1160611358 M * almak ok I think I see what is going on. Thx. 1160611464 M * Bertl you're welcome! 1160614537 M * waldi Fixing visibility of /prKernel unaligned access at TPC[4b724c] filldir64+0x54/0x134 1160614540 M * waldi Kernel unaligned access at TPC[4b7268] filldir64+0x70/0x134 1160614543 M * waldi Kernel unaligned access at TPC[4b7234] filldir64+0x3c/0x134 1160614545 M * waldi Kernel unaligned access at TPC[4b724c] filldir64+0x54/0x134 1160614548 M * waldi Kernel unaligned access at TPC[4b7268] filldir64+0x70/0x134 1160614550 M * waldi yeah 1160614623 M * Bertl cool 1160614640 M * waldi Linux zee.buildserver.net 2.6.18-1-vserver-sparc64 #1 SMP Wed Oct 11 22:57:32 UTC 2006 sparc64 GNU/Linux 1160614680 M * Bertl I always thought that was an alpha oddity ... 1160614696 M * waldi no 1160614706 M * waldi this is a problem on anything except i386 1160614739 M * Bertl even x86_64? 1160614759 M * waldi s/i386/x86/ 1160614782 M * Bertl i.c. but it seems to be handled well on most archs 1160614796 M * Bertl never saw such messages on parisc for example 1160614811 M * waldi yes, the kernel emulates it and on some arches spit 1160614837 M * Bertl ah, so fixup handlers are on all archs, only some of them are 'verbose' enough 1160614858 M * waldi yes 1160614868 M * waldi on arm you can even decide what happens 1160614882 M * waldi spit, ignore, segfault or so 1160614934 M * Bertl cool, if you discover issues (with the alignment) in Linux-VServer code, please let me know 1160614954 M * waldi this seems to be vprocunhide, did not yet check where 1160615007 M * Bertl that might be dietlibc then 1160615014 M * waldi possible 1160615211 M * waldi anyway, I have to sleep 1160615223 M * Bertl right, good night! 1160619269 Q * ensc Killed (NickServ (GHOST command used by ensc_)) 1160619279 J * ensc ~irc-ensc@p54B4D2C5.dip.t-dialin.net 1160622676 J * stefani ~stefani@c-24-19-46-211.hsd1.mn.comcast.net 1160623590 P * stefani parting (is such sweet sorrow) 1160625603 M * Bertl off to bed now .. have a nice one everyone! cya! 1160625611 N * Bertl Bertl_zZ 1160625797 J * marcfi1 ~mef@c-68-39-177-97.hsd1.nj.comcast.net 1160625802 P * marcfi1 1160625864 Q * kir Read error: Connection reset by peer 1160625967 J * kir ~kir@swsoft-mipt-nat.sw.ru 1160630567 Q * Rich_Estill Ping timeout: 480 seconds 1160631543 J * LeanCalPer Administra@200.114.3.36 1160632041 J * Piet hiddenserv@tor.noreply.org 1160633473 J * eric mzmFSox2Us@c-67-190-143-73.hsd1.co.comcast.net 1160633480 Q * eric 1160633497 J * jkl VSv91Us3Vk@c-67-190-143-73.hsd1.co.comcast.net 1160633515 M * jkl does anyone know how to get a usb device visible in a vserver? 1160633913 Q * LeanCalPer Quit: Abandonando 1160633969 N * Piet MyBrotherIs98 1160634005 N * MyBrotherIs98 Piet 1160634063 Q * cdrx Read error: Operation timed out 1160635661 Q * pzYsTorM Ping timeout: 480 seconds 1160638186 J * dna_ ~naucki@p54BCE29F.dip.t-dialin.net 1160638437 J * cdrx ~legoater@242.32.96-84.rev.gaoland.net 1160639426 J * dna___ ~naucki@p54BCF03F.dip.t-dialin.net 1160639806 Q * dna_ Ping timeout: 480 seconds 1160640867 Q * shedi Quit: Leaving 1160640941 J * meandtheshell ~markus@85-124-232-50.work.xdsl-line.inode.at 1160643134 M * matti :) 1160643627 M * h01ger moin 1160643665 M * h01ger jkl, you need to create the device for the vserver (from outside the vserver) first, then you can access it from inside 1160643882 M * h01ger setrlimit limit #11 (or #12 #13 #14) to soft=-1, hard=-1 failed - how do i map those numbers to http://oldwiki.linux-vserver.org/Resource+Limits or better the names in /etc/security/limits.conf? is there somewhere a list? 1160644549 Q * Curus Quit: Client exiting 1160644720 J * lilalinux ~plasma@dslb-084-058-197-116.pools.arcor-ip.net 1160648696 M * h01ger can i run a vserver-kernel in vmware and put vservers in/on it? 1160648782 Q * cdrx Ping timeout: 480 seconds 1160648885 M * cehteh h01ger: likely .. but why? 1160648910 M * cehteh you loose most vserver benefits (minimize resource usage) 1160648971 J * shedi ~siggi@dsl-149-109-85.hive.is 1160649009 M * h01ger cehteh, to test recovery... 1160649028 M * cehteh ah ok 1160649036 M * cehteh h01ger: you might try qemu as well 1160649070 M * h01ger its for someone who is a happy vmware user... 1160649076 M * cehteh imo qemu is in some aspects superior to vmware (easier configuration, commandline/scripting interface) 1160649085 M * cehteh ok 1160649085 M * h01ger & free software 1160650660 J * Piet_ hiddenserv@tor.noreply.org 1160650665 Q * SNy Quit: leaving 1160650756 Q * Piet Remote host closed the connection 1160650758 M * meandtheshell daniel_hozac: how do you generate your signature files for e.g. util-vserver 1160650758 M * meandtheshell "gpg --output file.asc --detach-sign file"? Why not "gpg --output file.sign --detach-sign file"? In short why do you use a .asc extension and not .sign? Just want to know if there's a special reason ... :) 1160650794 J * SNy 716f7f1200@bmx-chemnitz.de 1160651278 Q * phedny Ping timeout: 480 seconds 1160651646 J * cdrx ~legoater@cap31-3-82-227-199-249.fbx.proxad.net 1160652521 J * yarihm ~yarihm@whitehead2.nine.ch 1160652636 M * daniel_hozac meandtheshell: i use --armor as well. 1160652822 M * meandtheshell daniel_hozac: so .asc instead of .sign because you also supply --armor. Is this a convention I never heard about ... hm ... 1160652839 M * daniel_hozac .asc is the default when you use --armor ;) 1160653188 M * meandtheshell daniel_hozac: ahh - I see - of course I always used the --output option as well (which you didn't in this case) and specified a file name so ... 1160656446 Q * glut Ping timeout: 480 seconds 1160658325 J * glut glut@no.suid.pl 1160660680 Q * MrX Ping timeout: 480 seconds 1160660988 J * FireEgl FireEgl@2001:5c0:84dc:1:4::1 1160661277 N * Bertl_zZ Bertl 1160661284 M * Bertl morning folks! 1160661330 M * ray6 morning bertl 1160661392 M * Bertl hey ray6! how is it going? 1160661472 M * Bertl h01ger: only limits listed there with 'U' can be mapped to the /etc/security/limits.conf (I'd say) and please note that there are two sets of limits, ulimits _and_ rlimits 1160661478 M * ray6 fine,thanks :) 1160661701 J * MrX ~urk@219.95.3.191 1160662250 M * h01ger Bertl, ic. where i'm really lost is how to determine which limit #13 or #12 or any number is?... 1160662467 M * Bertl that depends on the architecture 1160662512 M * Bertl see /usr/include/asm/resource.h 1160662568 M * Bertl i.e. they are _not_ ordered by number on that page (because that changes from arch to arch) 1160662952 Q * FireEgl Ping timeout: 480 seconds 1160663093 M * h01ger Bertl, great ("see /usr/include/asm/resource.h") - thanks. but on i386 this list only goes to #11 while i have #12-14 in my logs, too 1160663105 N * dna___ dna_ 1160663132 M * Bertl h01ger: just means that your userspace (glibc/headers) are outdated and the kernel knows more 1160663157 M * Bertl you can peek into the kernel's userspace headers too 1160663174 M * harry mkay! 1160663185 M * harry the namespace cleanup stuff thingy is b0rK3d 1160663185 M * Bertl (e.g. make headers_install in the kernel tree) 1160663204 M * Bertl harry: hmm, really? or just not configured? 1160663233 M * harry i removed /etc/vservers/.default/namespace-cleanup 1160663237 M * Bertl i.e. daniel_hozac is working on a config file which excludes certain pathes, that should work for you, no? 1160663238 M * harry and now the new tools work 1160663275 M * Bertl yeah, not unexpected, the /usr is not handled as we figured 1160663283 M * harry uhu 1160663367 M * Bertl I'd say, the next version will work quite fine with the namespace cleanup 1160663376 M * Bertl (if configured properly of course) 1160663389 M * harry do you have to configure it?? 1160663427 A * harry looks @ cleanup code 1160663436 M * harry and what exactly it does/is supposed to do 1160663551 M * h01ger Bertl, /usr/src/linux-headers-2.6.16-2-vserver-686/include/asm-i386/resource.h tells me to look into ../asm-generic/resource.h :-) so the numbers _can_ be added to the page, right? 1160663552 M * Bertl harry: you will have to wait for a new release (or cvs/svn), the configuration is hardcoded atm 1160663570 M * h01ger .oO( if not "Note that the contents has been frozen and editing is disabled on all pages."... ) 1160663622 M * harry Bertl: so what's best to do for now? use the old tools, or the new, without namespace-cleanp? 1160663644 M * daniel_hozac harry: just don't enable namespace-cleanup. 1160663664 M * Bertl sorry, no, just grep/search for the following: RLIMIT_AS 1160663664 M * harry daniel_hozac: that's what i did for now ;) 1160663674 M * Bertl h01ger: sorry, no, just grep/search for the following: RLIMIT_AS 1160663680 M * Bertl include/asm-mips/resource.h:#define RLIMIT_AS6 1160663686 M * Bertl include/asm-generic/resource.h:# define RLIMIT_AS9 1160663691 M * Bertl include/asm-alpha/resource.h:#define RLIMIT_AS7 1160663722 M * h01ger Bertl, ah, ok. i386=generic, but this is not true for every arch... 1160663727 M * daniel_hozac harry: there's a reason it's not enabled by default. 1160663940 M * harry daniel_hozac: can i help? 1160663945 M * daniel_hozac with? 1160663952 M * harry fixing the problem 1160663965 M * daniel_hozac i suppose you could test the fix. 1160664012 M * daniel_hozac http://people.linux-vserver.org/~dhozac/p/uv/experimental/ns-cleanup.patch 1160664295 M * harry daniel_hozac: does that correctly deal with other mounts etc??? 1160664313 M * daniel_hozac hmm? 1160664328 M * harry like, i have a mount in: /vservsers/ftpserv => /vservers/ftpserv/home => /vservers/ftpserv/home/ftp/incoming 1160664441 M * daniel_hozac ok, and? 1160664493 M * harry will it try to unmount /vservers/ftpserv/home/ftp/incoming and /vservers/ftpserv/home ? 1160664500 M * harry (like it does now) 1160664517 M * daniel_hozac when you're starting another guest, it should, yes. 1160664729 M * harry gandalf:/usr/local/etc/vservers/.defaults# vserver stdserver start 1160664729 M * harry /usr/local/sbin/vserver: line 486: /usr/local/sbin/vserver-info: No such file or directory 1160664732 M * harry /usr/local/sbin/vserver: line 655: pushd: /usr/local/etc/vservers/stdserver/vdir: No such file or directory 1160664735 M * harry doesn't work 1160664767 M * daniel_hozac and you did apply the patch? 1160664780 M * harry sure did 1160664806 M * harry i don't have to recompile stuff, do i? 1160664813 M * harry (since they are just scripts) 1160664841 M * harry grmbl... 1160664844 M * daniel_hozac did you install it? 1160664855 M * harry yes 1160664861 M * harry patch => make install 1160664864 M * daniel_hozac (patch -p1 -d/usr/lib*/util-vserver < ~/ns-cleanup.patch should've patched your installed copy) 1160664937 M * harry uhu.. doesn't work :) 1160664988 M * daniel_hozac shit, i'm stupid. your errors are weird though. 1160664997 M * daniel_hozac line 1190, change "$i" to "$path". 1160665037 M * harry gandalf:/usr/src/util-vserver-0.30.211# vserver stdserver start 1160665037 M * harry umount: 0: not found 1160665037 M * harry umount: 0: not found 1160665041 M * harry but it starts allright... 1160665057 M * daniel_hozac after changing it to $path? 1160665068 M * harry yes 1160665075 M * daniel_hozac that's, umm, strange. 1160665090 M * harry i don't know... you unmount a lot of stuff, right 1160665101 M * harry what if you umount /vservsers/ftpserv first 1160665111 M * harry then unmounting /vservers/ftpserv/home and /vservers/ftpserv/home/ftp/incoming will be really hard 1160665133 M * daniel_hozac but those mounts should keep the parent mount busy. 1160665145 M * harry true 1160665163 M * daniel_hozac anyway, could you add an echo right about the UMOUNT to see what it's unmounting? 1160665175 J * phedny ~mark@volcano.p-bierman.nl 1160665300 M * harry http://paste.linux-vserver.org/478 1160665350 M * daniel_hozac what does cat /proc/mounts say on the host? 1160665385 M * harry http://paste.linux-vserver.org/479 1160665655 M * daniel_hozac hmm, very strange... 1160665679 M * harry i don't really get the logic behind the cleanup stuff ;) 1160665701 M * daniel_hozac could you do echo "d=$dev p=$path o=$opts" right after the while read dev path opts; do line? 1160665720 M * daniel_hozac it's basically to keep a minimum amount of mounts in the guest's namespace. 1160665728 M * daniel_hozac you don't need to have the other guests available in there, for instance. 1160665794 M * harry http://paste.linux-vserver.org/480 1160665819 M * daniel_hozac so it looks like we're losing input, somehow. 1160665869 M * harry indeed... extremely odd 1160666194 M * transacid hey! does anyone have experience with vserver guest and host running different distributions? I have a debian host system and want to run a gentoo vserver. I took the vserver-stage3 and build the system in chroot as far as possible. I think i set up all right. but the vserver won't start. here is the error message: http://phpfi.com/162895 Any ideas? 1160666242 M * Bertl first, different distros for guest and host are not a problem at all 1160666255 M * daniel_hozac transacid: you forgot to set the initstyle to plain. 1160666259 M * daniel_hozac which is required for Gentoo guests. 1160666270 M * transacid daniel_hozac: how do i do this? 1160666339 M * daniel_hozac --initstyle plain when building, or echo plain > /etc/vservers/.../apps/init/style 1160666355 M * transacid ah ok i try 1160666417 M * transacid yeah great! thx a lot 1160666529 A * harry doesn't get it... how can that read just... lose information 1160666577 M * harry maybe we can change it with: for line in `cat /proc/mounts`; do 1160666579 M * harry etc... 1160666837 M * daniel_hozac it seems very strange indeed... i suppose we could use a temporary copy or somesuch. 1160666957 M * harry the biggest q is: WHY is it going wrong? 1160666984 M * daniel_hozac i guess reading from /proc/mounts while unmounting lots of things might be kind of racy. 1160667015 M * harry maybe better to read it all into 1 array, and then, unmounting 1160667085 Q * yarihm Quit: This computer has gone to sleep 1160667094 M * Bertl A=( `cat /proc/mounts` ) ? 1160667120 M * daniel_hozac $(< /proc/mounts ) is what's used elsewhere in util-vserver :) 1160667148 M * harry smth like that, yes 1160667160 M * Bertl and ${A[n*6+c]} 1160667187 M * harry perl!!! we need perl!!! :) 1160667221 M * daniel_hozac i guess just mktemp + cat would suffice as well. 1160667288 M * Bertl probably 1160667399 M * daniel_hozac maybe that's why the unmounting was in a separate loop before... 1160667405 M * harry daniel_hozac: do i implement? or do you? :) 1160667418 M * harry probably :) 1160667428 M * daniel_hozac try reverting part of the patch that moved umount into the read loop. 1160667513 M * harry you try, or should i? ;) 1160667534 M * daniel_hozac refresh http://people.linux-vserver.org/~dhozac/p/uv/experimental/ns-cleanup.patch 1160667559 M * harry wiiiiii 1160667772 M * harry now i still get this: 1160667773 M * harry gandalf:/usr/local/lib/util-vserver# vserver stdserver start 1160667773 M * harry umount: /vservers/ftpserv/home: not found 1160667773 M * harry umount: /vservers/ftpserv/home: not found 1160667773 M * harry umount: /vservers/ftpserv/home/ftp/upload: not found 1160667775 M * harry umount: /vservers/ftpserv/home/ftp/upload: not found 1160667822 M * harry i don't get it 1160667828 M * daniel_hozac so am i wrong in thinking that /vsevers/ftpserv should still be around because those mounts are there? or is umount recursive by default? 1160667856 M * harry i don't think umount is recursive by default 1160667867 M * harry quite sure even... 1160667893 M * harry but... i think there is another problem... of the namespace not being aware/able to see that it's a recursive mount 1160668011 M * harry it seems like you're able to unmount stuff which is still mounted in another namespace 1160668020 M * harry well... recursively 1160668024 M * harry you know what i mean :0 1160668026 M * harry :) 1160668060 M * daniel_hozac so vnamespace -e stdserver cat /proc/mounts doesn't show the ftpserv/home... mounts? 1160668086 M * harry nope 1160668105 M * daniel_hozac strange... 1160668123 M * harry http://paste.linux-vserver.org/481 1160668140 M * Bertl daniel_hozac: maybe --bind mounts? 1160668207 M * daniel_hozac i don't see any bind mounts in http://paste.linux-vserver.org/479, seems to be separate LVs for everything. 1160668208 J * Heinzwurst ~i@dsl46-177.pool.bitel.net 1160668220 M * Bertl welcome Heinzwurst! 1160668231 M * Heinzwurst Hi Bertl! 1160668247 M * Heinzwurst nice homepage after that relaunch .... 1160668341 M * Bertl yeah, like it? 1160668344 M * SNy Nice nick. d; 1160668375 M * Heinzwurst BUT (there is ever a but) .... I still don't find help to my probs ... :/ 1160668442 M * Bertl let's hear .. 1160668453 M * Heinzwurst and I'm an absolute N00b 1160668455 M * Heinzwurst ;) 1160668456 M * Bertl (not everything is there yet, still a lot of stuff missing) 1160668492 M * Heinzwurst earlier I could start a vserver with "vserver name enter" and starting up some process .... < if nothing else helps 1160668526 M * Heinzwurst now I updated to version 2.6.17.13-vs2.0.2.1 and can't get the vserver starting up 1160668532 M * Bertl you still can do that, but it it is not advised to do so (so we made it a little harder :) 1160668552 M * Heinzwurst ahhhh :) ... how is it still possible? 1160668552 M * Bertl what tools do you use? 1160668568 M * Heinzwurst util-vserver-0.30.211 1160668579 M * Heinzwurst Guest/Host are Gentoo Systems 1160668588 M * Bertl good, what errors do you get when you do a 'normal' vserver start ? 1160668614 M * Heinzwurst I tried a lot with /etc/vservers/name/apps/init stuff .... cmd.start etc but don't get it 1160668646 M * Heinzwurst nothing. no errors ... just a new command line 1160668667 M * Bertl so maybe it started quite fine then? have you checked with vserver-stat? 1160668691 M * Heinzwurst hey, i'm not such a big noob ... ;) 1160668706 M * Heinzwurst no vserver running .... 1160668740 M * Heinzwurst I don't understand the conf of apps/init ..... what to use for gentoo-guests 1160668771 M * Bertl okay, could you upload the output of 'vserver --debug start' to paste.linux-vserver.org? 1160668782 M * Heinzwurst I tried style plain + sysv .... but there is nothing in docs that shows the meaning ... and also in the source code I don't understand the context 1160668793 M * Bertl plain is the style for gentoo right now 1160668817 M * Bertl plain means that an init process will be started, which takes care of starting the services 1160668835 M * Bertl sysv means that the vserver command will call the runlevel scripts (sysv) 1160668993 M * Heinzwurst ok ... paste done ... 1160669089 M * Heinzwurst I tried cmd.start with /sbin/rc default .... which worked some days ago .... but today I don't get it started 1160669152 M * Bertl no, you should not have to add any cmd.* files/entries by default 1160669163 M * Bertl they will just confuse the fragile gentoo startup scheme 1160669220 M * Bertl currently the wrong init style is selected (sysv) as it seems 1160669264 M * Heinzwurst ok .... after deleting the cmd.* files it's working now ..... hhmmnnn 1160669303 M * Heinzwurst how can I get the output of the init process to watch the vserver-starting-up-sequence .... that worked with sysv-style 1160669337 M * Bertl no easy way atm, you can add a fake /dev/console which points to the current terminal 1160669353 M * Bertl the plain init style logs only inside the guest 1160669360 M * Bertl (by default) 1160669414 M * Heinzwurst yeah .... and after using sysv I had the output again and changed some stuff (cmd.start) to get it starting .... that worked some days ago with maybe other util-vserver version 1160669479 M * Bertl no, the problem with sysv is, the gentoo dependancy system doesn't work with that properly (at least I was told so by gentoo folks) 1160669518 M * Bertl there was a special 'gentoo' init style which would have been the pendant to the sysv (for gentoo) but gentoo folks dropped that at some point 1160669535 M * Heinzwurst yes, i found that in the bash-sources ... ;) 1160669586 M * Heinzwurst ok ... thanks for your help .... I guess have to live without the start-log now ... ;) 1160669610 M * Bertl well, you can bug the gentoo folks to revive something like the 'gentoo' init style :) 1160669642 M * Bertl I still think it would be 'nice to have' 1160669646 J * bronson ~bronson@adsl-64-161-106-11.dsl.snfc21.pacbell.net 1160669663 M * Bertl especially as you can save a process (the init) 1160669666 M * Bertl welcome bronson! 1160669668 M * Heinzwurst I've already to bug the because of lots of other stuff .... 1160669710 M * Heinzwurst them < that is 1160669921 J * bonbons ~bonbons@83.222.36.111 1160670130 M * ensc Bertl: hi, are there some problems with the scheduler in patch-2.6.18-vs2.0.2.1-t8.diff? Processes in ctx are higher prioritized than e.g. host's X which moves a software mouse cursor 1160670138 M * Bertl ah, finally finished with my taxes ... so now back to usefull stuff 1160670160 M * Bertl ensc: nope, it is not the priority 1160670181 M * Bertl ensc: we are investigating this 'phenomenon' well we added the schedmon interface 1160670207 M * Bertl ensc: chances are good that it is some kind of hidden bug in the scheduler code 1160670225 M * ensc ok; some workaround? 1160670243 M * Bertl none known yet, as we don't know _what_ actually happens 1160670267 M * Bertl i.e. we need more input and/or a good idea what happens :) 1160670351 M * Bertl btw, you say 2.6.18-*, it doesn't happen with 2.6.17.x? 1160670525 M * ensc no; did not happened with 2.6.17.11 + vs2.02 1160670528 P * gluk So Long, and Thanks for All the Fish! 1160670565 M * Bertl hmm, interesting .. tx for the hint! 1160670926 M * bronson does linux-vserver work with aliased IP addresses? 1160670937 M * Bertl yes, if you configure them 1160670951 M * bronson chbind --ip eth0:2 echo hi -- Invalid IP number or host name 'eth0:2' 1160670959 M * bronson But the interface is up. 1160670968 M * Bertl no, you do not specify interfaces, just ips 1160670982 M * Bertl so --ip 10.0.1.1 (e.g.) 1160670998 M * bronson Ah, better correct that in the manpage. ;) 1160671003 M * bronson OPTIONS --ip ip_number/interface 1160671054 M * Bertl didn't even know that was there :) 1160671061 M * bronson chbind --ip 64.161.106.11 echo hi --- chbind: vc_net_create(): Invalid argument 1160671078 M * bronson But I can ping that IP addr just fine. 1160671136 M * Bertl depending on the kernel and config, try with --nid 42 1160671158 M * Bertl chbind --nid 42 --ip 64.161.106.11 -- echo hi 1160671182 M * Bertl btw, the man pages seem outdated to some extend, daniel_hozac? 1160671306 M * bronson Bertl: that works! Thanks. 1160671316 M * bronson um... how do I tell vserver-start to use that? 1160671432 M * Bertl it uses that by default, well not the 42 unless your context id is 42 1160671459 M * Bertl you can specify a separate nid too, by default the context id is used 1160671536 Q * shedi Quit: Leaving 1160671792 M * jkl good morning! 1160671804 M * Bertl good morning jkl! 1160671842 M * jkl how are you today? 1160671858 M * Bertl better, thanks! and you? 1160671875 M * jkl tired :/ 1160671880 M * jkl but good! 1160671913 M * jkl h01ger mentioned that if i need access to a usb device within a vserver I need to create it outside the vserver first 1160671947 A * h01ger waves 1160671951 M * jkl oh hello! 1160671966 M * jkl h01ger: how does one do that? 1160671992 M * bronson jkl, I'd guess mknod? 1160672001 M * Bertl hmm, well, usually the hotplug helpers (and udev) do that for you when the device is plugged in 1160672021 M * jkl yes i can see /dev/usb/lp0 just fine on the host 1160672022 M * Bertl i.e. it kind of 'magically' appears 1160672049 M * Bertl so then you can 'simply' copy that into the guest too, no? 1160672049 M * matled is it possible to disallow new hardlinks to device nodes inside a vserver? 1160672071 M * bronson Bertl: doesn't seem to be working for me... vserver hydra start -- Cannot find device "eth0:2" chbind: vc_net_create(): Invalid argument 1160672077 M * Bertl matled: I don't think device nodes can be hard-linked ... 1160672105 M * matled well 1160672105 M * Bertl bronson: what does your interfaces config look like? 1160672118 M * matled mh 1160672121 M * bronson Bertl: I did create this machine waith the wrong device, but I fixed it in ./interfaces/0/dev 1160672133 M * Bertl bronson: and what context do you use for the guest? 1160672155 M * matled you can create hardlinks (e.g. ln lp0 lp1 creates lp1 with the same major/minor id as lp0) 1160672158 M * jkl literally copy it? it seems to be hung trying to do so 1160672181 M * matled a mechanism to give a vserver access to a device would be nice, but also having a way to remove the device again 1160672182 M * bronson ...erm, context? I don't know. That wasn't in the howto I followed. 1160672194 M * Bertl matled: right, that works, didn't know that yet :) 1160672194 M * bronson Is there an easy way for me to check? 1160672232 M * Bertl should be in /etc/vservers//context 1160672255 M * Bertl (if it is missing, choose a context id between 2 and 49151 and add it there) 1160672295 M * matled there is a bind-mount ro patch, isn't it? perhaps one could mount /dev in the vserver using bind with ro 1160672306 M * Bertl bronson: did youw rite eth0:2 into the 'dev' file? 1160672310 M * Bertl *write 1160672318 M * matled I dunno if this causes any problems, but it should disallow renaming/copying 1160672337 M * Bertl matled: yes, you could make /dev readonly with that 1160672354 M * Bertl (just needs an entry in the fstab file) 1160672577 M * jkl bertl: so i should be able to copy /dev/usb/lp0 to /vservers/X/dev/usb ? 1160672592 M * Bertl yes, cp -va will do that 1160672622 M * jkl aaah, -va 1160672656 M * bronson Bertl, cat /etc/vservers/hydra/interfaces/0/dev -- eth0:2 1160672661 M * bronson Is that what you mean? 1160672747 M * bronson Gotta go to work... Back in a bit. 1160672752 Q * bronson Quit: Ex-Chat 1160672971 J * s0undt3ch_ ~s0undt3ch@81.193.56.135 1160673362 J * shuri ~shuri@64.235.209.226 1160673366 M * shuri hello 1160673376 M * Bertl hey shuri! LTNS! 1160673383 M * shuri hey Bertl 1160673384 M * shuri !!! 1160673393 M * shuri i got a strange problem today 1160673398 M * shuri tiwh only one vserver 1160673401 Q * s0undt3ch Ping timeout: 480 seconds 1160673401 M * shuri vcontext: execvp("/etc/init.d/rc"): No such file or directory 1160673408 M * shuri when vserver name start 1160673455 M * Bertl well, I see a bunch of options there 1160673467 M * Bertl - mispaced barrier (on vserver dir) 1160673467 M * shuri vserver voiceip-1.2 stop 1160673471 M * shuri vserver 'voiceip-1.2' is not running 1160673483 M * Bertl - missing script/dependancies (libraries/bash) 1160673501 M * Bertl - old tools with the 'env' issue 1160673520 M * shuri up 111 days 1160673537 M * shuri it was working and all other vserver is working fine stop and start 1160673547 M * Bertl so something changed then 1160673551 M * shuri about mispaced barrier ? 1160673552 M * Bertl reiserfs? 1160673559 M * shuri non ext3 1160673563 M * shuri no ext3 1160673588 M * Bertl what did you change rencently there? 1160673596 M * shuri nothing! 1160673601 Q * s0undt3ch_ Ping timeout: 480 seconds 1160673609 M * Bertl then you have a hardware problem 1160673619 M * shuri hummm 1160673627 M * shuri but it happen only with one vserver... 1160673641 M * shuri i got 10 vserver running 1160673647 M * shuri that i can satrt and stop 1160673653 M * Bertl well, if nothing was changed, it must be a hardware issue, no? 1160673661 M * Bertl i.e. silent disk corruption or so 1160673689 M * shuri maybe hachink... 1160673715 M * Bertl well, yes, somebody could have messed with the guest (from inside too) 1160673722 M * shuri ok 1160673724 M * Bertl would be enough to wipe out the bash 1160673730 J * stefani ~stefani@tsipoor.banerian.org 1160673740 M * Bertl welcome stefani! 1160673745 M * shuri bash is ther 1160673747 M * shuri bash is there 1160673755 M * stefani bash is everywhere 1160673769 M * Bertl but to rule out the obvious causes, please double check the barrier :) 1160673777 M * shuri how? 1160673808 M * Bertl showattr -d /vservers/voiceip-1.2 /vservers/voiceip-1.2/.. 1160673824 M * Bertl (adjsut the /vservers to your guest's location) 1160673830 M * shuri ---buiX /vservers/voiceip-1.2 1160673830 M * shuri ---buiX /vservers/voiceip-1.2/. 1160673846 M * Bertl .. instead of . 1160673864 M * Bertl but looks good so far 1160673885 M * shuri ok 1160673896 M * meandtheshell Bertl: is there a place where one can reach all 1160673896 M * meandtheshell - GnuPG key ID's 1160673896 M * meandtheshell - keyserver(s) 1160673896 M * meandtheshell - key fingerprints 1160673896 M * meandtheshell of all good guys listed in people.linux-vserver.org? 1160673923 M * Bertl meandtheshell: good question, I usually use the MIT server 1160673924 J * s0undt3ch ~s0undt3ch@81.193.56.135 1160674004 M * meandtheshell Bertl: I see - you know I just thougt of it because I download stuff here and there and searching for keys, fingerprints is really messy if everybody has it's own thing going on :) 1160674090 M * Bertl maybe we could add the fingerprints there (and/or the key ids) 1160674183 M * meandtheshell Bertl: that would be great if http://linux-vserver.org/Developers folks could find key ids and fingerprints there - the keyserver URL should be there as well 1160674198 M * meandtheshell *if on 1160674282 A * meandtheshell writes very confusing sentences today - omg 1160674307 M * Bertl np, please send me a pgp signled email to remind me to do that :) 1160674312 M * Bertl *signed 1160674363 M * meandtheshell Bertl: sure ... 1160674554 M * doener evening! 1160674558 M * meandtheshell Bertl: hm ... do I really need to send an email since the content would barely a copy paste of the last 20 lines writen to channel :) 1160674565 M * doener lost connection again yesterday... :( 1160674566 M * meandtheshell doener: evening too :) 1160674578 M * doener my mother's line is really flaky lately 1160675023 Q * shuri Remote host closed the connection 1160675488 M * meandtheshell daniel_hozac: where can I get your public pgp key? keyserver URL? I already got the key ID and fingerprint from mail attachments at Vserver@list.linux-vserver.org. I need the public key in order to verify the util-vserver sources. Would be great if you could add that info http://linux-vserver.org/Developers too for other folks. TIA 1160675544 J * bronson ~bronson@adsl-64-161-106-11.dsl.snfc21.pacbell.net 1160675641 M * daniel_hozac meandtheshell: my release email to the mailing list included the info, it's on pgp.mit.edu. 1160675672 M * meandtheshell daniel_hozac: sure - all but the keyserver URL :) thank you 1160677345 Q * bubulak Ping timeout: 480 seconds 1160677449 J * bubulak ~bubulak@whisky.pendo.sk 1160677471 M * Bertl ensc: could you provide your kernel .config somewhere? 1160677483 M * Bertl (for the 2.6.18 with issues, that is) 1160677582 M * ensc ensc: http://ensc.de/kernel-ensc-pc-config 1160677591 M * Bertl tx 1160677681 M * ensc I can not measure the effect from remote; only the slow mouse cursor indicates a problem 1160677697 M * ensc (and it should be a hard-, not software mouse cursor) 1160677704 M * Bertl yes, it is quite funny, it seems like the mouse is the only affected 1160677728 M * Bertl but I can (somewhat) reproduce it here on my powerbook too 1160677732 M * ensc fwiw, it's a 01:00.0 VGA compatible controller: ATI Technologies Inc Radeon RV100 QY [Radeon 7000/VE] 1160677751 M * Bertl ATI here too, the funny thing is this: 1160677767 M * Bertl - start a CPU hog on the host, everything is fine 1160677788 M * Bertl - start the same hog with chcontext in a context and the mouse is almost unusable 1160677802 M * Bertl - keyboard input in X is completely unaffected though 1160677839 M * ensc yes 1160677844 M * Bertl I don't see any difference in priorities or so 1160677858 M * ensc I see a difference in 'chcontext sh -c 'while true; do : ; done' 1160677871 M * ensc and 'while true; do : ; done' 1160677873 M * ensc on host 1160677887 M * ensc host gets 95%, ctx only 5% load in 'vtop' 1160677896 M * ensc with 2.6.17, both have 50% 1160677897 M * Bertl yeah? different priorities? 1160677912 Q * Piet_ Remote host closed the connection 1160678044 J * Piet_ hiddenserv@tor.noreply.org 1160678255 M * Bertl I think it could be caused by the scheduler changes, which have a number of special cases for single processes and weighting (which might get wrong when the process 'disappears' in a context 1160678511 M * ensc mmh... ctx should be created with same priorities on 2.6.17 and 2.6.18 1160678609 M * ensc I see it with a single process (tar/zcat or gcc) 1160678637 M * ensc the shell-command is a single process, too 1160680070 Q * Piet_ Ping timeout: 480 seconds 1160680677 J * Piet_ hiddenserv@tor.noreply.org 1160680943 M * harry daniel_hozac: any ideas on the error? 1160680950 M * harry well... on the unmount problem? 1160680986 M * daniel_hozac harry: which one? 1160680998 M * daniel_hozac oh, the seemingly recursive unmount? no. 1160680998 M * harry from ftpserv/home etc... 1160681001 M * harry yes 1160681009 M * harry any ideas on how we can troubleshoot it? 1160681079 M * daniel_hozac not really... seems very strange. 1160681092 M * daniel_hozac i assume you can reproduce it with vnamespace -n bash? 1160681103 M * harry ? 1160681141 M * daniel_hozac if you create a new namespace, and execute umount -l -n /vservers/ftpserv, does it work? 1160681157 M * Bertl does X request/have realtime priorities? 1160681171 M * harry yes 1160681181 M * daniel_hozac meaning it hangs around? 1160681191 M * harry and it seemingly unmounts the recursive directories 1160681225 M * harry http://paste.linux-vserver.org/483 1160681228 M * harry check it out :) 1160681251 M * daniel_hozac _very_ strange... but at least it's easily reproducible. 1160681260 M * daniel_hozac how about if you don't use -l? 1160681269 M * harry i think it's because the filesystem isn't mounted/used in that namespace, that it's able to unmount it 1160681289 M * harry gandalf:~# umount -n /vservers/ftpserv 1160681289 M * harry umount: /vservers/ftpserv: device is busy 1160681289 M * harry umount: /vservers/ftpserv: device is busy 1160681292 M * harry then it doesn't work 1160681302 M * daniel_hozac so it's the lazy unmount that causes it... good. 1160681337 M * harry uhu 1160681577 M * harry daniel_hozac: the biggest q is... why is this possible? 1160681585 M * harry this shouldn't happen imho 1160681599 M * harry and, if it's solved, how are we gonna fix the recursive mounts "order" ? 1160681609 M * harry we need to "sort" the unmounts too in that case 1160681621 M * daniel_hozac well, it actually looks like this might be expected behaviour with lazy unmounting. 1160681640 M * harry so lazy unmount is doing recursive mounts? 1160681708 M * daniel_hozac "Perform a lazy unmount: make the mount point unavailable for new accesses" so i suppose, yes. 1160681744 M * harry http://paste.linux-vserver.org/484 1160681750 M * harry strace from an umount -l 1160681776 M * harry http://paste.linux-vserver.org/485 1160681788 M * harry strace from a not lazy umount 1160681838 M * daniel_hozac hmm, strange. 1160681854 M * daniel_hozac no MNT_DETACH in the flags. 1160681872 J * yarihm ~yarihm@84-75-123-221.dclient.hispeed.ch 1160682028 Q * weasel Quit: leaving 1160682044 J * weasel weasel@asteria.debian.or.at 1160682099 Q * Piet_ Remote host closed the connection 1160682250 J * Piet_ hiddenserv@tor.noreply.org 1160682341 M * daniel_hozac hmm, looks like as long as there are no busy mounts, sys_umount will always unmount the entire tree. 1160682384 M * daniel_hozac or am i misinterpreting it? 1160682481 Q * yarihm Quit: Leaving 1160682533 M * harry they certainly arent busy in that namespace :) 1160682549 M * daniel_hozac right. 1160682552 M * Bertl daniel_hozac: would make sense 1160682583 M * harry daniel_hozac: i think it might be a good idea to reverse the /proc/mounts 1160682587 M * Bertl especially as they are 'just' bind mounts 1160682590 M * harry since that allways solves the problem :) 1160682591 M * daniel_hozac i was thinking that. 1160682616 M * harry or just reverse the unmount forloop 1160682619 M * harry BUT... comment it ;) 1160682650 M * daniel_hozac well, i was more thinking along the lines of adding entries at the beginning of the list_umount array :) 1160682669 M * harry the race aswell as the "reversing" 1160682677 M * harry daniel_hozac: mkay, 1160682684 M * harry same result :) 1160682700 M * harry make it an unmount stack! ;) 1160682709 M * harry push/pop unmounts!! 1160682717 M * daniel_hozac well, that's basically what it is :) 1160682732 M * harry yeah, i know, just wanted to signal some buzzwords here :) 1160682765 M * Bertl is 'showstopper' still a buzz word nowadays? 1160682775 M * harry Bertl: naaaaaah :) 1160683720 Q * bonbons Quit: Leaving 1160684241 M * gdm hi all 1160684281 M * Bertl ensc: do you know if you can observe the same behaviour with the hard scheduler disabled? I would think so 1160684324 M * gdm Bertl: i just got this error: http://paste.linux-vserver.org/486 1160684336 M * gdm and i can still enter the vserver afterwards 1160684342 M * gdm do you (or anyone) have any ideas? 1160684404 J * yarihm ~yarihm@84-75-123-221.dclient.hispeed.ch 1160684424 M * Bertl gdm: your guest could benefit from a cleanup, but recent tools should fix that IIRC 1160684471 M * gdm Bertl: how do you mean? 1160684481 M * gdm Bertl: that is, really,, a very new guest 1160684492 M * gdm it hasn't had that much done to it at all 1160684502 M * Bertl well, the guest seems to execute hardware related stuff 1160684511 M * Bertl i.e. unmount and hwclock and such 1160684527 M * gdm oh, ok 1160684534 M * gdm so maybe that is what the problem is... 1160684540 M * gdm it seems to have timed out actually 1160684543 M * Bertl but the fact that the reboot -f inside kills the shutdown script 1160684549 M * gdm immediately, vserver-stat showed that it was there 1160684554 M * gdm but a few minutes later, it doesn't any more 1160684558 M * Bertl is IIRC, something which should be fixed in 0.30.211 1160684614 M * Bertl if not, please file a bugreport to daniel_hozac 1160684626 M * gdm ok, am still running 0.30.210 1160684631 M * gdm so will have a check for that 1160684635 M * gdm otherwise i will 1160684638 M * gdm thanks for looking :) 1160684644 M * Bertl you're welcome! 1160684654 M * Bertl gdm: thanks for reporting! 1160684673 M * gdm no probs - it's great working with ppl who are so responsive :) 1160684826 Q * h01ger Quit: h01ger 1160684864 J * h01ger ~holger@socket.layer-acht.org 1160684899 M * Bertl daniel_hozac: do you know, is sched_prio default for chcontext? 1160684975 M * daniel_hozac i don't think so. 1160685176 M * daniel_hozac IMHO state_init is the only flag enabled by a simple chcontext. 1160685252 M * Bertl yup, tx 1160685698 Q * cdrx Quit: Leaving 1160686252 J * shedi ~siggi@inferno.lhi.is 1160686348 J * virtuoso_ ~s0t0na@shisha.spb.ru 1160686371 Q * virtuoso Read error: Connection reset by peer 1160688831 J * derjohn2 ~aj@dslb-084-059-002-253.pools.arcor-ip.net 1160688844 J * mire__ ~mire@111-167-222-85.adsl.verat.net 1160689084 P * stefani I'm Parting (the water) 1160689092 Q * mire Ping timeout: 480 seconds 1160689134 J * Aiken ~james@tooax6-147.dialup.optusnet.com.au 1160689206 Q * mire__ Quit: Leaving 1160689219 Q * lilalinux Quit: Leaving 1160689268 M * Bertl evening derjohn2! Aiken! 1160689309 M * derjohn2 Bertl, nice to see you! I seems i misseed rc36, rc37 and rc38 ... more than rebaseing? 1160689323 A * derjohn2 was in chianti for some days 1160689398 J * DreamerC_ ~dreamerc@59-115-48-131.dynamic.hinet.net 1160689452 M * Bertl derjohn2: yep, fixes and enhancements 1160689478 Q * almak Remote host closed the connection 1160689529 M * derjohn2 Bertl, this chnagelog-writer guy didnt do his job ... :( 1160689551 M * derjohn2 did anythign change that could have to do with the NFS problem? 1160689562 M * derjohn2 And: The mysql locked fixed? 1160689587 M * derjohn2 *mysql locks bug ? 1160689646 M * Bertl nfs should be fixed, mysql is still in unknown state 1160689661 M * Bertl but we have a new phenomenon to hunt for atm :) 1160689686 M * derjohn2 what? the NFS problem, was a real bug ? 1160689705 M * Bertl we are probably talking about different things 1160689706 M * derjohn2 I still thought it was my fault in the config ... 1160689719 M * Bertl I'm talking about the inability to do mounts from inside a guest 1160689737 M * Bertl (which was a missing capabilitly check) 1160689739 M * derjohn2 ah, that a different one, yes 1160689756 M * derjohn2 but it was a bug that was long there ... 1160689779 M * derjohn2 at least in rc19 is was not able to mount inside .... so I did it outside 1160689799 M * Bertl yeah, well, it seems we didn't understand the bug reports at this time 1160689806 Q * DreamerC Ping timeout: 480 seconds 1160689822 M * Bertl the issue was so trivial, that we just didn't see it :) 1160689838 M * Bertl luckily daniel_hozac figured it, and so it was finally fixed 1160689986 M * Bertl derjohn2: btw, how far away from Köln are you (usually) 1160690023 M * derjohn2 Bertl, Köln is about 90 mins by car. 1160690040 M * derjohn2 (maybe 60 .. depends on the car and the traffic) 1160690061 M * derjohn2 Bertl, why? are you in koeln ATM ? 1160690147 M * Bertl http://www.linuxworldexpo.de/vkf_06.php?ID=122 1160690235 M * derjohn2 ah .. Bertl : do they offer free tickets if I register soon ? 1160690266 M * Bertl absolutely no idea ... but might be worth asking ... 1160690300 M * derjohn2 http://www.linuxworldexpo.de/reg.php?Zielgruppen_ID=2&ID=30 1160690302 M * derjohn2 there it is 1160690401 Q * h01ger Quit: h01ger 1160690403 J * h01ger ~holger@socket.layer-acht.org 1160690994 Q * dna_ Quit: Verlassend 1160691981 Q * meandtheshell Quit: exit (0); 1160693761 Q * yarihm Remote host closed the connection 1160693920 Q * Heinzwurst Ping timeout: 480 seconds 1160694506 M * bronson How do I remove a guest from the system entirely? 1160694524 M * bronson just rm -rf /etc/vservers/guest-name? Or is there more? 1160694533 M * bronson (and, of course, rm -rf the root filesystem) 1160694549 M * doener that's all, but remember to stop the vserver first 1160694557 M * Bertl yep, should be enough, btw, you didn't get the solution earlier, right? 1160694593 M * Bertl bronson: eth0:2 is _wrong_ for 'dev' 1160694613 M * bronson I didn't; I'll try again tongith. 1160694634 M * Bertl you want either to have eth0 in dev and '2' in name 1160694650 M * Bertl if you really want the tools to create that alias 1160694669 M * bronson Really? Seems to me like eth0:2 is the device! Interesting. 1160694669 M * Bertl or (which is more likely) you want to use nodev instead of 'dev' and 'name' 1160694682 M * Bertl bronson: no, that is called an 'alias' 1160694692 M * Bertl i.e. eth0 is the device, and '2' is the alias 1160694704 M * bronson So I would have to name my machines 1, 2, 3... etc if I used the first technique you mention. 1160694728 M * bronson Ah. You're right. Mostly they're interchangable so I get lazy. 1160694729 M * Bertl no necessary, the 'name' in interfaces is independant from the guest name 1160694734 M * bronson OH. 1160694737 M * bronson OK. 1160694747 M * Bertl and btw, you do not need aliases at all 1160694757 M * bronson No? 1160694766 M * Bertl just specifying eth0 is enough, the addresses will be added 1160694771 M * bronson Awesome. 1160694780 M * bronson I should have thought of that! 1160694785 M * doener the "name" is attached to the address 1160694794 M * Bertl the only difference is, the aliases show up with 'ifconfig' (ancient tool) 1160694803 M * bronson Makes sense when you think about it... 1160694823 M * Bertl while the addresses used by default require newer tools (like ip from iproute2) 1160694841 M * Bertl OTOH, the normal addresses look nicer inside a guest 1160694858 M * bronson So, I can go back and try again with --ip 192.168.blah and --dev eth0 and it should just work? 1160694864 M * doener yep 1160694875 M * bronson I assume all the guest IPs must fall within the network interface's device mask... 1160694885 M * bronson netmask!! 1160694896 M * bronson device mask? I'm getting bleary. 1160694899 M * Bertl nope, netmasks are individual 1160694917 M * bronson So, just tell it the device and the packets just flow. 1160694923 M * bronson (assuming it passes iptables) 1160694938 M * Bertl yes, note that when you specify the device, the tools take control of that ip 1160694949 M * Bertl i.e. it will be added on guest start and removed on shutdown 1160694960 M * bronson Good. That's how it *should* work! 1160694966 M * Bertl if you don't want that, that's what you use the nodev for 1160694970 M * bronson Got it. So I just need to ensure the IP isn't in use anywhere else. 1160694976 M * Bertl precisely 1160694982 M * bronson Awesome. Thanks. 1160694985 M * bronson I'll try again in a few hours. 1160695006 M * Bertl okay, if you have issues/question, you know where to ask :) 1160695033 M * bronson yeah, you guys are great. I'll be back. 1160695036 Q * derjohn2 Read error: Operation timed out 1160695046 J * derjohn2 ~aj@dslb-084-059-002-253.pools.arcor-ip.net 1160695058 M * doener Bertl: regarding virtual loopbacks... maybe we could have addresses that just equal the xid (I guess they are invalid -> no conflicts *g*) and add routes to table 0 that are host local and are only matched for that single vserver (like fwmark routing) 1160695095 M * doener does that make any sense? 1160695177 M * doener hm, no... SELECTOR is used at least twice in ip(8) and the one that has fwmark is not for "route" :( 1160695178 M * Bertl yes, that is what I plan, the idea is like this (in case you missed my elaborations or they weren't clear enough) 1160695194 M * doener probably missed them... too many connection losses 1160695201 M * Bertl i.e. I would like to use 127.x.x.127 inside a guest (as loopback) 1160695211 M * Bertl assign that with x.x = xid 1160695228 M * Bertl and map it on every access from userspace to 127.0.0.1 inside 1160695246 M * Bertl but let it work as is inside the network stack/kernel 1160695314 M * doener ok 1160695363 M * Bertl IMH that should work quite fine 1160695367 M * Bertl *IMHO 1160695387 M * Bertl and it doesn't require any changes besides the remapping we already do 1160695419 M * Bertl (except for userspace querying, which will need tiny virtualizations) 1160695536 J * mire ~mire@111-167-222-85.adsl.verat.net 1160695835 M * Bertl welcome mire! 1160696074 M * Zaki Bertl, hi :) 1160696078 M * Zaki Bertl, i have a personal guess may i talk about it with you please? 1160696090 M * Zaki probably in pm? 1160696154 M * Bertl well, so be it :)