying of the config (which is now pointing to one guest) 1135971603 M * _are_ where is the old interface defined? 1135971612 M * derjohn Roey, on interfaces/0/ip you can only assign ONE IP 1135971641 M * derjohn Roey, copy interfaces/0/ip to interfaces/1/ip and change all necessary files ... 1135971655 M * Bertl _are_: http://www.13thfloor.at/vserver/d_rel26/v2.1.0/split-2.6.14.4-vs2.1.0/30_2.6.14.4_legacy.diff.hl 1135971694 J * G-ShocK ~wanna@85-65-16-7.barak-online.net 1135971699 M * derjohn Bertl, what the heck is 'exp 2.1.0.1 '? 1135971716 M * Roey derjohn: oh. 1135971721 M * Roey derjohn: well I have three clients, right? 1135971726 M * Bertl derjohn: experimental release 1135971729 M * G-ShocK how do I make the bird logo @ the main page of my T-bird have It's wings lift up instead of hugging the envelop? 1135971737 M * Roey derjohn: and I thought I could just have a separate interfaces/0/ip for each one 1135971743 M * Roey derjohn: is this not good? 1135971748 M * _are_ Bertl: vs_network.h there? 1135971755 M * Bertl G-ShocK: read the source :) 1135971784 M * derjohn Bertl, huh? 1135971795 M * Bertl _are_: no, legacy_net and legacy.h 1135971814 M * derjohn Roey, rule: one interface per IP. 1135971816 M * Bertl derjohn: new features, you know? 1135971834 M * derjohn Bertl, I meant you t-bird comment :) 1135971852 M * derjohn Bertl, features ... foo ... NG ? 1135971860 M * G-ShocK Bertl:It's a picture, you don't need to read the source in order to replace it 1135971870 M * _are_ Bertl: what exactly is the 'new' interface we talkk about? is this ngnet? 1135971907 M * Bertl G-ShocK: sure about that? 1135971980 M * Bertl derjohn: no, first a scheduler upgrade 1135971991 M * G-ShocK well, it can be extracted to an image 1135972025 M * G-ShocK can you please check and see what kind of logo you have in the main page of your thunderbird (assuming you have one) 1135972042 M * Bertl G-ShocK: no sorry, I have none ... 1135972046 M * G-ShocK I think it had it's wings lift at the debian version 1135972050 M * G-ShocK :/ 1135972072 M * Bertl G-ShocK: btw, how is that linux-vserver related? 1135972085 M * cehteh duh ... cant find the patch .. i have it certainly offline on some cd 1135972117 M * Bertl cehteh: did you test it (back then?) 1135972127 M * cehteh i had it in use 1135972130 M * Roey derjohn: hmm 1135972134 M * Roey derjohn: so I have two interfaces. 1135972138 M * G-ShocK the version where I saw the bird with the wings lift up was a debian 1135972149 M * Roey derjohn: and they are ifenslaved to bond0. 1135972159 M * Roey derjohn: now, I cannot assign multiple IPs to this bond0 ??? 1135972212 M * Bertl why not? 1135972234 M * yang2 Bertl: discovered any other bug in that patch or can i reboot? 1135972242 M * Roey Bertl: that's what I thought! 1135972255 M * Roey Bertl: so I do not need... three /physical/ ifaces then? 1135972258 M * Roey they can all share 1135972275 M * Roey provided that they use a different number for /etc/interface//ip ? 1135972278 M * cehteh http://www.paul.sladen.org/vserver/archives/200112/0039.html 1135972405 M * cehteh oh .. had only one mask .. well that describes it good 1135972460 M * Roey I don't get it. 1135972476 M * Bertl yang2: no, should work fine 1135972568 M * Bertl cehteh: well, I got the 'idea' I'm more interested in the patch and how to integrate that with util-vserver 1135972590 M * derjohn Roey, assign one IP -we call A- to bond 0: ip addr addr A/bla dev bond0 (or do it via etc network interfaces). Every vserver guest get lets say two IP. To we need /etc/vservers/foo/interface/0/ip (for B) and /etc/vservers/foo/interface/1/ip (for C). 1135972617 M * derjohn Roey, every guest add its ip to the bond0 itself. 1135972649 M * Roey hmm 1135972654 M * derjohn Roey, A, B and C will appear on bond0 (ip addr show) after foo was started. 1135972661 M * cehteh Bertl: that time i just added a parameter to the chbind syscall .. if you want it compatible .. is it possible to have optional parameters in syscall guess not, then add another syscall for it 1135972669 M * derjohn Roey, remeber: only one kernel to bidn them all .... 1135972683 M * Roey derjohn: so it is not enough to have a 1/ in the web client and 2/ in the dns client and 3/ in the mail client; they all need 1/ 2/ and 3/ 1135972687 M * cehteh and by default old behaviour is like the mask is 0.0.0.0 1135972708 M * derjohn Roey, what? a "1/" ?? 1135972716 M * Roey derjohn: interfaces/1/ 1135972750 M * Bertl cehteh: the syscall switch is quite flexible for that, but of course, the tools need changes 1135972761 M * Roey katzr@sink2:/etc/vservers/web$ cat /etc/vservers/web/interfaces/2/ip 1135972761 M * Roey 192.168.5.145 1135972761 M * derjohn Roey, i dont know ... if the dns client only need one IP you need only /0 in its config 1135972780 M * derjohn cat /etc/vservers/web/interfaces/0/ip == ? 1135972781 M * Roey derjohn: well each client has a /0/ and each has a different numbher in /0/ip 1135972795 M * derjohn Roey, yes, each client has a /0 1135972828 M * cehteh Bertl: yes .. i dont know how the syscall interface for vserver looks nowadays ... but i would vote for a single vserverctl() syscall similat to ioctl 1135972842 M * yang2 Bertl: excellent ! It works ! 1135972844 M * Roey derjohn: http://rafb.net/paste/results/tAna5t97.html 1135972849 M * derjohn Roey, and every /0/ip can^W must hold differnt values 1135972850 M * cehteh to add new features and migrate old ones 1135972861 M * Roey derjohn: OK so that is what I did earlier! and yet they all point to the same client!! 1135972865 M * Roey at least, when I ssh into it. 1135972893 M * Roey derjohn: now, I just changed it so that each client has a different N for /etc/vservers/.../interfaces/$N/ 1135972895 M * derjohn Roey, # cat /etc/vservers/vjohn/interfaces/dev 1135972895 M * derjohn eth0 1135972905 M * Bertl yang2: good, be careful, as the guest might now 'overlap' host services 1135972920 M * Roey katzr@sink2:/etc/vservers/web$ cat /etc/vservers/web/interfaces/2/dev 1135972920 M * Roey bond0 1135972930 M * Roey derjohn: did you see the above? 1135972937 M * Roey derjohn: they each have the same bond0. 1135972947 M * cehteh as i did it, it was just a few line patch to chbind .. some optargs and the additional parameters to the chbind syscall 1135972965 M * derjohn Roey, yes you pastebin is ok. 1135972983 M * derjohn Roey, are yoi talking about ssh problems? (ssh was mentioned above) 1135973003 M * Bertl cehteh: well, I won't extend the legacy interface, so that is only an option for the newer interfaces (the ones I pointed _are_ to) 1135973004 M * derjohn keep in mind that ssh will bind to _ALL_ IP is started on the host 1135973018 M * cehteh Bertl: exactly 1135973044 M * derjohn Roey, if ssh bind port 22 to all IPs on the host, the guest can not bind port 22 on its IP. 1135973077 M * cehteh so old interfaces set the mask to 0.0.0.0 and new interfaces start with 255.255.255.255 and let the user constrain it 1135973120 M * cehteh thats all really trivial 1135973173 M * _are_ Bertl: ok, checked these out, atm vc_set_ipv4root gets called, these get replaced by vc_net_{create,migrate,add,remove) 1135973174 M * Roey derjohn: I don't get it. 1135973175 M * derjohn Bertl, (concerning the #IP per guest problem) could you point out what the "new interface" is? Did you change the kernelapi while developing NG ? 1135973179 M * Roey cehteh: to whom are you talking? 1135973188 M * cehteh bertl 1135973198 M * Roey derjohn: when I ssh... into the different IPs, I end up logging into the same client!@ 1135973202 M * Roey cehteh: ok. 1135973243 M * derjohn Roey, see private chat 1135973300 M * Roey derjohn: http://rafb.net/paste/results/8lVkaW88.html 1135973350 M * Roey derjohn: I don't have the ip command when I use vserver enter. 1135973352 M * derjohn Roey, and inside the giesz 1135973353 M * Roey derjohn: only when I ssh into the host. 1135973356 M * derjohn guest? 1135973357 M * Roey giesz?? 1135973363 M * Roey derjohn: what I pasted 1135973365 M * Roey er 1135973366 M * derjohn Roey, lol .. coll typo, he? 1135973367 M * derjohn cool 1135973369 M * Roey only when I ssh into the guest 1135973386 M * derjohn Roey, no, thats from the host 1135973402 J * shedi ~siggi@213-140-22-77.fastres.net 1135973408 M * derjohn Roey, I want the guests output 1135973443 M * derjohn Roey, 'vserver yourguest enter' 1135973469 M * derjohn Roey, do not login directly via ssh to the guest - thats not working 1135973485 M * bubulak ls 1135973507 A * michal_ was logging directly to the guests all the time 1135973514 M * michal_ *loging in 1135973521 M * michal_ via ssh 1135973527 M * michal_ it was working ;p 1135973542 M * derjohn michal_, but you bound the hosts sshd to only one IP ... not to * :) 1135973554 M * Bertl bubulak: my_first_file.txt 1135973556 Q * G-ShocK Quit: 1135973563 M * michal_ how did you know ;) (yes you are right) 1135973652 M * derjohn michal_, after Roey resistance to think about this theme I have to point his nose onto the fact why he has to do it ... 1135973655 M * derjohn michal_, :) 1135973680 M * michal_ i see 1135973693 M * bubulak Bertl: :))) 1135973712 M * michal_ bubulak: ls: no such file or directory 1135973724 M * bubulak hooops :) 1135973729 M * michal_ you have Reconfigured Main -ReadFrom / ? ;p 1135973733 M * derjohn bubulak, read mal really fast == rm -rf / 1135973742 M * derjohn s/mal/mail/ 1135973743 M * bubulak passwd 1135973747 M * bubulak don't look 1135973748 M * bubulak :) 1135973760 M * michal_ we do not have to look, we are already there ;p 1135973771 M * bubulak lool 1135973812 M * _are_ Bertl: ok, checked these out, atm vc_set_ipv4root gets called, these get replaced by vc_net_{create,migrate,add,remove), is this correct? 1135973915 M * Bertl _are_: yes, basically 1135973942 M * Bertl _are_: so this interface does not require the legacy stuff 1135973975 M * Bertl _are_: so basically if you make util-vserver work with the legacy networking turned off, your part is done ... 1135973995 M * Roey derjohn: hi 1135974008 M * Bertl _are_: CONFIG_VSERVER_LEGACYNET (in recent kernel configs) 1135974010 M * Roey derjohn: you're right. 1135974013 M * Roey derjohn: are you here? 1135974016 M * Roey derjohn: I was away for a bit 1135974018 M * Roey now I am back. 1135974025 M * Bertl Roey: we didn't notice :) 1135974044 M * derjohn Roey, yes 1135974069 M * Roey derjohn: hi! 1135974070 M * Roey Bertl: hi! 1135974075 M * _are_ Bertl: ufff, from my side it is probably a lot easier to make the kernel export the limit via sysctl interface and the tools read this. Don't kow if i have enough knowledge to do the LEGACY-removal stuff 1135974081 M * Bertl hey Roey! how are you? 1135974081 M * Roey derjohn: ok, the guests do not have the 'ip' command. 1135974090 M * Roey derjohn: what is the packagename? lemme download it. 1135974090 M * _are_ (if it was as easy probably it would have been done by now) 1135974094 M * Roey Bertl: down on my luck 1135974112 M * derjohn Roey, apt-get install iproute 1135974115 M * Bertl _are_: okay, np with that :) 1135974119 M * derjohn Roey, but ifconfig will do ... 1135974193 M * _are_ Bertl: but as i don't want to give up (yet): it is only the chbind tool that is affected by this legacy stuff? 1135974224 M * Roey http://rafb.net/paste/results/Mxgk9586.html 1135974227 M * Roey derjohn: see that. 1135974268 M * Roey root@mail:/# ip addr 1135974268 M * Roey 2: bond0: mtu 1500 qdisc noqueue 1135974268 M * Roey link/ether 00:02:b3:48:50:2c brd ff:ff:ff:ff:ff:ff 1135974268 M * Roey inet 192.168.5.141/24 brd 192.168.5.255 scope global secondary bond0 1135974303 M * derjohn Roey, ahh ... see? you have this one ip 192.168.5.141 in your guest, right? 1135974305 M * Roey root@web:/# ip addr 1135974305 M * Roey 2: bond0: mtu 1500 qdisc noqueue 1135974305 M * Roey link/ether 00:02:b3:48:50:2c brd ff:ff:ff:ff:ff:ff 1135974305 M * Roey inet 192.168.5.145/24 brd 192.168.5.255 scope global secondary bond0 1135974323 M * Bertl _are_: yes, AFAIK, everything but chbind works quite fine with legacy net disabled 1135974340 M * derjohn Roey, and now compare with ip addr of the host ... 1135974357 M * Roey root@dns:/# ip addr 1135974357 M * Roey 2: bond0: mtu 1500 qdisc noqueue 1135974357 M * Roey link/ether 00:02:b3:48:50:2c brd ff:ff:ff:ff:ff:ff 1135974357 M * Roey inet 192.168.5.146/24 brd 192.168.5.255 scope global secondary bond0 1135974358 M * Roey there. 1135974361 M * derjohn Roey, see that the host also shows the IP 192.168.5.145 and 192.168.5.141 ? 1135974382 M * derjohn Roey, no, the host pls ! 1135974402 M * Roey http://rafb.net/paste/results/lc4bha76.html 1135974404 M * Roey derjohn: :) 1135974412 M * Roey derjohn: yeah it has like... three additional IPs bound to bond0. 1135974414 M * derjohn Roey, .. are those Ip there? 1135974417 M * Roey derjohn: yes. 1135974419 M * Roey yes they are. 1135974428 M * Roey derjohn: exaclty as you said 1135974431 M * derjohn Roey, and now .. think what happens if you start sshd on the host ... 1135974447 M * derjohn Roey, on which IPs the sshd will listen on? 1135974457 M * derjohn Roey, Remember: _host_ ! 1135974520 M * derjohn Roey, which IPs will it bind port 22 to? I want to continue the quiz! 1135974547 M * Roey the host 1135974559 M * Roey well it appears taht the host is listening on all those IPs. 1135974578 M * derjohn Roey, yes ... did you look which IPs are in /etc/ssh/sshd_config ? 1135974579 M * Roey the host will accept sshd connections on port 22 from any of those addresses. 1135974583 M * Roey hmm 1135974585 M * Roey one second. 1135974604 M * derjohn Roey, and now ... what will happen if the guest starts an sshd to (same IPs on the host too) 1135974613 M * Roey # Use these options to restrict which interfaces/protocols sshd will bind to 1135974614 M * Roey #ListenAddress :: 1135974614 M * Roey #ListenAddress 0.0.0.0 1135974622 M * Roey that is what the host's says. 1135974629 M * Roey derjohn: so... the host will mask the clients right? :) 1135974636 M * Roey derjohn: so I should make it listen on only one address. 1135974643 M * Roey derjohn: and the clients should listen on their own address too. 1135974646 M * Roey does this make sense to you? 1135974656 M * derjohn Roey, yes, you got it :) 1135974678 M * Roey ahhhh ok :) :) 1135974680 M * derjohn btw: the guest can only bind it's own ip. it doenst "see" the others. 1135974706 M * derjohn Roey, but in the host you must ...that counts for all apps that bind von * (apache , bind, etc) 1135974713 M * jso Bertl, patch-2.6.15-rc7-vs2.1.0.1.diff applied cleanly against 2.6.15-rc7-pa0 (parisc linux flavour) and on going to rebuild (just need some time my system is slow ;-) ) 1135974722 M * Roey derjohn: von?? 1135974730 M * Roey derjohn: ich versteh nicht 1135974759 M * Bertl jso: if you encounter unusual warnings, (or if your just started) apply this patch http://vserver.13thfloor.at/Experimental/delta-mm64-fix01.diff 1135974768 M * derjohn Roey, and because you posted "dns" .... dns (bind) is very special (see: wiki) 1135974774 M * Roey derjohn: is there a way I can just say to the host 'dude, don't listen on any port except ssh' 1135974797 M * Roey derjohn: and 'dude, don't listen on any IP other than ' 1135974805 M * derjohn Roey, yes, this is called ngnet and currently under development by jedimaster^W bertl 1135974825 M * derjohn Roey, verstehst du was ich meine ? :) 1135975010 M * Roey derjohn: NEIN! 1135975011 M * Roey :)O 1135975036 M * Roey derjohn: Sprach bitte sehr sehr nicht-schnell... dass meint, slowly. 1135975058 M * Roey weil ich versteh nur ein >>>bischen<<< hehehehe 1135975066 M * Roey ngnet? 1135975068 M * Roey hmm ok. 1135975084 M * Roey derjohn: so if host has apache running... and guest has apache running... the host is the one that will respond? 1135975099 M * derjohn Roey, forget about ngnet. it's only a proof of concept by now (bertl dont hit me :)) 1135975131 M * derjohn Roey, I think the one who was faster to bind port 80 to the IP will respond. 1135975149 M * Bertl derjohn: why should I hit you, when you speak the truth? 1135975184 M * derjohn Bertl, because I thought its more than a proof of concept now.... 1135975224 M * jso Bertl, no unsual warning till now, though (but ok, I will apply for the next build, this one is about 80% finished) 1135975233 M * derjohn Bertl, Roey, i wanted to prevent Roey even think about ng by now. 1135975234 M * Roey derjohn: ok good enough. 1135975238 M * Roey derjohn: :) 1135975248 M * Roey ok I need to go brush my teeth 1135975251 M * Roey sugar = no good for teeth. 1135975265 M * derjohn Roey, linux = no good for health. 1135975296 M * Bertl jso: yeah, doesn't apply for archs which do not use atomic64 1135975302 M * Roey adminning = no good for sanity :) 1135975329 M * derjohn Roey, adminspotting :) /me not turns towards Java ... enough support for today :) 1135975362 M * derjohn Roey, Now i try ro get support in #java :) 1135975422 M * jso Bertl, tbh I have no clue (will check ;-) ) 1135975468 M * jso Bertl, just know that parisc has a lot of artefact ;-) 1135975573 M * dothebart ok, congress is over. sitting in the c-base ;) 1135975594 M * dothebart finaly stable networking ;-) 1135975849 M * _are_ Bertl: is this correct: vc_set_ipv4root gets called with all IPs/BCAST adresses in one go, vc_net_add should replace this and gets called with a structure vcmd_net_addr_v0 as argument once for each aditional IP? does ... 1135975849 M * _are_ something need to be vc_net_create-ed? 1135975894 M * Bertl _are_: yes, you first create the context (basically replaces the set_ipv4root) and then _add_ ip after ip 1135975918 M * Bertl (well, you can add them in bunches of up to 4 IIRC) 1135975969 M * Bertl this also allows to add or remove IPs while the guest is running (but that's a little trickier, actually) 1135976591 M * jso Bertl, goooood news my box boot fine and well the guest vserver: 1135976605 M * _are_ ok, guess I know what generally should be done now, seems i miss some details 1135976606 M * jso # uname -a 1135976606 M * jso Linux hpalin 2.6.15-rc7-vs2.1.0.1-pa0-c110 #1 Fri Dec 30 21:49:12 CET 2005 parisc GNU/Linux 1135976623 M * jso # vserver-stat 1135976623 M * jso CTX PROC VSZ RSS userTIME sysTIME UPTIME NAME 1135976623 M * jso 0 60 116.8M 37.7M 0m32s98 1m25s89 4m26s39 root server 1135976623 M * jso 49152 14 79.5M 22.1M 0m02s33 0m03s91 2m26s10 DebSid 1135976625 M * Roey derjohn: hey! things work now! 1135976627 M * Roey hehe 1135976647 M * Roey derjohn, Bertl: so eventually vserver WILL be made so that it blocks stuff coming to IPs other than the host's? 1135976658 M * _are_ vnet id in struct nx_info, is this what i know as context id / xid? 1135976693 M * derjohn Roey, in a tired moment Bertl estimated something about "in 6 month". 1135976704 M * Roey :) 1135976705 M * Roey ok 1135976710 M * jso Bertl, and ssh the guest without any pb: $ ssh hplvsrvr 1135976710 M * jso Password: 1135976710 M * jso The programs included with the Debian GNU/Linux system are free software; 1135976710 M * jso the exact distribution terms for each program are described in the 1135976710 M * jso individual files in /usr/share/doc/*/copyright. 1135976711 M * jso Debian GNU/Linux comes with ABSOLUTELY NO WARRANTY, to the extent 1135976712 M * Bertl _are_: yes, but for the network context 1135976713 M * jso permitted by applicable law. 1135976715 M * jso Last login: Fri Dec 30 20:58:07 2005 from sid.debian.home 1135976745 M * Roey derjohn: like... say someone compromises a host... then they start installing programs that listen on other IPs and make people think that they're logging into those other IPs when in fact they are logging into the compromised host!!!!!!1 1135976748 M * Roey this is bad for security!!!!!!11 1135976787 M * derjohn Roey, if the _host_ in compromise you open like goaste .... 1135976799 M * Roey derjohn: nono 1135976803 M * derjohn Roey, no matter what's about Ip stuff 1135976803 M * Roey derjohn: assuming a guest is compromised 1135976811 M * Roey derjohn: the guest can start listening in on other IPs. 1135976847 M * derjohn Roey, ahhh ... hm ... hehe .... try to add an Ip within the guest : ip addr add x.y.z/b dev bond0 or eth0 1135976882 M * derjohn Roey, same goes for sniffing .. try to sniff within a guest (e.g. tcpdump -X -i eth0 ...) 1135976948 M * jso Bertl, congratulation ;-) (if you thought to some more test/patch to do, please mail me soete.joel@tiscali.be or I am registered to m-l ;-) 1135977018 M * Roey derjohn: to sniff you need to be able to set the guest to be promiscuous 1135977022 M * Roey don't we love promiscuity? 1135977030 M * Roey derjohn: ohhhh right right vserver blocks that :) 1135977034 M * Roey awww you 1135977036 M * Roey hehehe :) 1135977036 M * derjohn Roey, sure we do, but we are the host :) 1135977038 M * jso Bertl, I will now gona spent some time with my wife, thanks again, cyl * and Happy New Year ;-) 1135977048 M * Roey see ya jso!! 1135977049 M * Bertl jso: have a great new year! 1135977140 M * derjohn Roey, would you please do a favor to us? 1135977140 J * Aiken ~james@tooax6-007.dialup.optusnet.com.au 1135977174 Q * jso Quit: cu *, Hapy New Year ... ;^) 1135977226 M * Bertl welcome Aiken! 1135977249 M * Aiken hello 1135977350 M * Roey derjohn: sure 1135977356 M * derjohn Roey, I mean all what I just now told you in here ... could please write that down in you own word and out it on the wiki? (or mail it to me and I put it on the wiki) ? 1135977369 M * derjohn Roey, this Q appered more than once ... 1135977384 M * Roey Q? 1135977385 M * Roey quiet? 1135977386 M * Roey ok 1135977415 M * derjohn Roey, even there are some hints about that on the wiki, something with a not so technical approch would be fine | Q=Question ;= 1135977422 M * dothebart hm. has somebody here experience with the new broadcom wireless drivers? 1135977443 M * dothebart it keeps telling me, my version of the wireless extension is to old... 1135977551 M * derjohn dothebart, maybe your ersion of the wireless extension is to old :) 1135977605 M * dothebart well, i actualy downloaded the latest wireles* from http://pcmcia-cs.sourceforge.net/ftp/contrib/ and compiled it... 1135977611 M * Doener dothebart: extension or tools too old? 1135977629 M * dothebart extension... 1135977641 M * dothebart the kernel has got 19, the iwconfig 17 1135977652 M * Doener then the tools are too old 1135977675 M * Bertl (but there is compatibility left :) 1135977680 M * dothebart yes, but it was the latest available from the url above, which seems to be the original site? 1135977695 M * dothebart i'm trying to get these http://bcm43xx.berlios.de/ to run... 1135977707 M * cehteh # chcontext --xid 2 --flag private chcontext --xid 0 bash 1135977707 M * cehteh New security context is 2 1135977707 M * cehteh vcontext: vc_ctx_migrate(): Operation not permitted 1135977723 M * cehteh ... so i can not get back to context 0? 1135977733 M * Bertl cehteh: yup, quite expected 1135977741 M * cehteh .. because 0 is not really a context? 1135977753 M * Bertl that, and for security reasons 1135977764 M * cehteh uhm that sux 1135977789 M * Bertl well, you can (somewhat) easily add such a 'feature' 1135977799 M * cehteh how? 1135977806 M * dothebart Doener: where else could i get a newer version? 1135977813 M * Bertl cehteh: first, define a flag for it 1135977829 M * cehteh you mean patching the kernel? 1135977830 M * Bertl cehteh: then, have a look at the 'migrate' functions 1135977835 M * cehteh ok 1135977861 M * Bertl and once you understood how 'entering' a context works, you can also 'leave' it :) 1135977863 M * cehteh well SYS_ADMIN can be used for that .. or would you define a new flag? 1135977878 M * Bertl cehteh: I'd go for a 'context' flag 1135977885 M * cehteh ok 1135977888 M * FaUl *gaehn* 1135977902 M * Bertl cehteh: let's see what is already there ... 1135977934 M * Doener dothebart: did you by chance install the stuff with prefix /usr/local and have your distros wireless tools left in /usr ? 1135977947 M * cehteh Bertl: thats what i dislike .. since we have 2 special contexts (0 and 1) where 0 isnt even really a context you have exceptional behaviour 1135977972 M * dothebart Doener: yep, they have, i called it with full path. 1135977977 M * Bertl well, you would not be able to do context 1 stuff without that? 1135977992 M * Bertl cehteh: ^-- for you :) 1135977995 M * dothebart or does it take some libs from else what place? 1135978022 M * Bertl cehteh: http://www.13thfloor.at/vserver/d_rel26/v2.1.0/split-2.6.14.4-vs2.1.0/07_2.6.14.4_context.diff.hl (see that one for flags) 1135978039 M * dothebart hm. i'll try the sid package... 1135978104 M * cehteh you can change the design a little .. means from the kernel viewpoint all contexts are the same with some capabilities (posix or vserver) .. and ctx 1 is setup at boot time with the desired flags .. (as well as ctx0 drops some) 1135978112 M * Roey derjohn: ok... you know... I will write a step-by-step. 1135978117 M * Roey derjohn: and this is from the beginning 1135978121 M * Roey with compiling the vserver. 1135978135 M * cehteh Bertl: but i have currently no time to hack on the kernel 1135978168 M * derjohn Roey, hm, the ssh things would be enough in the frist step ... step by step 1135978170 M * Bertl cehteh: adds unnecessary overhead ... 1135978202 M * cehteh Bertl: what is the overhead exactly .. i mean it is just one and probably the smallest (least processes) context 1135978208 M * Bertl cehteh: I agree that from the 'design' POV it's probably nicer to put xid=0 in a separate context 1135978232 M * Doener dothebart: do you actually experience problems? usually a version mismatch isn't critical 1135978232 M * cehteh i didnt looked at the source since 2001 .. cant remember much 1135978236 M * Bertl cehteh: I'm not worried about the overhead for the processes 1135978253 M * Bertl cehteh: just think about an arbitrary xid=0 check 1135978256 M * dothebart well, the iwconfig call is missing information as the base station. 1135978283 M * cehteh Bertl: why that? 1135978285 M * Doener did you set essid, channel and stuff? 1135978288 M * Bertl cehteh: now it's a single comparison (i.e. task->xid == 0) 1135978290 M * dothebart that bcm shit won't work with driverloader under amd64 ;-) 1135978307 M * Bertl cehteh: now consider this check for a 'context' property 1135978315 M * dothebart nope, tried to set it by hand, and got the version warning... 1135978340 M * Bertl cehteh: and similar for the spectator context, again, every check would have to look into the context 1135978345 M * cehteh Bertl: yes .. how many times is that done .. and any of them timing critical? 1135978372 M * Bertl cehteh: it is done on almost every syscall and/or file operation 1135978372 M * Doener dothebart: well, you need to set essid etc. before the card can associate with the ap 1135978374 M * cehteh i think not .. but as saied, i didnt looked at the source 1135978379 M * Doener just ignore the version warning 1135978435 M * Bertl cehteh: but feel free to change the design, we can then do some comparisons ... 1135978446 M * cehteh some capabilities (i mean here access tokens not posix caps) can be short-circruited directly into the task struct 1135978460 M * dothebart yep. 1135978468 M * dothebart ill try with the sid package. 1135978484 M * Bertl cehteh: that would reduce the flexibility, no? you could not change them easily for a context 1135978486 M * cehteh Bertl: i dont have endless time .. unless someone would fund that 1135978499 M * Bertl cehteh: me neither :) 1135978533 M * cehteh Bertl: sure, the few syscalls which mutate context flags just need to care for the ones in the task struct too 1135978562 M * Bertl cehteh: task structs! 1135978585 M * cehteh yes 1135978609 M * stefani sid package not have vhashify ? 1135978617 M * Bertl cehteh: so, and what was it again, what that would buy you? 1135978632 M * Bertl stefani: IIRC, there is no proper libbeecrypt on sid 1135978660 M * stefani ohoh. 1135978686 M * cehteh anyways .. how expensive is a additional double dereference and mask operation on each syscall ... compared to task->xid == 0 ... syscalls are already moderately expensive by themself .. and since they usually do some more stuff i think that wont weight much .. but thats left to be prooven 1135978716 M * Bertl cehteh: depends, on an SMP system it would cause cache issues 1135978726 M * stefani Bertl: sarge no, sid yes. 1135978752 M * Bertl stefani: what package did you get/choose? 1135978766 M * stefani Bertl: mixed version. 1135978767 M * cehteh Bertl: maybe 1135978782 M * Bertl stefani: try to get the one micah did (0.30.209) 1135978794 M * cehteh well putting some capability flags into the task struct would solve that 1135978828 M * Bertl cehteh: fine, what is the gain over the current design? for the user? 1135978857 M * cehteh exactly what i am trying to do .. making the root-server useable on a desktop system 1135978889 M * Bertl IMHO it is pretty useable on desktop systems 1135978910 M * cehteh another idea ... 1135978929 M * Bertl let's hear! 1135978944 M * cehteh si::sysinit:/etc/init.d/rcS .. i think i just plug chbind's into my initab 1135978966 M * cehteh si::sysinit:/sbin/chbind --ip ... /etc/init.d/rcS 1135978968 M * cehteh and so on 1135978999 M * Bertl hmm, good, what would that buy you? 1135979036 M * cehteh the desktop starts many servers listening on 0.0.0.0 1135979042 M * cehteh (on the root) 1135979047 M * Bertl IMHO all you really want is a way to escape from a chbind jail ... 1135979071 M * Bertl (and that should be pretty trivial to do) 1135979100 M * cehteh no i want to chbind anything which is started by the init (except the vservers) to some ip 1135979110 M * cehteh so that it wont block ports for the vservers 1135979113 M * cehteh thats all 1135979132 M * Bertl hmm, why not use the v_* wrappers then? 1135979175 M * Bertl (for those services you want to be restricted, except sshd) 1135979403 M * cehteh because i want to keep my installation in sync with debian package management and dont want to care for configuration issues on updates 1135979470 M * cehteh and i have servers which are not covered by v_* wrapers .. so i just search a generalized way to push a single button to make all work fine 1135979481 M * Roey derjohn: how do you install the base system again? 1135979493 M * Roey derjohn: to get rid of the LOCALE errors? 1135979515 M * Bertl cehteh: what about a small modification of the rc script? 1135979537 M * Roey Roey, do a apt-get install locales or dpkg-reconfigure locales I HTH :) 1135979539 M * Roey derjohn: thanks again 1135979547 M * Bertl cehteh: to jail everything 'except' ssh for example? 1135979589 M * cehteh # cat /sbin/vinit 1135979590 M * cehteh #!/bin/sh 1135979590 M * cehteh /sbin/chbind --ip 10.20.20.10/16 --bcast 10.20.255.255 -- /sbin/init $@ 1135979599 M * cehteh .. trying that now 1135979631 M * cehteh chbind works in context 0 i hope :) 1135979658 M * Bertl yes 1135979691 M * cehteh reboots 1135979775 M * Roey Bertl: why do I get this: 1135979777 M * Roey katzr@web:~$ sudo jed /etc/hosts 1135979777 M * Roey sudo: unable to lookup web.senet-int.com via gethostbyname() 1135979833 M * Bertl I'd say some part of your command tries to lookup your hostname or something similar 1135979834 M * cehteh mhm 1135979876 M * cehteh does chbinf fork before exec? 1135979901 Q * tso Quit: BitchX-1.1-final -- just do it. 1135979905 M * cehteh kernel panic .. init didnt liked that 1135979939 M * Roey Bertl: it works fine for the two other guests though. 1135979939 M * Bertl I'd say bash becomes ini 1135979944 M * Roey Bertl: what could be the difference?? 1135979960 M * Bertl Roey: entries like /etc/host or /etc/resolv.conf 1135979976 M * Bertl Roey: or just the fact that web.senet-int.com doesn't exist 1135980029 M * Roey it doesn't 1135980036 M * Roey Bertl: but still.. the other clients do not complain. 1135980087 M * cehteh Bertl: ah have to exec ;:) 1135980088 M * Roey Bertl: what's the base debian installer after you debootstrap? 1135980174 M * Bertl Roey: what do you mean? dpkg-reconfigure? 1135980202 M * Roey like, I am not sure that I have fully set-up my debian guests. 1135980207 M * Roey for instance, they have no /etc/hosts 1135980220 P * meandtheshell 1135980243 M * Bertl well, if they have none and work fine, so be it .. (I really don't know how debian guests should look like) 1135980277 M * Roey ok 1135980282 M * Roey well I stuck a /etc/hosts in there. 1135980284 M * Roey and all is quiet now. 1135980285 M * Roey hmm. 1135980286 M * Roey yeah. 1135980395 M * cehteh gotcha .. works 1135981884 A * cehteh dances 1135981894 M * cehteh works all now :) 1135981933 M * Bertl excellent, so no design change required :) 1135981960 M * cehteh yeah 1 line of code :P 1135981979 M * cehteh i should rm the v_* things from init.d ;) 1135981990 M * cehteh and write the vinit tip on the wiki 1135982013 M * Bertl IIRC, the v* wrappers already have been removed from the tools 1135982038 M * Bertl and of course you should add that to the wiki 1135982190 M * FaUl Bertl: is that vhasify-feature stable btw? 1135982199 M * FaUl seems to rock anyway 1135982200 M * FaUl :-) 1135982223 M * Bertl FaUl: vhashify is very stable, it doesn't use anything new (just plain old hardlinks :) 1135982252 M * Bertl FaUl: the CoW link breaking in the devel release is working but I do not consider it 'stable' yet 1135982364 M * FaUl ok, cow should work automagically if i use vhashify, won't it? 1135982886 M * derjohn Roey, base-config and apt-get install locacles 1135982906 M * Bertl in devel, yes, but it's not strictly required for 'typical' vhashifications (which exclude /etc and /vr by default) 1135983217 M * stefani Bertl: micah's util-vserver does install and much works on a sarge system, but the vhashify stuff does not happen. 1135983237 M * stefani no fault of micah's 1135983522 P * stefani I'm Parting (the water) 1135983526 M * FaUl Bertl: i think it's very importent for securtiy reasons 1135983557 M * FaUl Bertl: it would be to bad if someone get a rootkit and it would change every single sshd on any vserver because they are linked :-) 1135983577 M * mattg is vhashify installed by default when alpha is compiled from source? 1135983581 M * cehteh how about starting a "Recipes" subpage below 'Help' on the wiki? 1135983633 M * mattg nevermind.. I just need to learn to spell right 1135983688 M * Bertl FaUl: no, that's protected by the immutable flag 1135983726 M * Bertl cehteh: why not ... 1135983750 M * Bertl okay, I'm off for today ... have fun! 1135983759 N * Bertl Bertl_zZ 1135983827 M * mattg how about "Can not find file for 'RPMSTATEDIR'" on vhashify 1135983996 M * FaUl Bertl_zZ: so i can't change anything which is hashifyed on any release not 2.1.0? 1135984345 J * are|lunch ~are@dslb-084-056-135-171.pools.arcor-ip.net 1135984560 Q * michal_ Ping timeout: 480 seconds 1135984606 M * cehteh http://linux-vserver.org/Recipes 1135984649 J * michal_ ~michal@mprivacy-update.de 1135984660 M * mattg ok.. fixed RPMSTATEDIR.. /vservers/.pkg/[server] was in the wrong place 1135984750 Q * _are_ Ping timeout: 480 seconds 1135985067 M * derjohn cehteh, running the root server with chbind'ed ip address -> this works? so the guest IPs are not visible to the host then? 1135985233 J * comfrey ~comfrey@cpe-69-203-195-46.nyc.res.rr.com 1135985360 M * cehteh derjohn: they are .. but anything you start on the root which tries to bind 0.0.0.0 gets the configured address 1135985378 M * cehteh means you dont have to reconfigure servers in the root 1135985755 Q * comfrey Ping timeout: 480 seconds 1135985949 M * derjohn cehteh, thats excellent, why didnt anyone point that out till now? 1135985972 M * cehteh because i am smart :P 1135985990 M * derjohn cehteh, drawbacks? 127.0.0.1 works or do I have to bind it explicitly, too? 1135986004 M * cehteh didnt checked yet 1135986011 M * derjohn cehteh, is disabled smart on my machine :) 1135986429 J * comfrey ~comfrey@cpe-69-203-195-46.nyc.res.rr.com 1135986829 M * cehteh so when i ssh 127.0.0.1 from guest .. i connect to the guest .. seems to be ok or? 1135986935 M * derjohn cehteh, eh, no. in guest 127.0.0.1 is mapped to the first IP of the host, i.e. the realserver. So you shouldnt see the guest .... 1135986936 M * cehteh on old vservers i just put 'external-ip localhost' into /etc/hosts ... worked for some clients 1135986966 M * cehteh huh 1135986991 M * cehteh if i connect to 127.0.0.1 i expect to connect to myself not to the root server 1135987033 M * cehteh i can tell, it works at least as i expect 1135987046 M * derjohn cehteh, sure? I remeber the 127.0.0.2 trick in guests for simulating a "lookback" 1135987084 M * derjohn cehteh, so I think 127.0.0.1 in a guest is mapped to root server (unitl we have ngnet) 1135987120 M * derjohn so, is 127.0.0.1 there if you do a ip addr on host? 1135987135 Q * comfrey Ping timeout: 480 seconds 1135987177 M * cehteh derjohn: i hope that 127.0.0.2 thing was only a old workaround