1102118807 N * Bertl_oO Bertl 1102118820 M * Bertl evening folks! 1102118845 M * no_x hi bertl ! 1102118860 M * Bertl hey no_x! how are you? 1102118892 M * no_x fine ! thx und u ? 1102118896 M * no_x and (: 1102118926 M * Bertl fine too, but nobody wants to test the ngnet stuff :/ 1102119025 M * Bertl ah well, I just put it back on hold until folks come screaming for it ;) 1102119476 M * no_x hehe 1102119514 M * no_x sorry bertl i'm stuck to other stuff right know 1102119538 M * Bertl yeah, no problem with that ... 1102119583 M * no_x but will have 2 testmachines soon 1102122226 J * ntrs ntrs@SP2-24.207.228.55.charter-stl.com 1102122471 Q * ntrs Quit: 1102123360 M * Bertl Doener: what is on your linux-vserver schedule next? 1102123371 M * Bertl (if there is any such thing ;) 1102123387 M * Doener good question... i had something in mind... 1102123437 M * Doener guess i'll just (finally) test the legacy LOCK flag fix... 1102123462 M * Bertl ah, userspace stuff then? 1102123515 M * Bertl or do you think that needs fixing on the kernel side? 1102123559 M * Doener we once found out, that a change from 1.9.2 to 1.9.3 (?) was wrong. in vc_new_s_context a check for the private flag was made, where a check for lock was needed (in 1.9.3, was correct in 1.9.2) 1102123601 M * Doener the (correct) check needed for the private flag isn't there in any of them IIRC 1102123617 M * Bertl hmmm, IMHO, that was fixed in 1.9.3 1102123641 M * Bertl or at least right after 1.9.3 that is ... 1102123740 M * Bertl it seems that neither VXF_INFO_PRIVATE nor VXF_INFO_LOCK is used in recent patches ... 1102123752 M * Doener + if (!vx_check(0, VX_ADMIN) || 1102123752 M * Doener + !capable(CAP_SYS_ADMIN) || vx_flags(VX_INFO_PRIVATE, 0)) 1102123761 M * Doener + return -EPERM; 1102123778 M * Bertl obviously a typo ;) 1102123783 M * Doener there should be VX_INFO_LOCK 1102123821 M * Bertl ah, okay, that explains a lot ... let me check that ... 1102123830 M * Doener AFAICT it doesn't matter anyways 1102123873 M * Doener we need to be in context 0 and that one doesn't have a vx_info 1102123896 M * Bertl yeah, just verified that they _are_ in sync ... 1102123935 M * Doener and the check for private is missing, i.e. you can still enter a private context 1102124029 M * Bertl okay, let's look at this ... for me the codepath is only taken if ctx != -2, right? 1102124048 M * Doener yep 1102124085 Q * sannes Ping timeout: 480 seconds 1102124094 M * Doener hm, are the flags shown somewhere in proc? 1102124111 M * Doener ah, got it 1102124111 M * Bertl yep, they are in /proc/self/status 1102124119 M * Doener Flags: 0000000202000018 1102124128 M * Doener # vserver debian enter 1102124128 M * Doener mesg: /dev/pts/1: Operation not permitted 1102124128 M * Doener debian:/# 1102124163 M * Doener VX_INFO_PRIVATE 0x00000008 1102124173 M * Bertl yep 1102124224 M * Bertl so what does this show us? 1102124241 M * Bertl (except that your pts is 'owned' by a different context) 1102124244 M * Doener that we can still enter a private context, but it shouldn't be possible 1102124271 M * Bertl is this now legacy or new style? 1102124284 M * Bertl (new style doesn't care about those flags) 1102124376 M * Doener new style... but i can do the same with legacy chcontext-compat 1102124385 M * Doener # /lib/util-vserver/chcontext-compat --ctx 10 bash 1102124385 M * Doener New security context is 10 1102124463 M * Bertl please enable vserver switch debugging ... just to make sure .. 1102124573 M * Bertl echo 1 >/proc/sys/vserver/debug_switch 1102124573 M * Doener i don't have debugging enabled in my kernel config, strace? 1102124590 M * Bertl well, strace if your strace is patched ... 1102124616 M * Doener hm? we don't need to follow, do we? 1102124661 M * Doener SYS_273(0x3f010001, 0xa, 0xbffff7f8, 0xbffff874, 0upeek: ptrace(PTRACE_PEEKUSER,26846,44,0): No such process 1102124797 M * Doener that's VCMD_new_s_context 1102124808 M * Bertl okay, yeah, I guess I got it ... 1102124857 M * Bertl hmm, looks like I fixed that in the wrong place ... 1102124880 M * Bertl indeed that line should use the VX_INFO_LOCK 1102124902 M * Bertl and there should be a different check, somewhat further down, where the context has ben fetched 1102124908 M * Bertl s/ben/been/ 1102124977 M * Doener exactly 1102125485 M * Bertl okay, what about that: 1102125486 M * Bertl http://vserver.13thfloor.at/Experimental/delta-vs1.9.3.8-private-fix01.diff 1102125570 M * Bertl (sorry for the -p0) 1102125605 M * Bertl hmm, well broken anyway ;) 1102125683 M * Bertl please reload 1102125917 M * Doener looks good 1102125980 M * Bertl okay, you are going to test it? 1102126084 M * Doener will do 1102126111 M * Bertl okay, I added a comment to remind me why the (currently useless) VX_INFO_LOCK is there 1102126138 M * Doener i guess 1.9.3.8.2-private-fix01 will be my kernel? 1102126169 M * Bertl if you like, I have a 2.6.10-rc2-bk17 version for testing too ;) 1102126188 M * Bertl (and it does boot ;) 1102126211 M * Doener uhm, no, i'm pretty happy that 10-rc2 is stable for me... 1102126233 M * Bertl okay, then 1.9.3.8.2 + fix it is 1102126790 J * ntrs ntrs@SP2-24.207.228.55.charter-stl.com 1102126811 M * Bertl welcome ntrs! 1102126986 M * ntrs hi Herbert 1102127187 M * Bertl ah, rc3 is out ;) 1102127438 M * Doener ok, then i'll wait a little longer, was just about to reboot :) 1102127621 M * Doener hey, you're mentioned in the changelog Bertl 1102127634 M * Bertl hmm, I am? 1102127642 M * Doener - Herbert Poetzl's fixes for the following 1102127642 M * Doener -> break character recognition 1102127642 M * Doener -> Magic-SYSRQ handling 1102127642 M * Doener -> Uart software flow control (IrDA console) 1102127674 M * Bertl funny that is the s3c2410 stuff ... 1102127759 M * Bertl but good that they are in mainline now ... 1102128446 Q * tchan Remote host closed the connection 1102131990 J * tchan tchan@c-24-13-81-164.client.comcast.net 1102132069 M * Bertl wb tchan! 1102134348 Q * ensc Ping timeout: 480 seconds 1102135688 J * _no_x vps@213.39.193.41 1102135857 Q * no_x Ping timeout: 480 seconds 1102140445 Q * Nik Ping timeout: 480 seconds 1102141757 M * Bertl okay, I'm off have a nice whatever everyone! 1102141779 N * Bertl Bertl_zZ 1102142865 J * GhostXz lexdronix@CPE000d88a9ac16-CM014250035611.cpe.net.cable.rogers.com 1102152859 Q * albeiro Ping timeout: 480 seconds 1102152869 Q * matti Ping timeout: 480 seconds 1102152910 J * matti matti@linux.gentoo.pl 1102153002 J * albeiro albeiro@linux.gentoo.pl 1102153790 J * soopertrooper456 soopertro@pD9E1F2BF.dip.t-dialin.net 1102154012 P * soopertrooper456 1102159301 Q * berni jupiter.oftc.net uranium.oftc.net 1102159301 Q * UFOczek jupiter.oftc.net uranium.oftc.net 1102159301 Q * sladen jupiter.oftc.net uranium.oftc.net 1102159301 Q * mugwump jupiter.oftc.net uranium.oftc.net 1102159301 Q * v00dY jupiter.oftc.net uranium.oftc.net 1102159301 Q * Doener jupiter.oftc.net uranium.oftc.net 1102159301 Q * maharaja jupiter.oftc.net uranium.oftc.net 1102159301 Q * Pinnen jupiter.oftc.net uranium.oftc.net 1102159301 Q * UnionPivo jupiter.oftc.net uranium.oftc.net 1102159301 Q * mcp jupiter.oftc.net uranium.oftc.net 1102159301 Q * albeiro jupiter.oftc.net uranium.oftc.net 1102159301 Q * DuckMaster jupiter.oftc.net uranium.oftc.net 1102159301 Q * mboman jupiter.oftc.net uranium.oftc.net 1102159301 Q * Pazzo jupiter.oftc.net uranium.oftc.net 1102159301 Q * th jupiter.oftc.net uranium.oftc.net 1102159301 Q * no_maam jupiter.oftc.net uranium.oftc.net 1102159301 Q * TheSeer jupiter.oftc.net uranium.oftc.net 1102159301 Q * dsanta jupiter.oftc.net uranium.oftc.net 1102159301 Q * Plug jupiter.oftc.net uranium.oftc.net 1102159301 Q * pusling jupiter.oftc.net uranium.oftc.net 1102159301 Q * stupidawy jupiter.oftc.net uranium.oftc.net 1102159301 Q * Medivh jupiter.oftc.net uranium.oftc.net 1102159301 Q * logger jupiter.oftc.net uranium.oftc.net 1102159301 Q * flock jupiter.oftc.net uranium.oftc.net 1102159301 Q * cereal jupiter.oftc.net uranium.oftc.net 1102159301 Q * daniel_hozac jupiter.oftc.net uranium.oftc.net 1102159301 Q * _no_x jupiter.oftc.net uranium.oftc.net 1102159301 Q * _sebd jupiter.oftc.net uranium.oftc.net 1102159301 Q * grecea jupiter.oftc.net uranium.oftc.net 1102159301 Q * eyck_ jupiter.oftc.net uranium.oftc.net 1102159301 Q * lilo jupiter.oftc.net uranium.oftc.net 1102159321 J * cereal cereal@ns1.starhosting.de 1102159321 J * daniel_hozac daniel@h212n1fls33o829.telia.com 1102159339 J * albeiro albeiro@linux.gentoo.pl 1102159339 J * DuckMaster Duck@dyn-83-155-108-33.ppp.tiscali.fr 1102159339 J * mboman michael@cm168.sigma231.maxonline.com.sg 1102159339 J * Pazzo thomas@host130-250.pool8172.interbusiness.it 1102159339 J * th tom@pc-4092.ethz.ch 1102159339 J * no_maam erik@datenzone.de 1102159339 J * TheSeer theseer@border.office.salesemotion.net 1102159339 J * dsanta santa@c68.190.156.105.roc.mn.charter.com 1102159339 J * Plug plug@datadot.net 1102159339 J * flock restless@l192-117-111-12.broadband.actcom.net.il 1102159339 J * logger rs@vds.pas-mal.com 1102159339 J * Medivh ck@paradise.by.the.dashboardlight.de 1102159339 J * stupidawy foo@you.wish.you.were.pimp.olicio.us 1102159339 J * pusling pusling@cpe.atm4-0-7285.0x50c44806.boanxx19.customer.tele.dk 1102159353 J * _no_x vps@213.39.193.41 1102159353 J * _sebd konversat@81.56.247.131 1102159353 J * grecea grecea@h-195-22-237-74.mdl.net 1102159353 J * eyck_ eyck@81.219.64.71 1102159353 J * lilo lilo@lilo.usercloak.oftc.net 1102159358 J * berni berni@2001:1b18:202::2 1102159358 J * sladen paul@starsky.19inch.net 1102159358 J * UnionPivo union@clj8-137.dial-up.arnes.si 1102159358 J * maharaja maharaja@ipax.at 1102159358 J * v00dY v00dy@62.241.52.143 1102159358 J * mugwump sv@210-54-92-188.ipnets.xtra.co.nz 1102159358 J * Doener doener@193.24.208.131 1102159358 J * Pinnen pinnen@h194n2fls35o917.telia.com 1102159358 J * mcp hightower@81.17.110.148 1102159436 J * UFOczek ufoczek@hood.openbug.net 1102159925 Q * grecea Remote host closed the connection 1102160162 J * grecea grecea@h-195-22-237-74.mdl.net 1102166707 Q * virtuoso Ping timeout: 480 seconds 1102166856 J * virtuoso s0t0na@mutator.sot.com 1102168105 Q * flock Ping timeout: 480 seconds 1102168960 J * Hollow bene@home.xnull.de 1102169042 P * Pazzo 1102169445 Q * Hollow Ping timeout: 480 seconds 1102171207 N * Bertl_zZ Bertl_oO 1102172356 J * _cereal cereal@217.20.127.85 1102172462 Q * cereal Quit: 1102172465 N * _cereal cereal 1102172947 N * cereal cereal|away 1102173249 N * cereal|away cereal 1102173394 M * mcp moin 1102174239 N * _no_x no_x 1102177555 J * Hollow bene@home.xnull.de 1102177557 Q * Hollow Quit: 1102177680 J * Hollow bene@home.xnull.de 1102178165 Q * Hollow Ping timeout: 480 seconds 1102178176 J * sannes ace@home.skarby.no 1102178355 Q * grecea Remote host closed the connection 1102179927 J * Hollow bene@home.xnull.de 1102179933 M * Hollow sigh 1102180415 Q * Hollow Ping timeout: 480 seconds 1102181360 Q * sannes Read error: Connection reset by peer 1102182217 P * ccooke 1102183154 J * id_ test@relax-media.softwarezentrum.de 1102183718 N * Bertl_oO Bertl 1102183728 M * Bertl evening folks! 1102183754 M * id_ hi Bertl 1102183764 M * Bertl hey id_ how are you? 1102183791 M * id_ bit ill ... but still in the office :( 1102183794 M * Bertl mcp: morning! what happened to wolk? 1102184304 M * Bertl cereal: hey, how are you? 1102188326 J * sannes ace@home.skarby.no 1102188334 M * Bertl welcome sannes! 1102189614 J * Hollow bene@home.xnull.de 1102189624 M * Hollow hey Bertl 1102189636 M * Bertl hey Hollow! why did you sigh? 1102189653 M * Hollow the oftc network breaks way too much 1102189668 M * Hollow or my client does :P 1102189701 M * Bertl hmm ... i.c. ... yeah, the quit/joins are very hight in the last few days ... 1102189716 M * Bertl maybe related to that new virus out there ... 1102189725 M * Hollow but i setup my new dedi with the ngnet patches... seems to work... 1102189737 M * Bertl really? 1102189747 M * Bertl which one? 1102189773 M * Hollow i only have the problem, if i use your vnet2_setup.sh to create the devices in the context 'vserver foo start' complains about the context already exists 1102189812 M * Bertl yeah, it would be better to create them in the pre script 1102189832 M * Hollow i just copied it to prepre-start.d 1102189838 M * Hollow so, it should to pre-start.d? 1102189845 M * Bertl yeah, might be better ... 1102189869 M * Hollow without the sleep 1000? 1102189881 M * Bertl yes, the context should exist at this time 1102189884 M * Bertl (untested) 1102189893 M * Hollow ok.. 1102189907 M * Bertl in my script the sleep is just there to keep the context busy ... 1102189994 M * Hollow doesn't work 1102190018 M * Hollow vc_add_vndev: No such process 1102190019 M * Hollow vc_add_vndev: No such process 1102190050 M * Hollow i thought pre-start is executed inside the vserver already 1102190057 M * Hollow so this won't work then 1102190077 M * Bertl hmm, you might be right ... 1102190094 M * Bertl maybe we need a trickier approach for that ... 1102190111 M * Hollow hm, let's see how the interfaces are created normally 1102190111 M * Bertl let me have a look at the vserver script ... 1102190433 M * Bertl hmm, that isn't really easy ... it will need some support from the tools ... 1102190453 M * Bertl but I guess I have a quick and dirty workaround for now: 1102190514 M * Bertl write a small script which basically does the netdevice setup 1102190529 M * Bertl and make the first command a 'sleep 5' 1102190558 M * Bertl then start this script in the pre-start hook, but in the background 1102190584 M * Bertl in your runlevel script (inside the vserver) add a similar 'sleep 5' 1102190595 Q * Hollow Ping timeout: 480 seconds 1102191236 J * nonamious b80-70-812@nat226.splius.lt 1102191261 M * Bertl welcome nonamious! 1102191280 M * nonamious hi, Bertl 1102193865 J * DuckKing Duck@dyn-83-154-129-21.ppp.tiscali.fr 1102193901 M * Bertl welcome DuckKing! 1102194295 Q * DuckMaster Ping timeout: 480 seconds 1102197254 P * nonamious Leaving 1102198658 J * Hollow bene@home.xnull.de 1102198659 Q * Hollow Quit: 1102198921 J * Hollow bene@home.xnull.de 1102198929 M * Bertl wb Hollow! 1102198937 M * Hollow ehy Bertl 1102198952 M * Hollow any solution yet? 1102198969 M * Bertl yeah, I have two ... one can be read in the irc log 1102199005 M * Bertl and the other will need some changes in the kernel, but I guess the second is the way to go for now (nevertheless you can try the first) 1102199029 M * Hollow where is the irc log? 1102199065 M * Bertl http://irc.13thfloor.at/LOG/ (or various others, see linux-vserver wiki) 1102199134 M * Hollow > write a small script which basically does the netdevice setup 1102199143 M * Hollow this is done by vnet2_stup.sh no? 1102199155 M * Bertl yeah, well you can strip it down a little .. 1102199213 M * Bertl basically that is all you want: 1102199214 M * Bertl /tmp/vnet -x 100 -n 10 -d lo 1102199214 M * Bertl /tmp/vnet -x 100 -n 11 -d eth0 -D en0 1102199231 M * Bertl where 100 = of your server ... 1102199236 M * Hollow yeah.. 1102199241 M * Hollow in pre-start 1102199246 M * Bertl and the 10,11 should be unique over all vservers 1102199284 M * Bertl the idea is to put it in a script, which is started in the background from pre-start 1102199297 M * Bertl (with the additional delay with sleep 5 or so) 1102199338 M * Bertl so that the 'normal' vserver startup can continue and create the context _before_ the script is creting the devices 1102199373 M * Bertl then add a similar sleep 5 to the runlevel script, to make the device addition happen before the runlevels are called ... 1102199420 M * Bertl in addition you should remove any ips assigned to that vserver, and allow the vserver to setup en0 and lo properly from inside (from the runlevel scripts) 1102199480 M * Bertl I know this needs proper integration from the userspace tools, but enrico is busy with his thesis right now ... 1102199545 M * Hollow stoned pre-start.d # vserver --verbose gentoo start 1102199545 M * Hollow ipv4root is now 1102199545 M * Hollow XID: 0 1102199545 M * Hollow vc_add_vndev: No such process 1102199545 M * Hollow vc_add_vndev: No such process 1102199547 M * Hollow ipv4root is now 1102199547 M * Hollow New security context is 1253 1102199549 M * Hollow * Caching service dependencies... [ ok ] 1102199549 M * Hollow * Starting syslog-ng... [ ok ] 1102199551 M * Hollow * Loading key mappings... [ ok ] 1102199551 M * Hollow * Initializing random number generator... [ ok ] 1102199566 M * Hollow stoned pre-start.d # cat vnet 1102199566 M * Hollow #!/bin/bash 1102199566 M * Hollow cat /proc/self/vinfo 1102199567 M * Hollow _VNET=/usr/local/sbin/vnet 1102199567 M * Hollow CTX="1253" 1102199569 M * Hollow $_VNET -x $CTX -n ${CTX}0 -d lo 1102199571 M * Hollow $_VNET -x $CTX -n ${CTX}1 -d eth0 -D en0 1102199610 M * Bertl okay, move that stuff to /tmp/test.sh 1102199628 M * Bertl add a 'sleep 5' before the $_VNET 1102199639 M * Bertl and change the pre-start.d/vnet to 1102199661 M * Bertl ( bash /tmp/test.sh & ) 1102199685 M * Hollow ok... 1102199739 M * Hollow doesn't work... no error, but no interfaces too 1102199751 M * Bertl check with ifconfig -a from inside 1102199763 M * Hollow no... 1102199822 M * Bertl you waited the 5 seconds? 1102199850 M * Hollow yep 1102199863 M * Bertl okay, using namespaces I guess? 1102199892 M * Bertl anyway, let's try the script now (after the vserver was started) 1102199907 M * Hollow dunno... don't looked at namespaces yet, so probably not 1102199930 M * Hollow stoned pre-start.d # sh /tmp/test.sh 1102199930 M * Hollow XID: 0 1102199930 M * Hollow vc_add_vndev: Invalid argument 1102199930 M * Hollow vc_add_vndev: Invalid argument 1102199942 M * Bertl ah, so the devices are there ... 1102199972 M * Hollow ip addr ls doesn't show them 1102199984 M * Bertl on the host or inside the vserver? 1102199988 M * Hollow host 1102199996 M * Bertl ah, they are _inside_ the vserver ;) 1102200018 M * Hollow hm, ifconfig doesn't show them inside teh vserver... iproute is not installed yet inside 1102200041 M * Bertl is proc configured properly (security) to be available inside? 1102200059 M * Hollow i did vprocunhide, yep 1102200067 M * Bertl okay, just checking ... 1102200174 M * Hollow i'll try with iproute2, maybe they are showing up.. 1102200192 M * Bertl okay, but ifconfig should show them too ... 1102200216 M * Hollow ip addr ls doesn't show them... 1102200278 M * Bertl so it might be either a namespace issue or the result of some previous tries ... 1102200291 M * Bertl let's try the following right now: 1102200308 M * Bertl check with vserver-stat that your context indeed has xid=1253 1102200322 M * Hollow yep, it is 1102200340 M * Bertl then let's do vnet -x 1253 -n 10 -d lo 1102200348 M * Bertl (on the host) 1102200364 M * Hollow stoned root # vnet -x 1253 -n 10 -d lo 1102200364 M * Hollow vc_add_vndev: Invalid argument 1102200389 M * Bertl hmm, okay, let's try with something different than 10 1102200396 M * Bertl for example 20 1102200408 M * Hollow no matter... always the same 1102200420 M * Bertl okay, do you ahve dummy configured on the host? 1102200435 M * Bertl is the dummy device available? 1102200441 M * Hollow compiled it, but not booted yet.. sec 1102200537 M * Hollow so, what should i do with the dummy device then? 1102200561 M * Bertl try: vnet -x 1253 -n 30 -d dummy0 1102200575 M * Hollow ok, without creating dummy0 before? 1102200585 M * Bertl ifconfig -a should display dummy0 1102200615 M * Bertl (just without an assigned ip, if the module is loaded) 1102200624 M * Hollow hm... i think i killed the machine :( 1102200653 M * Bertl happens ... reboot and try again ... 1102200675 M * Hollow forgot to call lilo before reboot... this means no reboot untill monday... 1102200700 M * Hollow i hate doing kernels on remote machines 1102200703 M * Hollow :P 1102200708 M * Bertl unfortunate test setup ... 1102200714 M * Hollow yes 1102200725 M * Bertl but thanks for testing anyway ... 1102200764 M * Hollow sure, perhaps sb is available to reboot my machine tomorrow, dunno... it's quite late anyway 1102200939 M * Hollow Bertl: dunno if i already said: it already worked today... i could see en0 inside context 100 and 200 with your initial vnet2_setup.sh but now it didn't work anymore, so let's see tomorrow/day after if it works.. 1102200958 M * Bertl yeah, great! 1102200974 M * Bertl I hope that I have a better solution for the setup tomorrow ... 1102200994 M * Hollow perhaps the enableInterface option could be rewritten to use ngnet 1102201023 M * Bertl no, unfortunately not ... 1102201041 M * Bertl the issue here is, that the context has to exist before the vnet stuff is executed 1102201069 M * Bertl and currently the context is created _after_ the network is configured 1102201070 M * Hollow so it should be called before /sbin/rc? 1102201082 M * Hollow but after the ctx is created 1102201091 M * Bertl yeah, before /sbin/rc but from outside the context 1102201095 M * Hollow yep 1102201115 M * Bertl (which isn't currently possible without some backgrounding and sync tricks) 1102201124 M * Hollow it seems that pre-start.d is where to put it, as you see above i put /proc/self/vinfo in teh script and it said XID: 0 1102201158 M * Bertl yes, but the context is created after that ... and the next hook is post-start, which is from _inside_ the context 1102201232 M * Hollow no 1102201243 M * Hollow zeus post-start.d # vserver gentoo start 1102201243 M * Hollow * Caching service dependencies... [ ok ] 1102201243 M * Hollow * Starting syslog-ng... [ ok ] 1102201243 M * Hollow * Loading key mappings... [ ok ] 1102201243 M * Hollow * Initializing random number generator... [ ok ] 1102201245 M * Hollow * Starting local... [ ok ] 1102201245 M * Hollow XID: 0 1102201247 M * Hollow zeus post-start.d # cat foo.sh 1102201247 M * Hollow #!/bin/bash 1102201249 M * Hollow cat /proc/self/vinfo 1102201271 M * Bertl ah, yes, correct, but it is also after the vserver booted ... 1102201285 M * Hollow yeah, so the services cannot bind 1102201293 M * Bertl exactly ... 1102201311 M * Bertl it would require to make the rc call a separate 'enter' into the context 1102201325 M * Bertl and setup the devices just before that ... 1102201360 M * Bertl but I guess I have a simple workaround for that ... 1102201364 M * Hollow the setup script should be started inside the vcontext script no? 1102201406 M * Bertl yes, but if you use --debug, you'll see that it looks like this: 1102201438 M * Bertl chbind .... chcontext .... chcontext ... call to rc 1102201469 M * Bertl so the context is created and various stuff is configured in one single callchain 1102201477 M * Bertl (including the rc script) 1102201523 M * Hollow just put the call before the /bin/env ? 1102201524 M * Bertl and this becomes even more complicated when you have a real init running ... 1102201580 M * Bertl vserver.start 1102201587 M * Bertl this is executed on startup ... 1102201591 M * Hollow yep 1102201629 M * Bertl just below is_configured=1 the context creation and init script call is done 1102201661 M * Bertl so you would need to hook in there ... being in context 0 1102201688 M * Bertl which isn't really possible right now ... 1102201730 M * Bertl once enrico switches to a multithreaded synchronized C startup code ... this (and other things) can be done easily ... 1102201762 M * Hollow mhm 1102201769 M * Bertl until then, we'll use a simple trick, and make a sync call to userspace with the vshelper ... 1102201785 M * Hollow ok... 1102201791 M * Bertl (which will happen _after_ the context is created, but before the syscall returns) 1102201807 M * Bertl this way the vshelper can setup the devices properly ... 1102201840 M * Hollow ok, let's try this when the machine is up again ;) 1102201849 M * Bertl yeah! 1102202325 Q * Hollow Ping timeout: 480 seconds 1102203261 J * soopertrooper456 soopertro@pD9E1F2BF.dip.t-dialin.net 1102203271 M * Bertl welcome soopertrooper456! 1102203617 P * soopertrooper456 1102203647 M * Doener evening! 1102203658 M * Bertl evening Doener!