1169510472 M * PowerKe From that documentation it doesn't seem to be related to any file attributes 1169510494 M * PowerKe (from the mount man page) 1169510526 M * Bertl_oO will check that shortly, but it will take a little, have to get some work done right now, will you be around in a few hours? 1169510580 M * PowerKe I'll be next online in 8 hours 1169510592 N * sladen_ sladen 1169510651 M * PowerKe I just realize that I'm using rather low context numbers for my vguests: 2 and 3, I hope there's nothing reserved about those? 1169510665 M * Bertl_oO nope, not yet 1169510739 M * PowerKe I'll be back online in about 8 hours, on and off until tomorrow this time 1169510757 M * Bertl_oO okay, will investigate 1169510765 M * PowerKe If you like you can have my gmail / google talk account? 1169510795 M * Bertl_oO should be fine if you show up again ... 1169510803 M * PowerKe Will do 1169510807 M * PowerKe Thanks already! 1169510812 M * Bertl_oO you're welcome! 1169511257 Q * PowerKe Quit: off to bed, cya tomorrow 1169514439 M * Bertl_oO okay, off to bed now ... 1169514445 N * Bertl_oO Bertl_zZ 1169517318 Q * pusling Ping timeout: 480 seconds 1169520659 Q * Aiken Ping timeout: 480 seconds 1169521925 J * Aiken ~james@tooax6-132.dialup.optusnet.com.au 1169521937 Q * Aiken Read error: Connection reset by peer 1169521941 J * Aiken_ ~james@tooax6-132.dialup.optusnet.com.au 1169529058 Q * Piet Quit: Piet 1169530392 J * pusling pusling@195.215.29.124 1169536509 Q * kugg Ping timeout: 480 seconds 1169536942 J * gab ~gab@158.36.45.236 1169538737 J * dna ~naucki@35-243-dsl.kielnet.net 1169539295 J * rikkit ~rikkit@77.50.50.148 1169539425 J * PowerKe ~tom@d54C13E4B.access.telenet.be 1169540281 Q * dna Quit: Verlassend 1169540289 Q * rikkit Quit: Leaving 1169540377 J * Aiken__ ~james@tooax8-136.dialup.optusnet.com.au 1169540445 Q * dlezcano Ping timeout: 480 seconds 1169540730 Q * Aiken_ Ping timeout: 480 seconds 1169542085 J * rgl ~Rui@84.90.10.107 1169542091 A * rgl aves 1169543317 P * click [IRSSI] 1169543372 J * click click@ti511110a080-3802.bb.online.no 1169543604 M * daniel_hozac PowerKe: vhashify should only care about mtime. 1169543745 M * daniel_hozac PowerKe: and failing to set the times is expected, the file is immutable after all. 1169543779 M * PowerKe Hmm, I might've mixed up ctime and mtime 1169543836 M * PowerKe From what I've noticed, emerging a package in 2 gentoo guests and running vhashify doesn't unify the files because of different timestamps 1169543851 M * daniel_hozac sounds likely. 1169543855 M * PowerKe Which looks like normal behaviour, it's just annoying 1169543876 M * PowerKe So that's why I'm changing the timestamps on different guests 1169543882 M * waldi no, it is a result of the gentoo way of live 1169543892 M * daniel_hozac waldi++ :) 1169543926 M * daniel_hozac of course, you could just use a guest to build binary packages for the rest. 1169543959 M * PowerKe No, Gentoo assigns the timestamp when merging the files. So using a binary package doesn't help 1169543980 M * daniel_hozac i.e. it doesn't store the timestamps? 1169544001 M * PowerKe I guess it could use the timestamps in the archive, but it doesn't. 1169544027 M * daniel_hozac sounds like you're trying to fix the wrong thing then. 1169544057 A * waldi sets the gentoo-is-useless counter a littlebit higher 1169544087 M * PowerKe Maybe, but since I don't know python, it seemed easier to work around it instead of fixing it at the source. 1169544152 M * daniel_hozac python's really easy. and if you don't want to do it, i'm sure a bug report would get someone's attention. 1169544196 M * PowerKe Unless it's not considered a bug, but I might give that a try. 1169544225 M * daniel_hozac the purpose of binary packages is to get identical installs of the package across multiple boxes, no? 1169544261 M * PowerKe yes 1169544298 M * daniel_hozac so [cm]time ought to be included, just like uid and gid. 1169544359 J * DavidS ~david@vpn.uni-ak.ac.at 1169544448 M * PowerKe At least it would make running vhashify easier on gentoo guests. 1169545401 J * dlezcano ~dlezcano@blueice1n1.uk.ibm.com 1169546020 J * fabrice ~fabrice@86.68.27.99 1169546028 M * fabrice hi 1169546044 M * daniel_hozac hello 1169546111 M * fabrice i'm migrating guest with old style configuration (one file) to new style (many files) 1169546124 J * debugger ~Rui@84.90.10.107 1169546129 M * fabrice do you know how to set ulimt with new style ? 1169546133 M * fabrice ulimit 1169546138 M * daniel_hozac the ulimits directory. 1169546147 M * fabrice ha 1169546170 M * daniel_hozac that is, _if_ you want to set ulimits. 1169546180 M * daniel_hozac for per-guest limits, you want the rlimits directory. 1169546188 M * fabrice ok 1169546205 M * fabrice i made a rlimits dir but it doesn't work 1169546213 M * fabrice something more to add in config ? 1169546236 M * fabrice rlimits/as:90000 1169546236 M * fabrice rlimits/nofiles:512 1169546266 M * fabrice but iin the vserver nofiles stay at 1024 (as it is on the host) 1169546287 M * fabrice same for as, it stay at no limit 1169546305 Q * shedi Quit: Leaving 1169546308 M * daniel_hozac the rlimits won't show up in ulimit -a. 1169546362 M * fabrice ok 1169546370 M * daniel_hozac check /proc/virtual//limit for the current values, the limit, etc. 1169546394 M * daniel_hozac it's nofile too, not nofile_s_. 1169546460 M * fabrice argh, my mistake 1169546466 M * fabrice ok i try 1169546514 Q * rgl Ping timeout: 480 seconds 1169546551 M * fabrice VM: 6587 6587 90000 0 1169546551 M * fabrice VML: 0 0 -1 0 1169546551 M * fabrice RSS: 1778 1778 -1 0 1169546552 M * fabrice ANON: 433 433 -1 0 1169546552 M * fabrice FILES: 132 132 512 0 1169546554 M * fabrice nice :) 1169546562 M * fabrice thanks for your help daniel_hozac 1169546572 M * daniel_hozac you're welcome! 1169546815 M * fabrice one more question: you said ulimit dir apply to all vserver, but where is that dir ? 1169546868 M * daniel_hozac no, i said that rlimits apply to the entier guet. 1169546882 M * daniel_hozac the ulimits directory is just for a default ulimit -a. 1169546891 M * fabrice hmm, ok, i see 1169546896 Q * dlezcano Read error: Connection reset by peer 1169546908 M * daniel_hozac (granted, the guest cannot raise the values above the hard limit) 1169546938 M * fabrice i can set /etc/security/limits.conf and that conf will apply to the host and all guest, if no rlimits set 1169546955 M * daniel_hozac not with recent util-vserver versions. 1169546970 M * daniel_hozac 0.30.212 will reset all ulimits by default. 1169547282 M * daniel_hozac Hollow: copying resolv.conf is already done, if you put the file in /etc/vservers/.defaults/files/resolv.conf. 1169548057 J * dlezcano ~dlezcano@blueice1n1.uk.ibm.com 1169548094 M * PowerKe Just found something about the mtime problem: http://bugs.gentoo.org/show_bug.cgi?id=153633 1169548098 M * PowerKe We use a container format called xpak and it only stores the raw files (no 1169548099 M * PowerKe metadata such as timestamps). 1169548158 M * PowerKe Status: RESOLVED CANTFIX 1169548442 Q * Aiken__ Quit: Leaving 1169548487 Q * dlezcano Read error: Connection reset by peer 1169548641 M * PowerKe Hmm, just looked in a binary package and there is mtime info in the archive, so it must be something else... I'll just log a new bug report. 1169549312 J * dlezcano ~dlezcano@blueice2n1.uk.ibm.com 1169550126 J * rgl ~Rui@84.90.10.107 1169550520 Q * debugger Ping timeout: 480 seconds 1169552478 M * matti _@/" 1169552479 M * matti :) 1169552489 Q * cdrx Ping timeout: 480 seconds 1169552852 J * dna ~naucki@202-239-dsl.kielnet.net 1169553740 J * kugg ~kugg@90-227-91-196-no120.tbcn.telia.com 1169553947 Q * ensc Ping timeout: 480 seconds 1169554701 J * Piet hiddenserv@tor.noreply.org 1169554708 P * chaotika 1169555061 Q * Smutje Remote host closed the connection 1169555664 M * daniel_hozac wow, looks like there is some interest in 2.6.16-vs2.0. 1169556594 M * daniel_hozac there's just one reject though, and two fuzzes. 1169556726 Q * kugg Ping timeout: 480 seconds 1169557176 M * daniel_hozac (using 2.0.2-rc22, that is) 1169557315 J * max22 ~programsp@i3ED6E4F7.versanet.de 1169557315 M * max22 S@luto tutto il Ch@nnel [#vserver ] --[ Siciliani Script v.1 ]-->> 1169557367 Q * max22 1169557482 M * DreamerC ... 1169557489 M * DreamerC spam ... 1169558320 J * kugg ~kugg@90-227-91-196-no120.tbcn.telia.com 1169558635 J * ensc ~irc-ensc@p54B4D22F.dip.t-dialin.net 1169560963 J * cdrx ~legoater@blueice2n1.uk.ibm.com 1169561849 M * renihs ugly spam too 1169562849 J * shedi ~siggi@v10-222-142.lhi.is 1169563269 A * nox suggests a /mode +c 1169563317 M * nox to strip lousy colors 1169563429 M * daniel_hozac you know, most clients let you disable colors... 1169563475 M * nox ofcause but i prefer it per channel 1169563525 M * nox also i like topiclock (; 1169563544 M * daniel_hozac Bertl_zZ: http://people.linux-vserver.org/~dhozac/p/k/patch-2.6.16.37-vs2.0.2.1.diff 1169563563 M * daniel_hozac most people are mature enough to leave it alone. 1169563573 M * daniel_hozac same policy we use with the wiki... 1169563649 M * nox yes i know ... trust is a good thing 1169565422 Q * DavidS Remote host closed the connection 1169565602 M * ard daniel_hozac : namespace id, is that the same as context id? 1169565617 M * daniel_hozac ard: hmm? 1169565632 M * ard I mean: there is only a network seperation and a process seperation? 1169565645 M * ard I was looking at vnamespace, chcontext and chbind 1169565655 M * daniel_hozac yes. 1169565662 M * ard I want to understand what is going on ;-) 1169565663 M * ard ah ok... 1169565699 M * ard and when I understand it, I have to explain my collegaes :-) 1169566808 M * Hollow daniel_hozac: another update: http://paste.linux-vserver.org/913 1169566853 M * Hollow did not test the chrootsed() yet.. 1169566896 M * daniel_hozac as i said last time, /etc/resolv.conf doesn't belong there. 1169566932 M * Hollow ah, i didn't see that .. 1169566941 M * Hollow why does resolv.conf not belong here? 1169566975 M * daniel_hozac 11:14 < daniel_hozac> Hollow: copying resolv.conf is already done, if you put the file in /etc/vservers/.defaults/files/resolv.conf. 1169566984 M * Hollow ah, neat 1169567169 M * daniel_hozac net.vserver is missing from distrib/Makefile.am? 1169567211 M * daniel_hozac other than that, looks fine to me. 1169567222 M * daniel_hozac assuming chrootsed works, of course. 1169567247 M * Hollow ah, good point... will test chrootsed in a sec.. 1169567840 Q * gab Remote host closed the connection 1169568706 Q * Piet Remote host closed the connection 1169568807 J * Piet hiddenserv@tor.noreply.org 1169570275 J * stefani ~stefani@tsipoor.banerian.org 1169570478 M * Hollow daniel_hozac: ok, chrootsed works, have commited the changes 1169570497 M * daniel_hozac okay, cool. 1169570593 M * Hollow much automagic already .. :) 1169570595 M * Hollow http://paste.linux-vserver.org/914 1169570683 M * daniel_hozac nice! 1169570817 M * harry the grsec patch is updated to fix the grsec bug that was in it... 1169570887 Q * shedi Quit: Leaving 1169571566 M * daniel_hozac Hollow: btw, have you seen the vlogin thread? runaway CPU usage when losing the terminal? 1169571688 M * Hollow on the ML? 1169571701 M * daniel_hozac yeah, was a while ago now, but i have been unable to track it down. 1169571711 M * Hollow no, didn't see it .. 1169571743 M * daniel_hozac http://archives.linux-vserver.org/200701/0054.html 1169571957 M * Hollow daniel_hozac: i.e. the vlogin process did not exit on logout? 1169571961 M * daniel_hozac right. 1169572076 M * Hollow and even in your tests vlogin did not vasnish? just not eating up cpu? 1169572107 M * daniel_hozac it depends. sometimes it just sticks around forever, sometimes it'll also consume CPU. 1169572131 M * daniel_hozac it seems to be stuck in the select(), terminal_copy() loop. 1169572245 M * Hollow looks like the signal handler is not called to kill the terminal .. 1169572264 J * bonbons ~bonbons@83.222.37.103 1169572267 M * daniel_hozac i don't think vlogin is getting any signal. 1169572285 M * Hollow SIGCHLD? 1169572294 M * daniel_hozac that's for the parent. 1169572299 M * daniel_hozac vlogin is the child. 1169572314 M * daniel_hozac but the parent bash process seems to stick around too. 1169572321 M * Hollow you mean the vlogin that actually calls itself "login"? :) 1169572335 M * daniel_hozac right. 1169572345 M * daniel_hozac http://paste.linux-vserver.org/915 1169572380 M * daniel_hozac strace shows http://paste.linux-vserver.org/916 over and over. 1169572391 Q * ntrs Remote host closed the connection 1169572420 M * daniel_hozac (this has http://people.linux-vserver.org/~dhozac/p/uv/experimental/delta-vlogin-efds.diff applied to attempt to fix the problems) 1169572544 Q * fabrice Quit: using sirc version 2.211+KSIRC/1.3.12 1169572551 M * daniel_hozac which obviously doesn't do the trick :) 1169572766 M * Hollow well, if the program invoked by vlogin vanishes it should exit in the SIGCHLD exit handler too.. so, i still think the signal handler does not get invoked... 1169572780 M * Hollow s/exit/signal/ 1169572793 M * daniel_hozac it's not the program vlogin starts that's dying. 1169572799 M * daniel_hozac it's the parent. 1169572814 M * Hollow the vserver ... enter script? 1169572834 M * daniel_hozac no, the su proces. 1169572838 M * daniel_hozac +s 1169572854 M * daniel_hozac http://paste.linux-vserver.org/917 1169572864 M * daniel_hozac i kill -9 6734 1169572910 M * Hollow 6743 is now parent of init? 1169572915 M * Hollow the bash process 1169572919 M * Hollow ehm 1169572921 M * Hollow child of init 1169572958 M * daniel_hozac right, since the parent dies, it gets reparented to init. 1169572983 M * Hollow but the FDs are dead .. 1169573661 J * yarihm ~yarihm@vpn-global-dhcp1-230.ethz.ch 1169573981 Q * dlezcano Read error: Connection reset by peer 1169574433 M * daniel_hozac so, any ideas? i thought using the exception fd_set would let us detect EOF, but apparently not. 1169574575 Q * cdrx Read error: Connection reset by peer