1143417668 N * bertl_oO Bertl 1143417672 M * Bertl back now ... 1143420325 J * dos000 ~dos000@i216-58-67-245.cybersurf.com 1143420377 M * dos000 im getting ": No space left on device" when i try to copy stuff on /tmp .. any idea ? 1143420448 Q * doener Quit: leaving 1143420480 M * Bertl dos000: probably the device is full? 1143420489 M * Bertl what does df say? 1143420621 M * dos000 Bertl, in the vserver ? 1143420632 M * Bertl yes, for example 1143420638 M * dos000 /dev/hdv1 68G 642M 67G 1% / 1143420639 M * dos000 none 0 0 0 - /proc 1143420639 M * dos000 none 16M 16M 0 100% /tmp 1143420639 M * dos000 none 0 0 0 - /dev/pts 1143420649 M * dos000 oops 1143420683 M * dos000 i need more than 16m then 1143420691 M * dos000 how do you get it up ? 1143420702 M * Bertl you can configure that in the guest config (fstab) 1143420727 M * Bertl cat /etc/vservers/test01/fstab 1143420732 M * Bertl none /tmp tmpfs size=16m,mode=1777 0 0 1143420735 M * Bertl for example 1143420751 M * dos000 thanx 1143420758 M * Bertl np 1143420867 J * entroposcope ~entroposc@user-0c992og.cable.mindspring.com 1143420984 M * Bertl welcome entroposcope! 1143421298 J * click click@ti511110a080-0682.bb.online.no 1143421369 M * Bertl welcome click! 1143421658 Q * virtuoso_ Remote host closed the connection 1143421676 J * virtuoso ~s0t0na@shisha.spb.ru 1143422303 Q * pflanze Quit: [x]chat 1143423182 M * Bertl kir: ping? 1143423375 M * dos000 Bertl, care to help .. i am trying to setup vhcs in a vserver. However i am getting (98)Address already in use: make_sock: could not bind to address 1143423398 M * dos000 from apache ... i had to comment some of the iptables code that came with vhcs 1143423414 M * daniel_hozac dos000: does the host already use the port it tries to bind? 1143423417 M * Bertl no idea what vhcs is, but maybe the host is already using a port? 1143423441 M * dos000 no one is onthat port. but i think it is going on the host maybe 1143423497 M * dos000 i even changed the apache2 file to cpecifically listen on the guest ip:port 1143423524 M * dos000 and no one on the guest is using port 80 1143423550 M * Bertl dos000: you do not have to change anything inside the guest 1143423566 M * Bertl dos000: but!, you ahve to restrict services on the host 1143423581 M * Bertl dos000: an apache running on the host, will bind _all_ IPs 1143423596 M * dos000 Bertl, i have changed the etc/hosts to remove 127.0.0.1 1143423607 M * dos000 ah ! 1143423636 M * Bertl dos000: the guest will _automatically_ be limited to his ips 1143423727 M * dos000 ok .. thanks guys. apache on the host was binding on all ips as told by Bertl 1143423759 M * Bertl that's a common mistake, so no worries ... 1143423813 M * dos000 i knew it for sshd but forgot it for apache :-( 1143432173 M * Bertl okay, I'm off for tonight ... back tomorrow! 1143432181 M * Bertl have a good one everyone, cya! 1143432191 N * Bertl Bertl_zZ 1143432728 M * ntrs How do I enable CFQ for the vservers? Is there some vserver config parameter or something similar, or is it enough foir cfq to be compiled in the kernel? 1143435444 M * dos000 is it a good idea to just replace 127.0.0.1 in the host to the actual server ip ? it looks like all services bind to 0.0.0.0 1143436629 M * daniel_hozac ntrs: just compile it into the kernel. 1143436932 M * Hollow morning 1143436943 M * Hollow *yawn* 1143436943 M * daniel_hozac morning! 1143436986 M * dos000 *yawn* .. good night 1143436991 M * Hollow heh 1143436995 M * Hollow night 1143437728 J * dos000_ ~dos000@i216-58-25-69.cybersurf.com 1143438083 Q * dos000 Ping timeout: 480 seconds 1143438138 M * anonc Hollow: where is the util-vserver package for gentoo heading these days? I've tried using the latest ~x86 version with 2.6.16-vs2.1.something and it doesn't seem to work too well. 1143438153 Q * Dr4g Ping timeout: 480 seconds 1143438783 M * daniel_hozac anonc: what do you mean? 1143438907 M * Hollow anonc: please elaborate, it works pretty well here 1143439007 M * Hollow daniel_hozac: btw, can you check the killer part in vcd spec? hope i got it right this time 1143439009 M * Hollow http://home.xnull.de/work/vserver/vcd.spec.html#vps-killer 1143439060 M * anonc util-vserver-0.30.210-r6 on 2.6.16-vs2.1.1-rc14. Um - by any chance have dynamic context IDs had their support removed from the utils? 1143439077 M * Hollow hm, no, but you should not use them 1143439083 M * Hollow and probably they are disabled in the kernel 1143439100 M * anonc no - i explicitely enable them in my kernel config 1143439105 M * Hollow dynamic context ids are a Bad Thing (tm) 1143439114 M * daniel_hozac anonc: what makes you think that dynamic contexts aren't supported anymore? 1143439115 M * anonc I know - but I am lazy (tm) 1143439118 M * daniel_hozac Hollow: looks fine. 1143439126 M * Hollow daniel_hozac: good :) 1143439167 M * anonc vserver-stat sometimes says my vservers have wonky uptimes: # vserver-stat 1143439168 M * anonc CTX PROC VSZ RSS userTIME sysTIME UPTIME NAME 1143439168 M * anonc 0 71 174.4M 120M 41m06s60 14m18s90 1d13h51 root server 1143439168 M * anonc 49152 4 12.9M 2.2M 0m00s47 0m00s54 1m25s41 corp 1143439168 M * anonc 49195 1 2.1M 436K 0m00s00 0m00s00 INVALID users 1143439169 M * anonc 49196 4 6.4M 2.1M 0m00s16 0m00s80 0m01s59 mail 1143439170 M * anonc grr - sorry 1143439184 M * Hollow daniel_hozac: ah.. just one last thing.. reboot 4) has to happen in background, right? the helper should return immediately after calling the rc scripts? 1143439205 M * anonc and when you try to enter them, sometimes they let you in - then quickely print out 'killed' and I'm booted out again. 1143439213 M * daniel_hozac Hollow: no, that should be executed by the helper. 1143439224 M * anonc then the context is non-existent for a little while, then vserver-stat says its back again 1143439240 M * Hollow sounds like a reboot, especially the "Killed" thing 1143439260 M * Hollow daniel_hozac: ah, so the killer also kills the helper when the rc scripts loop inifnite.. 1143439270 M * daniel_hozac Hollow: exactly. 1143439274 M * Hollow ok, fine then 1143439386 M * daniel_hozac anonc: does that also happen for the corp guest? 1143439436 M * daniel_hozac anonc: you sure you didn't set something like runlevel 6 as the default for your guest(s)? 1143439614 M * anonc gentoo host and clients - all initdefault 3 1143439664 M * anonc it happens occasionally to all the guests. this is on a test machine - I have many others on older versions running nicely differing only in kernel and util-vserver versions. 1143439723 M * anonc just wondering whether libvserver or anything has been incorporated in the later util-vservers or anything like that. 1143439948 M * anonc the obvious difference would be that I switched over to 'plain' init style since 'gentoo' has been depreciated. 1143440743 J * doener ~doener@i5387F76A.versanet.de 1143444240 J * ||Cobra|| ~cob@146.50.22.204 1143444625 M * Hollow anonc: are these newly created guests or migrated ones? 1143444631 M * anonc migrated 1143444634 M * Hollow i.e. is baselayout-vserver up to date 1143444682 M * Hollow look at the blue box at the top of http://linux-vserver.org/GentooGuestHowto 1143444697 M * Hollow there are instructions how to migrate/upgrade an older version or baselayout 1143444781 M * anonc Hollow: that's probably it. the guests still use baselayout 1143444796 M * Hollow oh well, ancient times :) 1143444804 M * Hollow you really should upgrade 1143444821 M * anonc yeah - didn't realize a util-vserver upgrade would be affected. 1143444821 M * Hollow you also know http://www.gentoo.org/doc/en/vserver-howto.xml ? 1143444837 M * Hollow well, everything before baselayout-vserver was a hack 1143444845 M * anonc yeah - i know all that stuff - i just don't tend to update the guests very often. 1143444867 M * Hollow yeah, but at least baselayout-vserver should be used, it really eases things 1143444908 M * anonc i think the guests went in maybe 1.5 years ago 1143444949 M * Hollow the old hacked baselayout won't work correctly with plain, that's your problem i guess, but the gentoo init style is not possible like util-vserver does it now so we had to deprecate it, and use plain for gentoo 1143444991 M * anonc that'll be it. shame about the gentoo init style - i liked seeing my guest init script output roll past... 1143445027 M * Hollow yeah, many users miss that, but it's a much cleaner solution, and maybe some day we'll even get the output of init again ;) 1143445037 M * Hollow but therefore a virtualized /dev/console is needed 1143445061 M * anonc i've been wanting to play with a 2.6.16 release for the fix to the kjournald namespace issue 1143445103 M * Hollow i guess you could give 2.0.2_rc14 a try, or wait for the final 2.0.2, will be in portage as soon as it is released 1143445152 M * anonc nah - for playing i use 2.1. i'm seeing whether drbd + cow links can assist in testing upgrades. 1143445164 M * Hollow ah, ok.. 1143445202 M * anonc are the vserver-sources ebuilds essentially gentoo-sources + vserver? 1143445204 M * Hollow hm, maybe i should add a warning to the deprecate message for baselayout-vserver 1143445214 M * Hollow anonc: yep, vanilla + genpatches + vserver 1143445306 M * anonc *sigh* vserver is so much easier to deal with than this @#!%#^@%^! nforce4 motherboard. sata + forcedeth|nvnet = no network. what a POS. 1143445558 M * anonc Hollow: any reason for this running inside my guest?: /usr/sbin/vcontext --silent --migrate --chroot --xid 52149 --uid root --vlogin -- /bin/bash -login 1143445594 M * Hollow yes, this is the new vlogin, a terminal proxy so you get /dev/pts/ entries right on vserver ... enter 1143445622 M * Hollow else you won't be able to use any app using /dev/pts, like screen or mc 1143445700 M * anonc aha. thanks for the info. must have missed the discussions on that one 1143445724 M * Hollow http://list.linux-vserver.org/archive/vserver/msg12117.html 1143445732 M * Hollow gtg to school now, cu later 1143445769 M * anonc bye - thanks again 1143446679 M * eyck hmm 1143446801 M * BartVB is there a version of vtop that also dispays which process belongs to which vserver? 1143447820 A * BartVB just discovered vapt-get, cool! :) 1143448038 M * SiD3WiNDR BartVB: not at the moment afaik :/ 1143448283 J * Smutje ~Smutje@xdsl-87-78-40-144.netcologne.de 1143448382 M * BartVB ah, bummer. Would be nice 1143448754 J * FireEgl Atlantica@Atlantica.US 1143449092 N * Bertl_zZ Bertl 1143449096 M * Bertl morning folks! 1143449120 M * daniel_hozac morning Bertl! 1143449154 M * FireEgl g'morning Bertl. =) 1143449407 J * Dr4g ~Dr4g@82-40-202-68.stb.ubr06.uddi.blueyonder.co.uk 1143449857 M * Bertl okay, off for now .. back later ... 1143449861 N * Bertl Bertl_oO 1143451205 Q * dos000_ Quit: Leaving 1143451489 J * cryo ~say@psoft.user.matrix.farlep.net 1143454941 N * Bertl_oO Bertl 1143454945 M * Bertl back now 1143454969 M * Bertl hey daniel_hozac! didn't see you :/ 1143455255 M * Hollow back too 1143455267 M * Bertl hey Hollow! 1143455294 M * Hollow hey Bertl 1143455311 M * Hollow the new vps killer chapter is now in the vcd spec 1143455339 M * Bertl btw, is that linked from the wiki somewhere? 1143455387 M * Hollow not yet, but i will publish the first public draft soon to the ML i guess, can also link it on the wiki then 1143455405 M * Bertl okay, make that an entry on the devel section 1143455410 M * Hollow yup 1143455443 M * Hollow if have any comments to the rest of the spec feel free to tell me 1143455447 M * Hollow +you 1143455546 M * Hollow Bertl: maybe you can help me a bit with the Data Collector, i'm not sure which information can be collected 1143455551 M * Hollow or should be 1143455574 M * Hollow maybe we need some addiotnal syscalls that too? 1143455578 M * Hollow +for 1143455579 M * Bertl okay, regarding the userspace stop/restart/reboot interaction 1143455604 M * Bertl (as you seem to have been somewhat tired yesterday) 1143455608 M * Hollow :) 1143455633 M * Bertl we came to the 'informal' agreement to leave the sync vs async as is 1143455641 M * Hollow yeah, i read that 1143455647 M * Bertl and provide an additional syscall command to read the vci 1143455648 M * Hollow get config via VCI etc 1143455667 M * Bertl yep, also this means for userspace that the procedure got slightly complicated 1143455706 M * Hollow what has to be done beside the async check? 1143455709 M * Bertl but, what I see in your list is the point 1. Record reboot request 1143455746 M * Bertl in the sync case, as ensc mentioned, it isn't necessary to put that somewhere, it is sufficient to pass this information to the killer 1143455770 M * Hollow that's what the spec does atm 1143455789 M * Hollow record reboot request means: tell the killer to reboot in the end 1143455809 M * Bertl in the legacy 'async' case, you have the 'new' choice of selecting reboot_kill too 1143455824 M * Bertl (starting with 2.0.2/2.1.1) 1143455836 M * Bertl currently this is exclusive 1143455874 M * Hollow hm, can you (again?) elaborate what's the difference between async and sync... i mean what is in sync or not? 1143455899 M * Bertl the helper will be spawned before the reboot_kill happens, but for the async behaviour this doesn't necessarily mean that userspace is executed before the kill happens 1143455930 M * Hollow i.e. in the async case the helper returns immediately 1143455936 M * Bertl yes, it's quite simple: in the sync case, the kernel waits for userspace to handle the request (i.e. until vshelper finishes) 1143455955 M * Bertl once that is done, the reboot_kill (if flagged) will kick in 1143455993 M * Bertl it also allows (in theory) to pass back values to the kernel, which might for example 'deny' the sys_reboot() 1143455996 M * Hollow ok, do you think it is a problem if vcd only supports sync? 1143456018 M * Bertl well, yes and no, depends on how you want that to be used ... 1143456032 M * Bertl any legacy kernel, or new kernel with legacy enabled will use the async 1143456060 M * Bertl so, even if vcd doesn't really support it well, it should at least handle it gracefully 1143456126 M * Hollow hm, tbh i'd like to skip the CONFIG_LEGACY completely and just bail out if legacy is in VCI 1143456135 M * Bertl the simplest thing would be to check for version/vci and just do that 1143456152 M * Hollow ok, fine 1143456155 M * Bertl that's fine for me ... 1143456444 M * Bertl daniel_hozac, Hollow, ensc: here is the first proposal for the vci command, I thought as the vci info is 32bit (atm) the following should be fine, we can extend that in the future if we really need to 1143456452 M * Bertl http://vserver.13thfloor.at/Experimental/delta-vcvci-feat01.diff 1143456505 M * daniel_hozac looks fine. 1143456538 M * Bertl not 100% sure if VERSION is the right category, but I didn't want to make a new one 1143456619 M * daniel_hozac well, they are both primarily used for determining what features kernel has. 1143456641 M * daniel_hozac i think VERSION is fine. 1143456736 M * Bertl okay, good ... 1143457166 J * michal` ~michal@www.rsbac.org 1143457333 M * k3mper when i try to restart init i get init: /dev/initctl: No such file or directory 1143457507 M * daniel_hozac is your guest running an init? 1143457514 M * phreak`` morning Bertl, daniel_hozac :) 1143457629 M * daniel_hozac morning phreak`` :) 1143457642 M * Bertl k3mper: it is _very_ likely that you use sysv init style, which doesn't start a separate init in the guest 1143457671 M * Bertl k3mper: the reboot command usually contacts init via /dev/initctl, which does the shutdown procedures 1143457685 J * meandtheshell ~markus@85-125-227-109.dynamic.xdsl-line.inode.at 1143457702 M * Bertl k3mper: in an init-less guest, you can use reboot -f to accomplish similar (modulo userspace tools and kernel config) 1143457703 M * Hollow Bertl: any comments for the data collector? 1143457711 M * Bertl Hollow: not yet ... sorry 1143457735 M * Hollow ok, no problem, but we should think about a way for collecting all that data in an easy way, so if you have some spare time.. 1143457749 M * Bertl Hollow: I planned to do some testing with rrd and rrdcollect 1143457768 M * Bertl but my test machine is unreachable for some days now .. so that got delayed 1143457770 M * Hollow ok, that's what the data collector will use too, so seems we're heading in the right direction 1143458555 M * Bertl yeah, we should make some tests and see if we can come up with a better interface, if so, we should use that one instead 1143458597 M * daniel_hozac http://phpfi.com/109568 :) 1143458689 M * daniel_hozac http://daniel.hozac.com/vserver/__init__.py for those who want to play around. 1143458762 M * Bertl cool! 1143459105 J * dos000 ~dos000@i216-58-25-69.cybersurf.com 1143459114 M * Bertl welcome dos000! 1143459134 M * dos000 hey Bertl .. you dont sleep much as well hein ? 1143459172 M * k3mper anyone using daemontools + vserver? inittab entry doesnt work :/ 1143459175 M * dos000 i am still fighting with running vhcs (a cpanel like) inside vserver 1143459179 M * Bertl dos000: hehe, well, I had to get up early today ... 1143459202 M * Bertl k3mper: probably the same reason, you do _not_ have an init running 1143459224 M * Bertl k3mper: you want to change your config (the init style) to 'plain' instead of 'sysv' 1143459250 M * Bertl k3mper: this will start an init, which can handle the inittab 1143459257 M * dos000 the guest is querying the localhost. and i am not sure why it does that 1143459267 M * k3mper Bertl: sysv = /etc/init.d/ ? 1143459281 M * daniel_hozac k3mper: yes. 1143459297 M * Bertl sysv init style means that the tools (on the host) will execute the runlevel scripts, but will not start an init inisde the guest 1143459316 M * Bertl dos000: well, what does your /etc/resolv.conf contain? 1143459320 M * k3mper well i want to use sysv + daemontools :) 1143459340 M * Bertl k3mper: sysv or sysv init style? 1143459368 M * Bertl k3mper: look, the difference is precisely this: 1143459374 M * dos000 Bertl, my provider dns server. why is the guest trying to locate localhost ? 1143459390 M * Bertl k3mper: when you start a guest, you can basically decide how that will happen: 1143459407 M * k3mper Bertl: i want to start all entries in /etc/rc2.d + all daemontools services 1143459428 M * Bertl k3mper: a) the init process inside the guest is started, and does whatever it does on a normal system. this is called the 'plain' init style 1143459454 M * Bertl k3mper: b) a shortcut, where no init is required, and the tools 'just' execute the sysv runlevel scripts instead 1143459462 M * Bertl (this one is called sysv init style) 1143459513 M * Bertl k3mper: so your options are: a) use 'plain' init, and be happy, b) write some kind of runlevel script which starts the daemontool stuff (if that is possible) 1143459542 M * Bertl you won't be able to get inittab functionality without init :) 1143459570 M * Bertl dos000: do you have an entry in /etc/hosts ? 1143459655 M * dos000 Bertl, yes .. it is localhost .domain 1143459666 M * dos000 .domain.com in fact 1143459693 M * Bertl maybe add localhost.localdomain there too ... for the resolver 1143459797 M * dos000 umm .. mind tell how 1143459809 M * Bertl vi? 1143459879 M * dos000 i know .. i am tring to figure the syntax localhost.localdomain hostname.domain.com ? 1143459908 M * Bertl yes, something like: localhost localhost.localdomain 1143459920 M * Bertl and if you have a real domain name for that host, add that too 1143460019 M * k3mper Bertl: oki thanks, where do i set it to plain? 1143460075 M * Bertl http://linux-vserver.org/InitStyles 1143460173 M * Bertl i.e. just do echo plain >/etc/vservers//apps/init/style 1143460183 J * lilalinux ~plasma@80.69.35.186 1143460543 Q * shedi Quit: Leaving 1143460754 J * shedi ~siggi@inferno.lhi.is 1143461290 Q * Greek0 Read error: Connection reset by peer 1143464471 J * harti ~hw@83-215-237-5.seek.stat.salzburg-online.at 1143464481 M * Bertl welcome harti! 1143464540 M * harti hi 1143465937 J * Smutje_ ~Smutje@xdsl-87-78-0-141.netcologne.de 1143466043 Q * cehteh Remote host closed the connection 1143466043 Q * Smutje Ping timeout: 480 seconds 1143466043 N * Smutje_ Smutje 1143466325 Q * romke Read error: Connection reset by peer 1143466483 Q * Aiken Ping timeout: 480 seconds 1143469110 J * Viper0482 ~Viper0482@p54975C02.dip.t-dialin.net 1143469386 M * Bertl welcome Viper0482! 1143469440 Q * virtuoso Ping timeout: 480 seconds 1143470196 Q * matta Ping timeout: 480 seconds 1143470504 Q * harti Quit: Leaving 1143471474 J * matta ~matta@c-68-32-239-173.hsd1.pa.comcast.net 1143472967 J * dos000_ ~dos000@i216-58-48-49.cybersurf.com 1143473069 Q * dos000 Read error: Connection reset by peer 1143474261 J * bonbons ~bonbons@83.222.39.180 1143474286 M * Bertl welcome bonbons! 1143474302 M * bonbons Hi Bertl! 1143474332 M * bonbons Bertl, is there somewhere doc about how COW works? 1143474375 M * Bertl not really, what do you want to know? 1143474410 M * restill daniel_hozac: I compiled the new src.rpm. The results are in the same place. Plz wait a couple hours before you kill my cable connections. I am working on something else. 1143474419 M * bonbons As a first step, what limitations it has 1143474505 M * Bertl hmm, what kind of limitations do you think of? 1143474535 M * bonbons e.g. does the copy happen automagically on any write to a file? 1143474538 Q * derjohn Quit: by(t)e 1143474565 M * Bertl bonbons: yes on writes, appends and such, not on attribute/ownership changes yet, though 1143474576 M * Bertl but it should be easy to add that too 1143474629 M * bonbons so when the attributes/ownership change in one guest it also changes in the other guest? Or did I miss-understand something? 1143474645 M * Bertl no, it just gives you an EPERM 1143474798 M * bonbons hm, ok, that's not really the behavior one would want :), I would have thought that attributes and ownership are inode-related and COW works with tagged hard-links; having copy happen on data change 1143474849 M * Bertl attributes _are_ inode related, and it works that way, that's why I said it is quite simple to add it 1143474868 M * Bertl currently we have just no check/break for those changes 1143474888 Q * cryo Ping timeout: 480 seconds 1143474945 M * bonbons so why is it currently rejected with EPERM? 1143475042 M * Bertl look, we have a single check (in open_namei) atm, which 1143475068 M * Bertl when something tries to open the inode for writing, gives an error 1143475090 M * Bertl which, in case of a unified link, will do the break 1143475095 M * Bertl and retry the open 1143475109 M * Bertl a similar check would be required for attributes/perms 1143475127 M * bonbons ok, I see 1143475148 M * Bertl if you are going to test it, probably 30min work 1143475171 M * bonbons yep, willing to give it a try 1143475227 M * Bertl first thing is, get a COW enabled kernel, create an unified file, and try to change the perms 1143475237 M * Bertl of this one, please make me an strace -fF 1143475244 M * bonbons what I would like to try with COW is setup a master guest, then COW-hardlink it for each new required guest, but without attribute change it can be problematic (especially for /etc and /var content) 1143475256 M * Bertl np 1143475276 M * Bertl it's something on my todo list anyways 1143475410 J * doener_ ~doener@i5387C490.versanet.de 1143475509 M * Bertl welcome doener_! 1143475617 J * mkhl ~mkhl@200-153-181-73.dsl.telesp.net.br 1143475650 M * Bertl welcome mkhl! 1143475820 Q * doener Ping timeout: 480 seconds 1143475875 M * blizz there's no loopback by default, right? 1143475946 M * blizz do i have to create it as an ordinary interface in /interfaces? 1143476000 M * bonbons Bertl: does COW need multiple guest, or can it be done host<->host, which file flag is required for COW, IUNLINK? 1143476192 Q * ||Cobra|| Remote host closed the connection 1143476218 M * Bertl bonbons: doesn't require any guest 1143476237 M * Bertl hard links, and --iunlink should be enough 1143476280 M * bonbons fine :) 1143476631 M * bonbons Ok, I did a touch $X, creates a new file named $X\251, on next touch $X\250, ... 1143476644 M * bonbons Most important line of trace: open("dsdt.dat", O_WRONLY|O_NONBLOCK|O_CREAT|O_NOCTTY|O_LARGEFILE, 0666) = -1 EMLINK 1143476702 M * Bertl okay, here we go the EMLINK is there, just no proper action is taken 1143476752 M * bonbons What I don't understand is where the new file comes from (like $X\251) 1143476768 M * Bertl that's part of the cow link breaking 1143476782 M * Bertl what happens if you do chmod instead of touch? 1143476795 M * bonbons just doing it 1143476901 M * Bertl another question, the file has link count>1 yes? 1143476903 M * bonbons chmod g+w just changes the permission on both files, all ops in strace return 0 (stat64, open, chmod, close) 1143476929 M * Bertl ah, careful, you are root/admin now 1143476938 M * Bertl you have to do something like: 1143476944 M * bonbons the file ending in \250 has refcount == 1, dsdt.dat (and it's companion dsdt2.dat) have 2 refs 1143476949 M * Bertl vcontext --create --xid 42 -- .... 1143477000 M * Bertl but the touch is interesting ... 1143477225 M * Bertl jost to verify, the following works for you as expected? 1143477236 M * Bertl echo "test" >/tmp/x 1143477240 M * Bertl ln /tmp/x /tmp/y 1143477243 M * Bertl setattr --iunlink /tmp/y 1143477250 M * Bertl ls -la /tmp/[xy] 1143477255 M * Bertl -rw-r--r-- 2 root root 5 Mar 27 16:32 x 1143477255 M * Bertl -rw-r--r-- 2 root root 5 Mar 27 16:32 y 1143477262 M * Bertl echo "hansi" >/tmp/y 1143477269 M * Bertl ls -la /tmp/[xy] 1143477269 M * Bertl -rw-r--r-- 1 root root 5 Mar 27 16:32 /tmp/x 1143477269 M * Bertl -rw-r--r-- 1 root root 6 Mar 27 16:33 /tmp/y 1143477321 M * Bertl because on touch I get 'permission denied' 1143477544 Q * dos000_ Quit: Leaving 1143477677 J * brc_ bruce@20151209200.user.veloxzone.com.br 1143477689 M * Bertl welcome brc_! 1143477728 Q * brc Ping timeout: 480 seconds 1143478178 M * brc_ bertl! 1143478180 M * brc_ :) 1143478356 M * bonbons Bertl: on you procedure I get Too many links on the echo hansi part and /tmp/y\251 appeared 1143478373 M * Bertl bonbons: filesystem and kernel version? 1143478385 M * bonbons My difference of before: I did set IUNLINK on both files (original and copy) 1143478409 M * bonbons Filesystem: reiserfs, kernel 2.6.16-rc5-vs2.1.1-rc13 1143478451 M * Bertl okay, so reiserfs might have a bug then, interesting 1143478471 M * Bertl could you try with ext2/3 just to verify that? 1143478503 M * bonbons Was just going to try with tmpfs 1143478515 Q * matti Ping timeout: 480 seconds 1143478664 M * bonbons Same effect on tmpfs as on reiserfs 1143478812 M * bonbons and again same effect on ext2 1143479254 M * bonbons on reiserfs, 2.6.16 + rc14 it looks better 1143479501 J * Greek0 ~greek0@85.255.145.201 1143479962 M * bonbons Bertl, I have more insight: the difference commes from the IMMUTABLE attribute. With IMMUTABLE it works, without it doesn't 1143480045 M * bonbons vattr (vserver-utils) just sets IATTR_IUNLINK, setattr (util-vserver) implicitly sets IATTR_IMMUTABLE in addition to IATTR_IUNLINK 1143480622 M * bonbons with both set (immutable and iunlink) files are unlinked on touch, chmod, ... 1143480700 M * bonbons speaking to quickly, touch unlinks, chmod failss with op not permitted 1143480845 M * bonbons but If I have X, Y Immutable+iunlink, touch Y (works fine) I cannot touch X anymore afterwards 1143480932 M * phedny are you working on a kind of copy-on-write thing for shared files? 1143480951 M * bonbons yep, COW is the current subject 1143480964 Q * Bertl Ping timeout: 480 seconds 1143480982 J * ptl_xid ~iuuuju@host207.201-252-53.telecom.net.ar 1143481061 M * ptl_xid Hi, I'd like to know if I have to patch the kernel to use per context disk limits + quotas ??? I'm using 2.6.14.4 + vserver 2.1.0 1143481277 J * liquid3649_ ~Viper0482@p549754EF.dip.t-dialin.net 1143481382 Q * liquid3649_ Quit: 1143481649 M * ptl_xid Hi, I'd like to know if I have to patch the kernel to use per context disk limits + quotas ??? I'm using 2.6.14.4 + vserver 2.1.0 1143481713 Q * Viper0482 Ping timeout: 480 seconds 1143481739 J * Bertl herbert@212.16.62.52 1143481765 M * Bertl great, I'm back now ... 1143481780 A * Bertl reading up ... 1143481844 M * Bertl bonbons: ahem, between rc13 and rc14 is no filesystem related fix IIRC 1143481948 M * ptl_xid Hi !!!. I've recently installed a brand new box, with sarge + 2.6.14.4 + vserver patch 2.1.0. I'd like to have context disk limits + quotas, do I have to apply another kernel patch ? 1143482017 M * Bertl disk limits are default, even on shared partitions, user/group quota are not yet supported on a shared partition 1143482039 M * Bertl (there was no real demand for that yet) 1143482078 M * Bertl bonbons: ah, I see, the COW break code does not check for IS_IMMUTABLE (yet) 1143482093 M * ptl_xid oh... but does it work: http://linux-vserver.org/Standard+non-shared+quota ? 1143482334 M * Bertl yep, on a non shared partition, quota works just fine 1143482421 M * ptl_xid sorry, what is a shared partition ?... its when you do xid tagging ? 1143482435 M * Bertl yep, when you have more than one guest per partition 1143482538 M * ptl_xid ohh... I see. So if you have lvm volumes, and you use 1 vserver / volume, does it count as non shared partition ? 1143482559 M * Bertl yep, that's exactly what is meant here ... 1143482572 M * Bertl you have quota and disk limits on that one if you like 1143482622 M * phedny but there is a limit for 8 of these virtual root devices? 1143482631 M * ptl_xid ok, that's why its not so important to have quota over shared partition... I'm starting to understand ! 1143482645 Q * bragon Ping timeout: 480 seconds 1143482656 M * Bertl phedny: no, the default is 8 devices, the maximum is 256 devices 1143482677 M * phedny that's a kernel config option? 1143482692 M * ptl_xid so, using LVM the xid tagging becomes useless 1143482694 M * phedny or is it just like how many devices I create? 1143482703 M * Bertl phedny: you can select that at kernel boot time if the vroot is compiled into the kernel or when loading the module 1143482713 M * phedny okay, I see :) 1143482731 M * Bertl ptl_xid: well, yes and no, the xid tagging has a lot of advantages over lvm per guest 1143482748 M * ptl_xid oh, for example ? 1143482770 M * Bertl well, if you have, let's say 20 guests, all sarge, then you can save a lot of resources 1143482786 M * Bertl the sarge install will be, let's say 300MB (probably more) 1143482803 M * Bertl where roughly 280MB can be shared between the guests 1143482826 M * ptl_xid ok... I understand 1143482842 M * Bertl so 20 guests with lvm, min 6GB, with xid and uification, 900MB 1143482860 M * Bertl and of course, the memory and caches benefit from that too 1143482892 M * ptl_xid yeah it's much better ... 1143483037 M * ptl_xid thank you Bertl !. Last question: to have cpu limits. Is this supported in the default 2.1.0 ? 1143483093 J * coocoon ~coocoon@p54A056EB.dip.t-dialin.net 1143483099 M * coocoon hello to all 1143483182 M * ptl_xid Hello ! 1143483204 M * Bertl welcome coocoon! 1143483709 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1143483781 J * coocoon ~coocoon@p54A056EB.dip.t-dialin.net 1143483974 J * Viper0482 ~Viper0482@p549754EF.dip.t-dialin.net 1143483993 P * Viper0482 1143484587 J * Viper0482 ~Viper0482@p549754EF.dip.t-dialin.net 1143484601 M * Bertl okay, off for now .. back later ... 1143484606 N * Bertl Bertl_oO 1143484621 P * Viper0482 1143484968 Q * shedi Read error: Connection reset by peer 1143485052 M * ptl_xid Byeee 1143485072 Q * ptl_xid Quit: 1143485823 J * shedi ~siggi@inferno.lhi.is 1143485952 J * Viper0482 ~Viper0482@p549754EF.dip.t-dialin.net 1143485965 P * Viper0482 1143486693 Q * rofel Quit: using sirc version 2.211+KSIRC/1.3.12 1143487773 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1143487859 J * ensc_ ~irc-ensc@p54B4F9C7.dip.t-dialin.net 1143487915 N * ensc_ Guest5131 1143487968 Q * ensc Ping timeout: 480 seconds 1143488574 Q * lilalinux Remote host closed the connection 1143489408 P * meandtheshell 1143490455 Q * eyck Ping timeout: 480 seconds 1143490613 J * eyck ~eyck@195.242.124.92 1143491105 Q * BartVB Quit: Leaving 1143491363 Q * michal` Ping timeout: 480 seconds 1143491702 J * michal` ~michal@www.rsbac.org 1143492191 J * derjohn ~derjohn@80.69.37.19 1143493804 J * coocoon ~coocoon@p54A056EB.dip.t-dialin.net 1143494546 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1143494903 J * Aiken ~james@tooax6-221.dialup.optusnet.com.au 1143496325 M * Hollow Bertl_oO: can you please take a look at http://bugs.gentoo.org/show_bug.cgi?id=127162 1143496473 J * coocoon ~coocoon@p54A056EB.dip.t-dialin.net 1143496653 Q * restill Quit: Leaving 1143497010 Q * bonbons Quit: Leaving 1143497164 J * cryo ~say@psoft.user.matrix.farlep.net 1143497629 J * yarihm ~yarihm@84-74-23-214.dclient.hispeed.ch 1143497905 M * derjohn Hollow, netfilter? 1143497941 M * brc_ Hollow 1143497957 M * brc_ Regarding util-vserver bug, does that mean a high risk ? 1143497960 M * brc_ Util-VServer Unknown Linux Capabilities Vulnerability 1143497961 M * brc_ The util-vserver package for the Linux-VServer project is susceptible to an unknown Linux capability vulnerability. The package fails to properly handle unknown Linux capabilities. 1143497961 M * brc_ The exact consequences of this issue are currently unknown. They depend on the nature of the unknown capabilities and on the nature of the applications that use them. Hosted virtual servers may possibly gain inappropriate access to the hosting operating system. 1143497980 M * derjohn Hollow, a I see: Tim Yamin 1143498040 M * derjohn brc_, where is it from? source? Do they mention a version? 1143498101 M * brc_ Just clsoed the window :( 1143498105 M * brc_ it is on securityfocus 1143498114 M * brc_ it says that the new version fixes this issue. 1143498136 M * derjohn " Hosted virtual servers may possibly gain inappropriate..." .. I am not a kernelman, but this sound very strange. as soon as the guest runs the utils dont do anything.... you could even delete therm 1143498156 M * derjohn well: do the newest version ;) 1143498328 M * brc_ that's what i thought 1143498332 M * brc_ really weird 1143498363 M * derjohn hm, well, sometimes those who explored the hole want to make look bigger ;) 1143498378 M * derjohn but: we let Bertl_oO decide 1143498381 M * derjohn :) 1143498426 M * brc_ hehehe :) 1143498821 J * mef_ ~mef@CPE-61-9-212-131.qld.bigpond.net.au 1143501611 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1143501815 J * coocoon ~coocoon@p54A056EB.dip.t-dialin.net 1143502396 J * mef__ ~mef@137.219.227.196 1143502420 Q * mef__ Read error: Connection reset by peer 1143502436 Q * mef_ Ping timeout: 480 seconds 1143502437 J * mef__ ~mef@137.219.227.196 1143503736 N * Bertl_oO Bertl 1143503743 M * Bertl back now for a few minutes ... 1143503793 M * Bertl Hollow: I see the bug page, but where is the disclosure? i.e. where is the issue described? 1143503837 M * Bertl brc_: what is that 'Unknown Linux Capabilities Vulnerability' about?