1150934740 J * shedi ~siggi@130.208.221.254 1150934898 Q * sid3windr Ping timeout: 480 seconds 1150935258 J * sid3windr luser@bastard-operator.from-hell.be 1150936403 J * mef ~mef@64.129.10.253 1150936466 Q * mef Remote host closed the connection 1150937669 Q * starlein Quit: changing servers 1150937772 J * starlein ~star@fo0bar.de 1150938465 Q * meandtheshell Quit: bye bye ... 1150941428 J * DarthVader ~Aniken@203.177.212.163 1150941835 Q * s0undt3ch Ping timeout: 480 seconds 1150941972 J * Brio ~Brio@pool-70-16-99-242.port.east.verizon.net 1150941985 M * Brio is this support channel? 1150942006 P * Brio 1150943761 J * s0undt3ch ~s0undt3ch@bl7-241-58.dsl.telepac.pt 1150944236 Q * FireEgl Ping timeout: 480 seconds 1150944294 Q * DarthVader Quit: Leaving 1150945125 Q * yelifu Quit: 1150945652 Q * Nam Ping timeout: 480 seconds 1150945767 Q * shuri Remote host closed the connection 1150945774 J * FireEgl Atlantica@Atlantica.US 1150946176 J * Nam ~nam@70.78.64.62 1150946662 Q * Nam Ping timeout: 480 seconds 1150946869 J * Aiken_ ~james@tooax6-096.dialup.optusnet.com.au 1150947163 J * Nam ~nam@70.78.64.62 1150947185 Q * Aiken Ping timeout: 480 seconds 1150948512 Q * Nam Ping timeout: 480 seconds 1150948690 Q * lilalinux Remote host closed the connection 1150948705 J * Nam ~nam@70.78.64.62 1150951493 J * _coocoon_ ~coocoon@84.160.105.178 1150951510 M * _coocoon_ morning 1150951577 Q * Nam Ping timeout: 480 seconds 1150951688 J * Nam ~nam@70.78.64.62 1150952252 Q * Nam Ping timeout: 480 seconds 1150952269 J * shuri ~shuri@64.235.209.226 1150952546 J * f_ ~f_@83-215-237-1.seek.stat.salzburg-online.at 1150952724 Q * stormy Ping timeout: 480 seconds 1150952735 M * _coocoon_ hollow r u there 1150952983 M * Hollow _coocoon_: yup 1150953128 M * _coocoon_ hollow do u have a document history of this side http://www.gentoo.org/doc/en/vserver-howto.xml#doc_chap4 1150953204 M * Hollow http://sources.gentoo.org/viewcvs.py/gentoo/xml/htdocs/doc/en/vserver-howto.xml?rev=1.7&view=log 1150953244 J * Nam ~nam@70.78.64.62 1150953252 M * _coocoon_ hollow: cool thanx 1150953333 M * _coocoon_ hollow: also wanted to know if shall use for my gentoo guest the make.conf from the host 1150953503 M * Hollow yeah, you can do this 1150954402 Q * Nam Ping timeout: 480 seconds 1150954649 N * otaku42_away otaku42 1150954988 J * Nam ~nam@70.78.64.62 1150955782 Q * Nam Ping timeout: 480 seconds 1150956169 Q * f_ Quit: This computer has gone to sleep 1150956232 Q * cdrx Ping timeout: 480 seconds 1150956241 J * f_ ~f_@83.215.237.1 1150956601 Q * Piet Quit: :tiuQ 1150956681 J * schimmi ~sts@212.202.73.176 1150956975 Q * s0undt3ch Remote host closed the connection 1150956977 J * s0undt3ch ~s0undt3ch@bl7-241-58.dsl.telepac.pt 1150959373 J * Nam ~nam@70.78.64.62 1150959393 J * meandtheshell ~markus@85-124-207-40.dynamic.xdsl-line.inode.at 1150959657 Q * schimmi Ping timeout: 480 seconds 1150960294 J * cdrx ~legoater@cimai.net4.nerim.net 1150961093 J * ||Cobra|| ~cob@pc-csa01.science.uva.nl 1150961149 J * schimmi ~sts@aquila.tcs.ifi.lmu.de 1150961564 J * dna ~naucki@dialer-181-216.kielnet.net 1150961917 Q * Nam Ping timeout: 480 seconds 1150962440 M * waldi oh, oh 1150962448 M * waldi util-vserver don't have a timeout 1150962457 M * waldi for starting marked domains 1150962962 J * Nam ~nam@70.78.64.62 1150963654 J * ecklock ~ecklock@219.148.74.66 1150963677 P * ecklock 1150963734 Q * Hollow Remote host closed the connection 1150964027 J * Hollow ~hollow@cl-52.muc-02.de.sixxs.net 1150964043 Q * eyck_ Quit: leaving 1150964127 J * DarthVader ~Aniken@203.177.212.163 1150964347 J * eyck_ eyck@ghost.anime.pl 1150964422 Q * sars Ping timeout: 480 seconds 1150964607 J * Arcticfox ~arcticfox@83.243.170.124 1150964694 M * Arcticfox is there any situation where "vserver server stop" would shutdown the host interfaces ... 1150965005 M * Loki|muh yes 1150965029 M * Loki|muh when capabilities are not in the kernel 1150965048 M * Arcticfox you mean the host kernel.. 1150965060 M * Loki|muh sure, there is only one kernel in vserver 1150965082 M * Arcticfox craps of crap .. sorry .. been up 26hrs .. 1150965235 M * Arcticfox ok thanks .. for the help.. 1150965248 M * Loki|muh np 1150965711 Q * Arcticfox Quit: . 1150965934 Q * _coocoon_ Quit: KVIrc 3.2.0 'Realia' 1150965973 J * _coocoon_ ~coocoon@84.160.105.178 1150966388 Q * _coocoon_ Quit: KVIrc 3.2.0 'Realia' 1150966876 J * cskarby ~cs@195.1.31.69 1150967164 P * glen 1150967262 Q * sladen Ping timeout: 480 seconds 1150967293 J * _coocoon_ ~coocoon@84.160.105.178 1150967535 J * sladen paul@starsky.19inch.net 1150968271 N * _coocoon_ coocoon 1150968790 J * sars ~sarnold@198.145.28.33 1150969149 J * Milf ~Miranda@141.12.9.205 1150969566 P * coocoon So Long, and Thanks for All the Fish! 1150969622 J * _coocoon_ ~coocoon@84.160.105.178 1150969652 N * _coocoon_ coocoon 1150969893 J * Arcticfox ~arcticfox@83.243.170.124 1150970061 M * Arcticfox hum the problem i presented earlier still persists 1150970463 Q * Aiken_ Quit: Leaving 1150970513 M * Arcticfox even with capabilities 1150971524 Q * shedi Quit: Leaving 1150973247 Q * Nam Ping timeout: 480 seconds 1150974319 Q * phedny Quit: Reconnecting 1150974323 J * phedny ~mark@volcano.p-bierman.nl 1150974842 Q * schimmi Ping timeout: 480 seconds 1150975891 J * Nam ~nam@70.78.64.62 1150976987 Q * Nam Ping timeout: 480 seconds 1150977191 J * schimmi ~sts@138.246.7.82 1150977207 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1150977216 N * Bertl_oO Bertl 1150977219 M * Bertl morning folks 1150977239 M * Bertl Arcticfox: host interface shutting down (is the problem which persists?) 1150977244 M * Arcticfox yes 1150977254 M * Arcticfox i really did the RTFM overnight 1150977255 J * _coocoon_ ~coocoon@84.160.105.178 1150977299 N * _coocoon_ coocoon 1150977314 M * Bertl Arcticfox: hmm, I'd assume you assigned the host ip to a guest (including the 'dev' part) can you proove me wrong? 1150977324 M * Arcticfox no 1150977330 M * coocoon hello bertl 1150977337 M * Arcticfox it uses its own ip 1150977360 M * Bertl and it doesn't have any capabilities like NET_ADMIN or such? 1150977371 M * Bertl hey coocoon! 1150977395 M * Arcticfox Bertl: oh ! do that does affect .. 1150977395 J * Nam ~nam@70.78.64.62 1150977406 M * Arcticfox i thought it would only affect the guest not the host 1150977436 M * Bertl well, it does only affect the guest, it allows to modify host network stuff :) 1150977481 M * Arcticfox yeah i had that one in bcapabilities earlier 1150977492 M * Arcticfox im running slack with latest 2.6 1150977494 M * Arcticfox btw 1150977564 M * doener if you, for whatever reason, need NET_ADMIN, make at least sure that your guest's initscript don't touch anything important ;) that includes removing "-i" from the halt and reboot scripts 1150977596 M * Arcticfox ok ill keep that in mind 1150977612 M * doener it's a good idea to remove that anyway, as it lowers the noise level on "vserver foo stop" 1150977680 M * Arcticfox yeah 1150977723 M * Bertl okay, having troubles with my network connectivity (again) will hopefully be fixed tomorrow ... will try to come back later ... 1150977737 N * Bertl Bertl_oO 1150978488 J * lilalinux ~plasma@dslb-084-058-211-143.pools.arcor-ip.net 1150981531 J * cattivik ~andrea@service.cab.unipd.it 1150982555 M * cattivik hi all 1150982967 M * doener hi cattivik 1150983116 J * mef ~mef@64.129.10.253 1150983125 Q * mef Remote host closed the connection 1150983128 M * cattivik maybe a silly question, made hundred of times (..) but I can't understand where I have to put the dev, ip, name and prefix files: in /etc/vservers//interfaces or in /etc/vservers//interfaces/0/ ... ? 1150983243 M * coocoon cattiv: maybe here u will find all answers http://deb.riseup.net/vserver/usage/ 1150983273 M * cattivik if I'm not wrong, www.13thfloor.at/vserver/s_rel26/v2.01/index.html and the flowers page do not agree about it... :] 1150983279 M * cattivik coocoon: ops, thanx... 1150983521 M * coocoon uw 1150984254 J * FaUl immo@shell.chaostreff-dortmund.de 1150984258 M * FaUl hiho 1150986625 Q * DarthVader Quit: Leaving 1150988061 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1150988106 J * coocoon ~coocoon@84.160.105.178 1150988114 Q * ||Cobra|| Remote host closed the connection 1150988117 Q * gerrit Quit: Client exiting 1150988119 Q * Milf Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org 1150988567 J * doener_ ~doener@i5387E772.versanet.de 1150988567 Q * doener Read error: Connection reset by peer 1150988688 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1150988736 J * coocoon ~coocoon@84.160.105.178 1150988882 Q * cdrx Ping timeout: 480 seconds 1150989420 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1150989437 J * Viper0482 ~Viper0482@p54977BE4.dip.t-dialin.net 1150990770 J * stefani ~stefani@tsipoor.banerian.org 1150990930 Q * Arcticfox Quit: connection reset by IRS 1150990951 P * cskarby 1150990958 M * shuri hi 1150991338 N * otaku42 otaku42_away 1150991632 Q * schimmi Ping timeout: 480 seconds 1150991669 J * coocoon ~coocoon@84.160.105.178 1150992259 Q * cryo Quit: Покидаю 1150992315 M * shuri anybody can help me with the qota features inside a vserver 1150992327 M * shuri i have setup de /dev/vroot/0 1150992443 M * shuri cp the device into the vserver 1150992449 M * shuri vrsetup the device 1150992470 M * shuri add the quota_clt intto tune bcaps of the verver 1150992489 M * shuri quotaon -a 1150992490 M * shuri quotaon: using //quota.group on /dev/hdv1 [/]: No such device 1150992491 M * shuri :( 1150992699 J * kir_home ~kir@swsoft-mipt-nat.sw.ru 1150992700 J * kir__ ~kir@swsoft-mipt-nat.sw.ru 1150992705 M * doener_ shuri: did you copy the device node as hdv1 into the vserver's /dev? 1150992718 M * shuri yes 1150992733 M * shuri cp /dev/vroot/0 /vserver/name/dev/hdv1 1150992734 M * doener_ what did your vrsetup command look like? 1150992761 M * shuri vrsetup /dev/vroot/0 /dev/sdb1 1150992855 M * doener_ hmm.. 1150992889 M * doener_ http://software.newsforge.com/article.pl?sid=06/06/15/1444210&from=rss 1150992891 M * doener_ wow! 1150992898 M * shuri brw-r--r-- 1 root root 4, 0 Jun 21 15:02 /dev/hdv1 1150992902 M * shuri inside de vserver 1150992923 M * shuri nice :) 1150993005 M * shuri so 1150993006 M * shuri quotacheck -maugv 1150993010 M * shuri work 1150993018 M * shuri quotacheck: Scanning /dev/hdv1 [/] done 1150993046 M * shuri quotaon no 1150993193 M * doener_ what's in your /etc/mtab? 1150993213 M * shuri . /dev/hdv1 / ufs rw,usrquota,grpquota 0 0 1150993303 M * cattivik grr... I still get "Operation not permitted" while pinging outside from withing the guest... any idea? :| The /etc/vservers//interfaces/0/* files seems to be OK... 1150993337 M * doener_ not permitted? sounds like you're missing a context capability... 1150993364 M * cattivik uhm... i did vprocunhide: does it matter? 1150993379 M * doener_ no, that's unrelated 1150993394 M * doener_ cat /proc/virtual//status | grep CCaps 1150993449 M * cattivik I get: 0000000000000101 1150993467 M * doener_ hm, that's fine... 1150993573 M * doener_ shuri: hm, did you add quota_ctl to bcaps or ccaps? 1150993598 M * shuri ccapabilities 1150993603 M * shuri quota_ctl 1150993619 M * doener_ ok, just wondered because you wrote bcaps above 1150993620 M * shuri vservers/vs5/ccapabilities 1150993665 M * shuri do i need someting special in the fstab of the host? 1150993708 M * shuri dev/sdb1 /vservers ext3 defaults,tagxid,usrquota,grpquota 0 0 1150993711 M * shuri i got this 1150993715 M * shuri on the host 1150993732 J * schimmi ~sts@port-212-202-73-176.dynamic.qsc.de 1150993766 M * cattivik doener_: is it ok if `route' from inside the guest does show eth0 as the interface? I have added the file /etc/vservers/vserver-name/interfaces/iface but the guest's interface still doesn't show in ifconfi (host side), dispite of what the Flowers's Page says... 1150993813 M * shuri cattivik, 1150993813 J * mkhl ~mkhl@200-153-153-17.dsl.telesp.net.br 1150993815 M * doener_ iface? 1150993828 M * cattivik doener_: sorry, I meant "name" 1150993851 M * doener_ so /etc/vservers/vserver-name/interfaces/0/name ? 1150993856 M * cattivik doener_: yes! 1150993879 M * doener_ and you also have: ip and dev files? 1150993885 M * doener_ s/:// 1150993887 M * shuri :) 1150993890 M * cattivik doener_: yes! 1150993898 M * doener_ excellent! ;) 1150993935 M * shuri what do you have inside dev 1150993937 M * doener_ and finally, you did (re)start the vserver after these changes were made? 1150993965 M * cattivik doener_: yes, of course... :] 1150993982 M * cattivik shuri: eth0 1150993984 M * doener_ ok, what are the contents of "name"? 1150993995 M * cattivik doener_: 0garg 1150994099 M * doener_ does "ip a" show the ip address? if so, what does the line look like? 1150994187 M * cattivik doener_: inet 147.162.213.19/24 brd 147.162.213.255 scope global secondary eth0 1150994195 M * shuri :::) 1150994204 M * shuri we got it 1150994227 M * shuri cattivik, ip must contain only the ip 1150994231 M * shuri 147.162.213.19 1150994251 M * cattivik shuri: the file ip? Yes, it does! 1150994253 M * doener_ shuri: hm? that's "ip a" output, and always looks quite like that 1150994281 M * shuri ok i think the ip file contain 147.162.213.19/24 brd 147.162.213.255 scope global secondary eth0 1150994294 M * cattivik shuri: no, no... 1150994298 M * shuri ok good 1150994299 M * shuri sorry 1150994309 M * cattivik shuri: :)) 1150994405 M * cattivik I also put broadcast and mask files but nothing changes, I only got a "...brd 147.162.213.255..." instead of "...brd 147.162.213.0..." output from `ip a' but still not networking from the guest... 1150994479 M * shuri hey doener_ look aqt this 1150994480 M * cattivik I also enabled debugging via /proc/sys/vserver but I can't read anything understandable from the host logs... 1150994483 M * shuri Module Size Used by 1150994483 M * shuri vroot 4872 0 1150994489 M * shuri Used 0 1150994494 M * shuri is it normal? 1150994556 M * doener_ cattivik: could you start the vserver using this line? vserver --debug name start 2> logfile 1150994576 M * cattivik doener_: i try... 1150994642 M * cattivik doener_: btw: starting and stopping is really slow... It doesn't seem very good to me... :\ 1150994646 M * doener_ then, please make the contents of "logfile" available somehow (eg. http://paste.linux-vserver.org) 1150994698 M * cattivik doener_: ...and I always get "RTNETLINK answers: Cannot assign requested address" ... 1150994702 M * cattivik doener_: ok... 1150994791 M * cattivik doener_: done! 1150994817 M * doener_ url? 1150994835 M * cattivik doener_: http://paste.linux-vserver.org :) 1150994847 M * coocoon cattivik: http://linux-vserver.org/Linux-Vserver+FAQ --> I. Issues When I stop a vserver its network alias does not go away, I am getting 'RTNETLINK answers: Cannot assign requested address' 1150994868 M * coocoon cattivik: 10. 1150994876 M * cattivik coocoon: I read it... 1150994881 M * coocoon ok 1150994883 M * doener_ cattivik: http://paste.linux-vserver.org/114 <-- that's the url I meant ;) 1150994908 M * cattivik doener_: ops, 1150994911 A * cattivik stupid 1150994915 M * doener_ but it looks a little short... mine is about 900 lines... 1150994945 M * doener_ even when I forget to stop the vserver it's about 160 lines 1150994999 M * cattivik coocoon: I ended up thinking that networking should work anyway from the inside, dispite of the message... 1150995018 M * cattivik doener_: I see... 1150995084 M * cattivik doener_: I apologize, I just pasted the last 10 lines... :P 1150995090 M * cattivik doener_: I will correct now... 1150995102 M * doener_ heh :) 1150995105 M * doener_ np 1150995362 M * cattivik doener_: it is 906 lines: http://paste.linux-vserver.org/115 1150995662 M * doener_ cattivik: did you add the "name" file while the vserver was running? 1150995708 M * cattivik doener_: Uhm, it could be <*)... But then I did restart it... ehm... 1150995723 M * doener_ well, that causes problems nevertheless... 1150995752 M * phedny hmm, this barrier flag must be set on the dir containing the vserver root dirs? 1150995760 M * doener_ ip a d 147.162.213.19/24 dev eth0 1150995761 M * phedny that-is: not on the vserver root dirs themselves 1150995777 M * doener_ ip a a 147.162.213.19/24 broadcast 147.162.213.255 label eth0:0garg dev eth0 1150995794 J * bonbons ~bonbons@83.222.39.166 1150995799 M * doener_ use these two commands to fix the situation 1150995808 M * phedny bonbons: nice ipv6 patch :) 1150995844 M * bonbons phedny: thanks :) 1150995909 M * phedny I only need to change the EXTRAVERSION in Makefile so ipv6 is in lowercase, since Ubuntu (or maybe even Debian) make-kpkg tool refused to compile it otherwise :) 1150995937 M * cattivik doener_: great, now the interface shows up both in the host and in the guest... Thanks!:) but still not networking... 1150996102 J * pisco ~pampel@80.135.133.188 1150996116 M * doener_ cattivik: which kernel version are you using? 1150996155 M * cattivik doener_: 2.6.14.3-vs2.0.1 :] 1150996192 M * doener_ old, but networking should work nevertheless... 1150996228 M * doener_ sorry, I'm out of ideas... guess daniel_hozac will pick up the topic when he returns... ;) 1150996239 M * shuri lol 1150996255 M * shuri doener_, you gave up with me too? 1150996290 M * cattivik doener_: I am really grateful, you've already spent a lot with me, thanks...! 1150996346 M * doener_ shuri: i have little experience with the vroot stuff... maybe you could try the following google search? site:irc.13thfloor.at quotaon "no such device" 1150996360 Q * kir_home Ping timeout: 480 seconds 1150996362 Q * kir__ Ping timeout: 480 seconds 1150996408 P * pisco 1150996430 M * shuri doener_, already done 1150996439 M * shuri and did no find any answer 1150996441 M * shuri hehe 1150996471 M * doener_ hm, too bad... 1150996531 M * doener_ shuri: the partition was mounted with {usr,grp}quota on the host, right? 1150996559 M * shuri vservers ext3 defaults,tagxid,usrquota,grpquota 0 0 1150996669 M * shuri do i need quota module load on the host ? 1150996715 M * doener_ no idea... 1150996754 M * coocoon shuri: yesterday i have read i do not know it exactly but on the host it also must be on 1150996791 M * coocoon ;-) 1150996804 M * daniel_hozac shuri: you did run vrsetup since rebooting the host, right? 1150996828 M * shuri i have not reboot the host 1150996846 M * shuri if i use vroot again it says device busy 1150996847 M * daniel_hozac i just meant that you've run it during this boot. 1150996853 M * shuri beacuse is already attached 1150996857 M * shuri yes 1150996859 M * daniel_hozac ok, that's fine then. 1150996863 M * bonbons phedny: that Ubuntu tool is really picky! :) 1150996923 M * shuri coocoon, Module Size Used by 1150996924 M * shuri quota_v2 10752 0 1150996924 M * shuri vroot 4872 0 1150996943 M * daniel_hozac bonbons: btw, q'n'd patch for IPv6 addresses in .../interfaces/*/ip: http://daniel.hozac.com/vserver/util-vserver-0.30.210-ipv6.patch 1150996987 M * daniel_hozac cattivik: what do you mean by no networking? 1150997000 M * bonbons daniel_hozac: ok, will look at it after meal or tomorror as I get time 1150997016 M * cattivik daniel_hozac: reaching the outside from the guest.. :) 1150997028 M * daniel_hozac cattivik: what's your networking setup like? 1150997068 M * cattivik daniel_hozac: uhm, wait, let me look at the firewall configuration .... ! 1150997120 M * coocoon shuri: sorry but i have no idea, maybe daniel_hozac can help u 1150997148 M * cattivik doener_: yes! 1150997175 M * doener_ heh, ok, right... let's get talk back on the channel... ;) 1150997320 M * cattivik doener_: :)) 1150997338 J * gerrit ~gerrit@67.160.146.170 1150997629 N * sars sarnold 1150997989 M * cattivik daniel_hozac: uhm, the firewall seems to be ok... 1150998274 M * cattivik PORK... 1150998276 M * cattivik ehm... 1150998365 M * cattivik somebody installed ipmasq on the host... WHO DID IT?? 1150998392 Q * f_ Quit: This computer has gone to sleep 1150998400 M * cattivik ...now it works :ц╬ 1150998425 M * cattivik thanks to all of you... 1150998705 M * cattivik bye bye, see you 1150998709 Q * cattivik Quit: Client exiting 1150999391 J * Brio ~Brio@pool-70-16-99-242.port.east.verizon.net 1150999396 M * Brio #MFGGIRC 1150999405 P * Brio 1150999543 J * f_ ~f_@83-215-237-1.seek.stat.salzburg-online.at 1151000449 J * nammie ~nam@70.78.64.62 1151000698 Q * bonbons Quit: Leaving 1151000762 Q * Nam Ping timeout: 480 seconds 1151001223 Q * sladen Ping timeout: 480 seconds 1151001478 J * sladen paul@starsky.19inch.net 1151002641 Q * baggins Ping timeout: 480 seconds 1151002837 Q * nammie Ping timeout: 480 seconds 1151002881 Q * lilalinux Remote host closed the connection 1151003632 J * baggins baggins@kenny.mimuw.edu.pl 1151005640 M * phedny how can I see what runlevel my guest is in? 1151005649 M * phedny ps ax | grep init --> show level 2 1151005656 M * phedny while the runlevel command returns 3 1151005698 M * daniel_hozac unless you have a real init inside your guest, the ps command will return your host's runlevel. 1151005700 Q * brc_ Quit: No windows for this server 1151005710 M * phedny okay, so my guest is in 3 1151005715 M * phedny and indeed it's true my host is in 2 1151005895 M * phedny by the way, is the concept of runlevel something the kernel knows about or is it purely userland and init-related? 1151005912 M * daniel_hozac userland. 1151005924 M * phedny so asking the runlevel of a guest is nonsense? 1151005929 M * phedny no, sorry 1151005932 M * phedny I need sleep :) 1151006119 M * cehteh runlevels are sysv - init concepts there are other concepts with other init-systems which dont use runlevels 1151006184 M * phedny okay, then I wonder where the runlevel command gets it's value from 1151006188 M * cehteh some even throw some kernel capabilities away for better security .. that could be possible in linux too but afaik its not used in most distris 1151006216 M * cehteh the init scripts have a $RUNLEVEL variable 1151006228 M * phedny ahh, from /var/run/utmp :) 1151006278 M * phedny and how does the runlevel 3 got in that file? 1151006286 M * phedny since default is 2 in inittab 1151006296 M * daniel_hozac the utils default is 3. 1151006299 Q * mkhl Quit: 1151006317 M * phedny the inittab file isn't used in guests? 1151006327 M * phedny or at least the default runlevel 1151006327 M * doener_ only for "plain" initstyle 1151006339 M * doener_ otherwise /etc/init.d/rc is called directly 1151006373 M * phedny okay :) 1151006477 M * cehteh runlevels have no much use in vservers i think 1151006496 M * cehteh mostly only present for compatibility ;) 1151006766 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1151006846 Q * f_ Quit: This computer has gone to sleep 1151007115 J * kir_home tis-65d01a@213.152.157.70 1151007149 J * kir__ tis-99f63a@213.152.157.70 1151007384 J * cdrx ~legoater@cap31-3-82-227-199-249.fbx.proxad.net 1151007814 Q * Viper0482 Remote host closed the connection 1151008178 J * f_ ~f_@83-215-237-1.seek.stat.salzburg-online.at 1151008334 J * piwosz_ ~piwosz@cup234.internetdsl.tpnet.pl 1151008522 P * piwosz_ Leaving 1151009700 J * coocoon ~coocoon@84.160.105.178 1151010485 J * Aiken ~james@tooax6-210.dialup.optusnet.com.au 1151011086 Q * f_ Quit: This computer has gone to sleep 1151011182 Q * Greek0 Read error: Connection reset by peer 1151011198 J * Greek0 ~greek0@85.255.145.201 1151011340 N * insomnia1 insomniac 1151011697 J * f_ ~f_@83-215-237-1.seek.stat.salzburg-online.at 1151011732 Q * meandtheshell Quit: bye bye ... 1151011837 Q * f_ Quit: 1151012013 J * shedi ~siggi@130.208.221.254 1151012320 M * harry anyone here? 1151012358 M * coocoon yes 1151012362 M * harry nice 1151012365 A * harry big prob here 1151012371 M * harry well... "big" 1151012381 M * harry i've got a vm with 2 interfaces 1151012387 Q * kir_home Ping timeout: 480 seconds 1151012388 Q * kir__ Ping timeout: 480 seconds 1151012394 M * harry 1: 134.58.241.32 1151012399 M * harry 2: 192.168.28.21 1151012411 M * harry now, i've set up source routing on the host 1151012424 M * mnemoc .oO( no greeting? )o 1151012436 M * harry so... all traffix is routet... to perfection :p 1151012440 M * harry routed 1151012445 M * harry and: traffic 1151012462 M * harry anyway... problem is... externam traffic goes to 192.x.x.x network 1151012466 Q * dna Quit: Verlassend 1151012466 M * harry and i don't want that! 1151012474 M * harry default traffic has to go over 134 address 1151012518 M * harry on the host, i got eth0 on webserver vlan, eth1 on management vlan, eth2 on mail vlan, eth3 not configged yet 1151012535 M * harry how can i set priorities on these virtual interfaces 1151013382 M * harry /etc/vservers//interfaces/dev setting that to eth2 doesn't solve the problem 1151013407 M * harry i thought, maybe it's the virtual name it needs... so putting 0 in there (134 address interface) doesnt work either 1151013461 M * waldi did you assign the ips to the correct interface? 1151013467 M * harry du'uh 1151013494 M * harry gandalf:/usr/local/etc/vservers/kromsin/interfaces# cat 0/* 1151013494 M * harry eth2 1151013494 M * harry 134.58.241.32 1151013589 M * waldi wait, linux don't use the interface to define where the route goes 1151013606 M * waldi it uses the source interface specified in the route 1151013622 M * harry output van ip route ls table 134.58.241-net: 1151013622 M * harry ------------------------------------------- 1151013622 M * harry 134.58.241.0/24 dev eth2 scope link 1151013622 M * harry default via 134.58.241.254 dev eth2 1151013631 M * harry output van ip route ls table neutral-net: 1151013631 M * harry ----------------------------------------- 1151013631 M * harry 192.168.28.0/23 dev eth1 scope link 1151013631 M * harry default via 192.168.29.254 dev eth1 1151013645 M * harry so... depending on the source address, i have different routes 1151013651 M * harry different default gateways 1151013689 M * harry advanced routing fun ;) 1151013707 M * harry but... it seems he prefers to take the 192.168 address for external traffic 1151013711 M * harry and i don't want that 1151014189 M * doener_ and "ip rule" output? 1151014224 M * harry 0: from all lookup local 1151014224 M * harry 1000: from 134.58.10.0/26 lookup 134.58.10-net 1151014224 M * harry 1001: from 10.33.113.0/24 lookup 10.33.113-net 1151014224 M * harry 1002: from 192.168.28.0/23 lookup neutral-net 1151014224 M * harry 1003: from all to 192.168.30.0/24 lookup internal-net 1151014227 M * harry 1004: from 134.58.241.0/24 lookup 134.58.241-net 1151014229 M * harry 32766: from all lookup main 1151014232 M * harry 32767: from all lookup default 1151014250 M * doener_ and the traffic you are concerned about comes from where? 1151014256 M * harry 00:10 < harry> 1004: from 134.58.241.0/24 lookup 134.58.241-net 1151014311 M * doener_ so the src address in the packets is 134.58.241.32? 1151014321 M * harry i want it to be that, yes 1151014341 M * harry but the virtual machine seems to prefer sending it with 192.168.28.21 as src address 1151014361 M * doener_ because they are on the same net 1151014383 M * harry ?????? 1151014424 M * doener_ ah, sorry, misread something above 1151014532 M * harry np 1151014535 M * harry do you see my problem? 1151014553 M * harry how can i set a "default" source address in virtual machines? 1151014561 M * harry or default interface, whatever 1151014585 M * doener_ in short: you have ip addresses a and b, for both there's a src-based routing that has a default route and the "wrong" src address is chosen 1151014593 M * harry yup 1151014602 M * harry (for THIS server) 1151014610 M * harry other servers may have other default routes 1151014635 M * harry the host itself needs to have 192.168 as default route 1151014646 M * harry virtual machine "stdserver" too 1151014661 M * harry other servers have other default addresses to use as src address 1151014730 M * harry at first i thought /etc/vservers/kromsin/interfaces/dev etc... but... i allready said that ;) 1151015075 J * Nam ~nam@142.179.93.180 1151015483 M * doener_ looking at the kernel source now... 1151016008 M * harry any luck? 1151016059 A * harry has to go to bed now... meeting tomorrow 1151016067 M * harry well... 2 actually 1151016076 M * harry 9-12 and 12.30 - 15u 1151016081 M * harry partyyyyy :s 1151016113 M * doener_ not yet, half-educated guess is that you could change the priority of the rules 1151016121 M * harry yeah 1151016128 M * harry but, that won't be good enough 1151016146 M * harry i also have servers that have other routing priorities 1151016164 M * harry i should be able to set a default src address for each virtual server 1151016173 M * doener_ or some netfilter magic in the prerouting chain 1151016201 M * harry that won't do 1151016209 M * harry that would be a really ugly hack 1151016218 M * doener_ there's no such thing for a plain linux host either, source address is always determined by routing 1151016221 M * harry i could source route it aswell then 1151016254 M * harry i don't get it... there should be a way to set "default source address" in your virtual system 1151016286 P * stefani I'm Parting (the water) 1151016287 M * harry now it's just... allmost randomly pick one 1151016296 M * harry that will be our source address :S 1151016302 M * doener_ no, your routing rules choose one 1151016317 M * doener_ same as on a vanilla kernel 1151016325 M * harry if i have 2 ip addresses, i want to be able to say which one is default imho 1151016334 M * harry vanilla has 1 routing table 1151016349 M * harry i have a default gateway for each network defined 1151016372 M * harry anyway, imho should the choice of "source address" be supported 1151016393 M * harry mailservers should route default over mailservers vlan 1151016398 M * harry webservers over webserver vlan 1151016417 M * harry management stuff should be routed over management vlan 1151016437 M * doener_ your program can choose a source address, if it doesn't do that, the kernel chooses one for you, based on your routing rules 1151016441 J * japan ~japan@dhcp185-3-151.dsl.ucc-net.ca 1151016446 M * doener_ and multi table routing is part of vanilla 1151016453 M * harry i know that 1151016457 M * harry but that's not good 1151016468 Q * schimmi Ping timeout: 480 seconds 1151016468 M * harry i perfectly understand what you're saying 1151016476 M * harry i'm just saying that it's a big problem 1151016523 M * harry this would mean there is no way to put a virtual host on mail AND webserver network with one machine defaulting to mailnetwork and another to webserver network 1151016547 M * harry while that's a perfectly plausible way to do it 1151016550 M * daniel_hozac why wouldn't there be? 1151016558 M * harry how would you do it? 1151016561 M * doener_ define routing rules that are less broad 1151016567 M * daniel_hozac a guest cannot use addresses that are not available to it. 1151016583 M * harry doener_: ??? 1151016590 M * harry daniel_hozac: ? 1151016612 M * ray6 harry; when you put each service in an own vserver, give each one IP, and each service has it's IP - route it on the host like you like :) 1151016622 M * daniel_hozac a guest will never choose a source address which is not available to it. 1151016635 M * harry daniel_hozac: wtf are you talking about??? 1151016641 M * ray6 mail to mail, web to web... can all be perfectly done with linux routing capabilities? 1151016656 M * harry i want a machine on mail AND webserver network 1151016665 M * harry well... 2 machines on both networks 1151016666 M * ray6 hary: a HOST or a vserver? 1151016667 M * daniel_hozac a single guest? 1151016669 M * harry vserver 1151016679 M * harry look... very simple 1151016687 M * harry 2 networks, eth1 on webserver vlan 1151016696 M * harry eth2 on mailserver vlan 1151016699 M * ray6 why web and mail in the same vserver when you have two subnets for that? 1151016714 M * daniel_hozac yeah, seems to defeat the point of separating the networks in the first place. 1151016715 M * harry i want a virtual machine on both networks 1151016728 M * harry which , to the outside world has the webserver source address 1151016738 M * doener_ harry: your neutral-net table allows the vserver to reach everything with the 192.168.x.x source address 1151016746 M * harry and another server, which has , as default outside worlda ddress, the one in mailnetwork 1151016752 M * harry how hard is that to understand? 1151016759 M * doener_ if the default route in that table weren't there, the other address should be chosen 1151016773 M * ray6 harry: are you trying to build some kind of failover setup? 1151016778 M * harry no 1151016782 M * harry i'm trying to get routing right 1151016793 M * harry which part don't you guys understand??? 1151016808 M * harry i really don't see what you don't understand here... it's as clear as a whistly 1151016812 M * harry whistle 1151016813 M * daniel_hozac given that you seem to contradict yourself consistently, i guess starting from the top would be wise. 1151016819 M * harry no i'm not! 1151016826 M * harry my situation??? 1151016834 M * harry i have a machine 1151016836 M * harry 4 interfaces 1151016844 M * harry eth0: vlan 49 webservers 1151016856 M * harry eth1: vlan 168: management + logserver 1151016860 M * harry eth2: mailservers network 1151016863 A * doener_ is out of the discussion... too flamy 1151016877 M * harry eth3: storage network 1151016878 M * shuri lol 1151016899 M * harry now... i have 1 machine, default installation, no public address, so: 1 ip on eth1, routed to management gateway 1151016905 M * ray6 (why don't you trunk it? :) - but OK, that's another topic :) 1151016924 M * harry machine 2: license server, running on eth0 , logging to eth1 1151016925 M * ray6 please don't mix up real and virtual machines :) You have 1 physical host right? 1151016941 M * harry ray6: let me finish first 1151016963 M * harry machine 3: webserver, runnin gon eth0 , loggin gon eth1 1151016963 M * ray6 use "guest" for the vservers, that makes it more clear, or are you taling about 2 physical machines now? 1151016981 M * harry machine 4: mailserver, running on eth2, logging to eth1 1151017003 M * harry machine 5: monitor stuff, running on eth1, but also addresses in eth0 and 2 1151017013 M * harry default on monitor should be over eth1 1151017018 M * harry ray6: machines are virtual machines 1151017037 M * harry since this is #vserver i assume you would be able to guess i'm talking about vserver stuff 1151017055 M * ray6 harry: sure, but many of us have many vservers on multiple real machines :) 1151017063 M * harry i'm talking about 1 "one" "ein" "un" "uno" machine 1151017085 A * doener_ .oO( oh oh, I have two hosts with multiple guests, there must be sth. wrong... ) 1151017089 M * harry ray6: if they were all physical machines, there wouldn't be a problem now, would there???? 1151017098 M * ray6 harry: you're using "machine" for your real machine as well as for your virtual machines and you're surprised nobody understands you 1151017098 M * harry djies.... 1151017114 M * harry 00:56 < harry> ray6: machines are virtual machines 1151017124 M * harry yeah... that's REALLY hard to understand 1151017126 M * ray6 [00:57] i'm talking about 1 "one" "ein" "un" "uno" machine 1151017132 M * harry must be my advanced english 1151017160 M * harry sorry 1151017163 M * ray6 yeah sure :) But OK. Which problem wouldn't be there if those would be real 1151017176 M * harry thanks!!!! 1151017193 M * harry i was kinda hoping you guys would be able to figure that one out yourself... 1151017205 M * ray6 so your real host has lots of IPs in all the 4 interfaces, which are distributed in the 6 vservers? 1151017209 M * harry since i'm explaining it for the 5th time 1151017213 M * daniel_hozac the problem? we're just supposed to infer the problem? 1151017214 M * harry yes 1151017225 M * harry daniel_hozac: yes 1151017234 M * harry read again... if you don't get it 1151017259 M * harry maybe i should have said : i have 1 physical machine 1151017259 M * daniel_hozac i'm not really interested, i'm just generally commenting. 1151017272 M * ray6 harry: I think it's a problem you would have also with 6 real machines. Source IP is based on the routing table, so the mailserver taling to the logging network will always use it's logging network IP for that 1151017339 M * doener_ ray6: actually not... the problem he described was related to traffic that hits the "default" route... that wouldn't appear with multiple hosts 1151017340 M * harry ray6: with a real machine, i just say: ifconfig eth0 ; ifconfig eth1 ; route add default gw 1151017353 M * harry that would make all external traffic go to eth1 1151017362 M * harry ray6: with a real machine, i just say: ifconfig eth0 ; ifconfig eth1 ; route add default gw 1151017371 M * harry that would make all external traffix go over eth0 1151017374 M * ray6 ok. You have external traffic over more than one of these networks? 1151017382 M * harry yes 1151017383 M * ray6 That was the point which was unclear to me so far :) 1151017390 M * harry hence the advanced routing 1151017392 M * doener_ ray6: his rules are just so broad, that everyone with a 192.168.0.0/24 address can use that address to reach external targets 1151017403 M * harry 00:00 < harry> output van ip route ls table 134.58.241-net: 1151017403 M * harry 00:00 < harry> ------------------------------------------- 1151017403 M * harry 00:00 < harry> 134.58.241.0/24 dev eth2 scope link 1151017403 M * harry 00:00 < harry> default via 134.58.241.254 dev eth2 1151017403 M * harry 00:00 < harry> output van ip route ls table neutral-net: 1151017405 M * harry 00:00 < harry> ----------------------------------------- 1151017408 M * harry 00:00 < harry> 192.168.28.0/23 dev eth1 scope link 1151017410 M * harry 00:00 < harry> default via 192.168.29.254 dev eth1 1151017431 M * doener_ s/everyone/every vserver/ 1151017452 M * harry true 1151017508 M * harry would be nice to be able to be able to have a choice when you specify more interfaces, to choose which you want traffic to come from by default 1151017535 M * harry i don't want to have to change my entire routing table every time i add a new host 1151017561 M * doener_ "come from"? 1151017564 M * ray6 harry: yeah you said that before. But that's something linux itself doesn't do. Using multiple tables on the host is the exact replication of what you would have in a physical world 1151017578 M * harry yes, doener_ , come from.... that's the same as: having source address... 1151017586 M * ray6 you also would have to add a new routingtable to each new physical host :) 1151017602 M * harry i come from home, means: my source was home 1151017613 M * doener_ harry: ah, you said that in a interface context, not in a ip address context, so I didn't understand what you meant ;) 1151017640 M * doener_ "come from" meant for me: from where it reaches the box 1151017653 M * ray6 I thine one routingtable per host is exactly what you want and I don't see why you don't like it? 1151017657 M * harry from where it reaches would not be: come from 1151017661 M * harry that would be: goes to 1151017704 M * harry doener_: incoming traffic is never a problem btw 1151017720 M * harry since you can send it to any ip you want 1151017724 M * harry if it's there it's there 1151017750 M * harry i want , in this case , ping to the outside world 1151017755 M * doener_ hm? traffic goes to eth0 (remember, interface context) means that it will reach the box? 1151017757 M * ray6 the alternative would be to use SNAT? 1151017775 M * ray6 for all connections origination from one of host As IP to host As main IP 1151017777 M * harry doener_: yes 1151017788 M * harry ray6: no, that would not be an alternative 1151017793 M * harry look... guys... 1151017807 M * harry i have virtual address 134.58.241.32 1151017812 M * ray6 but that really would just affect the source IP, not the routing. You want to have individual routing for each guest, so use multiple tables! 1151017819 M * harry if that's the only one, it gets routed through 13 1151017822 M * harry 134.58.241.254 1151017835 M * doener_ ray6: he has multiple tables 1151017844 M * ray6 doener: and why isn't that enough? :) 1151017849 M * harry if i now, add a SECOND virtual ip: 192.168.28.21.. it suddenly routes all external traffic over 192.168.28.254 1151017852 M * harry i don't want that 1151017861 M * doener_ 00:10:24 1002: from 192.168.28.0/23 lookup neutral-net 1151017865 M * harry on another machine, i have the other way around 1151017891 M * doener_ 00:00:30 192.168.28.0/23 dev eth1 scope link 1151017891 M * harry i have a 192.168.28.22 address, on which all traffic should be routed to 192.168.28.254 1151017891 M * doener_ 00:00:30 default via 192.168.29.254 dev eth1 1151017907 M * harry but, i also need an interface on 134.58.241 1151017916 M * harry so, routing works fine for the last one 1151017922 M * harry but not for the first virtual machine 1151017926 M * ray6 then add a rule to bring all packets from 192.168.28.22 to the table with default 192.168.28.254? 1151017934 M * doener_ ray6: he has multiple vservers using 192.168.28.0/23 addresses, some should use the default route in that table, some should use the default route in another table 1151017948 M * ray6 doener: then he can add rules for each individual IP? 1151017963 M * ray6 this IP to that table, this IP to that table...? 1151017963 M * harry ray6: you don't want to change routing table for every server 1151017965 M * doener_ but as this rule has higher priority than the other, the other available source address is never chosen 1151017985 M * doener_ ray6: I already suggested that once or twice... 1151017999 M * harry it would be extremely nice, if i could say in dthe vserver config: choose eth1 as default interface 1151018016 M * ray6 harry: but that's what your setup requires. You are mapping something that would have different routingtables for each physical host in the real world. It resulys in multiple routingtables in the virtual world 1151018017 M * harry or choose 0 as default source address 1151018035 M * ray6 harry: you can't just "configure" that. Linux IP selection doesn't work that way 1151018043 M * harry ray6: no kiddin' 1151018049 M * harry i know that too 1151018059 M * ray6 propose that to Linus to include in mainline. routing independent source IP default. 1151018067 M * harry "doesn't work that way" is not a sollution 1151018087 M * ray6 then a configuration option to isolate this to vservers could be added by Bertl in Minutes :) 1151018088 M * harry ray6: it's got nothing to do with linus's kernel 1151018097 M * doener_ ray6: how would that work? that might easily produce packets to which the destination host cannot reply 1151018111 M * harry this should be in vserver 1151018122 M * ray6 harry: of course it has. The feature you want is missing in LINUX. Not in Vserver. Vserver would have to SUPPORT this feature, if it existed, but it doesn't exist! 1151018136 M * harry since different "default source addresses" only makes sense if you have "different machines" 1151018141 M * harry in this case: virtual machines 1151018159 M * harry so you need different "default routes" for your virtual machines 1151018159 M * daniel_hozac harry: so create it. 1151018169 M * daniel_hozac harry: rather than just bitching at us. 1151018170 M * harry daniel_hozac: if only i would know how 1151018172 M * harry where to start 1151018190 M * harry daniel_hozac: i thought that was possible 1151018190 M * doener_ harry: if you set your vserver to choose 10.0.0.1 as default source address. what happens if you sent something to 192.168.0.15 which is a box that has no access to 10.0.0.0/8? 1151018195 M * ray6 harry: but selection of a "default source IP" (always use 10.5.5.5 as IP when nothing specified) could also make sense in the real world. That's the feature you want, and Linux doesn't have it 1151018221 M * ray6 I don't think it's a very useful feature, but that's the feature you want in Linux. 1151018245 M * harry doener_: ahm... it sends it to 10.0.0.1's default gw 1151018246 M * ray6 if it's in linux, it could be supported by vserver :) 1151018255 M * harry which, i don't know... knows the 192.168 network? 1151018277 M * harry ray6: you app has to support that, so that's no problem there 1151018280 M * ray6 but for now: having your tables and adding rules for each new hosts is the way to go 1151018281 M * harry it's supported 1151018299 M * harry well... the reason im bitching about this is http://www.nongnu.org/util-vserver/doc/conf/configuration.html 1151018304 M * harry the flower page 1151018313 M * harry /etc/vservers/vserver-name/interfaces 1151018313 M * harry * bcast 1151018313 M * harry The default broadcast address. 1151018313 M * harry * dev 1151018313 M * harry The default network device. 1151018315 M * harry * mask 1151018318 M * harry The default network mask. 1151018320 M * harry * prefix 1151018323 M * harry The default network prefix-length. 1151018326 M * harry * scope 1151018328 M * harry The default scope of the network interfaces. 1151018331 M * harry that part 1151018336 M * harry i thought... just use default network device eth2 for one host 1151018339 M * harry eth0 for another 1151018341 M * harry that'll do it 1151018342 M * daniel_hozac default as in "you can leave it out of the subdirectories". 1151018347 M * harry but... apparently not 1151018361 M * daniel_hozac you're aware that Linux-VServer networking does not have anything to do with interfaces, right? 1151018369 M * harry yeah 1151018373 M * ray6 these defaults talk about config defaults - writing it once vs. writing it in each config 1151018375 M * daniel_hozac so then wtf are you bitching about? 1151018400 M * daniel_hozac if you already know all that vserver knows is IP addresses? 1151018401 M * harry daniel_hozac: i was kinda hoping that maybe i was wrong 1151018413 M * harry i thought: default network device, being 0 in my case 1151018426 M * harry gandalf:/usr/local/etc/vservers/kromsin/interfaces# ls 1151018426 M * harry 0 1151018449 M * ray6 harry: no, it doesn't work like you want it and it's nothing which could be fixed by fixing vserver-config-tools. The setting you want to change is not there in linux, so it can't be set by vserver 1151018455 M * harry that it would use the address specified in 0 as default source address for that virtual machine 1151018460 M * ray6 think I said that often enough. Believe me or not :) 1151018471 M * harry ray6: i know that 1151018475 M * harry but i don't believe you 1151018485 M * harry since it IS possible 1151018494 M * harry it has to be 1151018495 M * ray6 You cold do it with xen - but there again you have one routing table per server 1151018497 M * daniel_hozac of course it is, but you won't accept the solution. 1151018501 M * ray6 no difference 1151018513 M * daniel_hozac a different routing table for each guest is IMHO the only sane way to go. 1151018515 M * harry you can give 1 virtual machine 1 ip 1151018520 M * ray6 one rule per IP is what you want. OR cluser your IPs in a way that one rule is enough 1151018525 M * harry then, there is no problem about the source the virtual machine uses 1151018532 M * daniel_hozac the only way that actually makes sense and lets you emulate a real machine further. 1151018543 M * harry if you use 2... it suddenly becomes a problem to use that same address as source 1151018561 M * harry daniel_hozac: that's a sollution, i know that too 1151018599 M * ray6 harry: because you are mixing servers which would have different routingtables in the real world, and then want to fix that by using some strange default magic which is not there in linux 1151018604 A * ray6 is off :) 1151018606 M * harry just think it 's strange... you define 1 virtual ip for a vm: no prob, that's the source address for the host 1151018622 M * harry you add another, suddenly it's impossible to keep your source addrses the same 1151018625 M * daniel_hozac how is it strange? it's the expected behaviour. 1151018634 M * ray6 harry: it's the way ip binding works. As long as there's only one, that is used - in real as in virtual servers 1151018639 M * daniel_hozac your routing tables dictate the behaviour you get. 1151018639 M * harry daniel_hozac: true, but not the "wanted" behaviour 1151018648 M * ray6 as soon as there are more IPs, the routingtable decides which to use 1151018673 M * harry yup 1151018683 M * ray6 harry: you have expectations not backed by the way the linux kernel does IP selection 1151018700 M * harry so, my question stays the same... how hard can it be to be able to choose which "virtual interface" is chosen as source ? 1151018703 M * ray6 and it doesn't help that you don't like that :) 1151018718 M * daniel_hozac harry: it's not. you just won't accept the solution. 1151018737 M * harry daniel_hozac: mkay then 1151018751 M * ray6 harry: add a silly feature to the linux kernel. Possibly not too hard. But nothing to be done by vserver, and nothing which would be easily accepted in mainline either :) 1151018772 M * harry daniel_hozac: as i said a million times, changing routes for every virtual machine is just... not an option 1151018787 M * ray6 why not. Do you have millions of vservers? 1151018793 Q * tokkee Remote host closed the connection 1151018802 M * ray6 harry: AND: the alternative is to group your IPs in a usefull manner. 1151018806 J * tokkee ~tokkee@ssh.faui2k3.org 1151018816 M * harry ray6: no, but it 's hell to understand the situation now 1151018816 M * ray6 harry: then one rule would get them all for each set 1151018821 M * doener_ ray6: it would never be accepted ;) it'd mean that you need to scan the whole routing tables to find out if you must use anything but the default, and even then you could get "surprising" results 1151018832 M * harry i'm not the only admin of server park, you konw... 1151018849 M * ray6 doener: I know. daniel knows. But harry doesn't like it :) 1151018877 M * ray6 and just the fact that we're all kernel hackers here doesn't make him accept that. Can't help it :) 1151018917 M * harry tssssss... 1151018937 M * ray6 group your IPs or have one rule per IP 1151018960 M * ray6 that's the advice I can give you. There won't be something better as long as your setup is given 1151018970 M * harry hmmm... 1151018976 M * harry that would mean... 1151018985 M * harry bwerk... 1151019002 M * harry rewrite routing every time a vm starts/stops 1151019052 M * harry ray6: i do see your point ... don't get me wrong 1151019070 M * harry but applications can say which address to use as source address 1151019084 M * doener_ ray6: depending on how easily decipherable the rules should be and the exact requirements, maybe even moving some of the default routes to extra tables that get lower priority rules might work... but that can easily get ugly 1151019089 M * harry so why can't all the processes in a certain context do the same? 1151019125 M * harry doener_: my plan now... since the 192.168 is not used very often for external traffic, move that one last 1151019157 M * harry that would only be a problem for hosts which are on that interface, and another, and need to get to the world over 192.168 1151019171 M * harry or, per ip routign 1151019174 M * harry routing 1151019195 M * doener_ hmm... I wonder if my last statement was plain bullshit... can multiple rules be chosen for a single routing attempt? 1151019217 M * harry doener_: my guess would be: no 1151019225 M * doener_ dito 1151019232 Q * click Ping timeout: 480 seconds 1151019238 M * harry packet matches rules, it stops there 1151019246 M * harry (same as iptables/xtables) 1151019255 M * harry afaik 1151019287 M * coocoon good night to all 1151019292 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1151019298 M * ray6 harry: because the application can define that or leave it undefined. You could (possibly easily) hack that syscall so it makes undefined to some configureable default, but that's such a strange feature nobody would put it into the linux kernel. 1151019326 M * harry would be easy in my case tough :) 1151019332 M * doener_ harry: which kernel are you using right now? 1151019333 M * harry 10 different networks 1151019344 M * harry put yourself on as many networks as you want 1151019352 M * harry and specify which is your default to use 1151019373 M * harry all is there, except the last one, unless you route every ip independently 1151019380 M * harry doener_: 2.6.16.20 1151019395 M * ray6 harry: normally one wants to go to each network with an IP from that network in case one has one 1151019395 M * doener_ vs2.0.2-rcX? 1151019397 M * harry i remembered bertl implementing it for lo btw :) 1151019411 M * harry Linux gandalf 2.6.16.20-vs2.0.2-rc22-grsec2.1.9 #1 SMP Thu Jun 15 11:46:58 CEST 2006 i686 GNU/Linux 1151019424 M * harry ray6: i know 1151019443 M * doener_ harry: the lo-thing is quite different, as you have a special case of routing there 1151019450 M * harry but i've got many virtual machines running on the same hosts 1151019457 M * doener_ you know all the details beforehand 1151019464 M * daniel_hozac harry: i guess include/net/route.h:ip_find_src is a good place to start adding your changes. 1151019486 M * harry doener_: isn't it possible to just check what context you're in, and, depending on that, checking which ip to use as source 1151019497 M * harry then let linux do your routing based on src address 1151019499 M * harry like it normally does 1151019515 M * harry only pick a "default src" per virtual machine 1151019533 M * doener_ if you want a fixed ip for _every_ destination address, yes. 1151019549 M * harry hmmm... 1151019553 M * harry damn, you've got a point 1151019565 M * harry then you'll have to check the destination again first too :s 1151019590 M * doener_ and checking the destination in that case means that you have to walk the whole routing table and not stop on the first match 1151019607 M * harry damn, didn't think that far :s 1151019674 M * doener_ that strikes you in every case where your program doesn't specify a source address. I'm not sure if it will even _badly_ hit sockets that listen in 0.0.0.0:port 1151019689 M * doener_ s/in 0.0/on 0.0/ 1151019731 M * harry i hate having "default gateways" on each network 1151019760 M * harry administration of 1 routing table per ip is a bit too much i think... 1151019770 M * harry not maintainable i'm afraid 1151019788 M * harry i will change some routing stuff tough... 1151019943 M * doener_ maybe you could write some pre-start/post-stop scripts that take care of the rules. the tables can probably stay as they are, you just have to be concerned about the priority of the matching rules as far as I understood your problem 1151019963 M * harry doener_: that's the only problem i have , yes :0 1151019964 M * harry ;) 1151020044 M * harry thanks you guys! 1151020057 M * harry too bad i'm sure there aren't any girls here now ;) 1151020142 M * harry is there a way to dynamically change pref value of a table? 1151020278 J * koft ~koft@cpe-066-057-020-207.nc.res.rr.com 1151020527 M * doener_ ip rule add from `cat interfaces/0/ip`/32 priority `cat context` table `cat interfaces/0/table` 1151020538 M * doener_ harry: could something like that work as prepre-start script? 1151020554 M * harry doener_: i've tried something now 1151020556 M * harry as a test 1151020569 M * harry 1000: from 134.58.10.0/26 lookup 134.58.10-net 1151020569 M * harry 1001: from 10.33.113.0/24 lookup 10.33.113-net 1151020569 M * harry 1002: from 134.58.241.0/24 lookup 134.58.241-net 1151020569 M * harry 5002: from 192.168.28.0/23 lookup neutral-net 1151020569 M * harry 5003: from all to 192.168.30.0/24 lookup internal-net 1151020574 M * doener_ (assuming that your xids are low enough) 1151020581 M * harry doener_: they are ;)) 1151020599 M * harry so, i changed pref to ... welll higher value 1151020714 M * harry i want the address on 134.58.241.0/24 to have advantage over 192.168.28.0/23 addresses 1151020784 M * harry in other words, what do i have to do, to make the 192.168 network get the lowest priority ?