1138492835 M * blizz only hollow's stuffi nstalled so far 1138492840 Q * Doener Quit: Leaving 1138492853 M * Bertl then use vcontext 1138492854 M * Hollow vcontext -M -x 42 -- grep .. 1138492867 M * Bertl -C ? 1138492870 M * Hollow s/-M/-C/ probably 1138492935 M * blizz wait a sec 1138492964 M * blizz [root@anarchy:/var/www/ehd]# vcontext -M -x 1337 -- grep CapE /proc/self/status 1138492965 M * blizz CapEff: 00000000fffffeff 1138492996 M * Bertl Hollow: so no --secure option there yet? 1138493022 M * Bertl (to drop capabilites) 1138493022 M * Hollow no, but that's a job for the init scripts anyway 1138493028 M * Hollow ehm 1138493031 M * Hollow the vserver scripts 1138493038 M * Bertl okay, and they do that properly? 1138493044 M * Hollow not yet 1138493052 M * Bertl okay, that explains it then ... 1138493061 M * Hollow yeah, but i said that before 1138493071 M * Bertl oh, sorry, must have missed that 1138493075 M * Hollow when you were off 1138493109 M * Hollow i said it to blizz 1138493154 M * Bertl okay, sorry blizz, sorry Hollow ... I'm going to do something useful :) 1138493159 M * Hollow but the question is, why sshd cannot open /dev/pts/0 1138493170 M * Hollow guess that was the error 1138493188 M * Bertl on the host? 1138493201 M * Hollow hm, iirc in the guest 1138493210 M * blizz ah, my log entry? 1138493242 M * Hollow jup 1138493441 M * ScoobyD00 the last line of the exploit program says "Exploit seems to work. =)" - even though I have the barrier set? 1138493451 M * ScoobyD00 is that normal? 1138493452 M * Hollow yeah 1138493459 M * ScoobyD00 :o) 1138493463 M * Hollow you can verify it by looking at the filesystem 1138493499 M * ScoobyD00 ahh yes :o) 1138493518 M * ScoobyD00 so, this tagxid thing - why is it bad to set this for a root partition? 1138493532 M * Hollow you mean barrier 1138493546 M * ScoobyD00 no, back to something else i was rambling about earlier :o) 1138493551 M * Hollow ah 1138493570 M * Hollow well, you should set it for the partition your vserver are on 1138493577 M * ScoobyD00 i only have 1 big partition, and re-partitioning is going to be awkward 1138493587 M * ScoobyD00 so i would have to set it for "/" 1138493591 M * ScoobyD00 on the host 1138493619 M * Bertl I would not suggest to do that 1138493627 M * ScoobyD00 i know - whyso? :o) 1138493641 M * Bertl because administrative files will get tagged too 1138493655 M * Bertl which might lead to all kind of 'strange' issues 1138493709 M * Bertl but it is possible, just not made easy, so that folks don't do it without exactly knowing what they do 1138493713 M * ScoobyD00 wish i had read that before i shipped the server to the datacenter :o( 1138493777 M * Bertl you have no rescue partition on that machine? 1138493788 M * ScoobyD00 rescue partition? 1138493807 M * Bertl well what do you do if the filesystem gets corrupted and you need to check it? 1138493812 M * ScoobyD00 i have /boot 1138493831 M * ScoobyD00 which is quite small with just kernel on it and initrd 1138493872 M * ScoobyD00 but /etc, /usr and /var are on this big partition taking up rest of the disk 1138493900 M * ScoobyD00 /dev/md2 183G 433M 183G 1% / 1138493900 M * ScoobyD00 tmpfs 497M 0 497M 0% /dev/shm 1138493900 M * ScoobyD00 /dev/md0 228M 13M 203M 6% /boot 1138493980 M * Bertl not even swap space? 1138493991 M * Hollow ok, really off to bed now, cu tomorrow 1138493993 M * ScoobyD00 yeah, i have 3g swap partition 1138494024 M * ScoobyD00 i guess i could hack it - it is just that if it goes wrong i dont fancy taking a day off work to go to the datacenter and fix the machines 1138494085 M * Bertl yeah, I can understand that ... 1138494122 M * ScoobyD00 i have got it working with some mounted disk images which is ok for now i guess 1138494141 M * ScoobyD00 but performance suffers 1138494289 M * Bertl yes, it's a large overhead 1138494360 M * Bertl and you lose all the sharing usually done 1138494383 M * ScoobyD00 what gets shared? 1138494437 M * Bertl if you use unification, you will share unified files, inode caches and more 1138494456 M * ScoobyD00 oh right, didnt know that 1138494474 M * Bertl make properly configured and unified guests about 10MB in size :) 1138494478 M * ScoobyD00 so if one vserver changes a file, it makes a copy? 1138494496 M * Bertl with CoW link breaking yes 1138494827 M * ScoobyD00 does that happen automatically, or do i need to configure it somewhere? 1138495125 M * Bertl the CoW break or the unification? 1138495308 M * ScoobyD00 sorry, just found the wiki feature matrix and saw it is only on the latest release :o) 1138495320 M * Bertl that's right ... 1138495330 M * Bertl but unification works for older releases too 1138495339 M * Bertl (just the CoW link breaking is not there) 1138495351 M * ScoobyD00 so does it essentially lock the files on older releases? 1138495364 M * Bertl they are immutable but unlinkable 1138495373 M * Bertl i.e. you cannot change them, but remove them 1138495383 M * Bertl this works for 99% of all packaging systems 1138495867 M * blizz Bertl, i installed util-vserver now 1138495875 M * blizz but i cant stop the server ;) 1138495876 M * blizz [root@anarchy:/etc/vservers]# chcontext --xid 1337 --secure -- grep CapE /proc/self/status 1138495879 M * blizz CapEff: 00000000fffffeff 1138495902 M * Bertl that's not a good sign, but kind of expected for the broken debian kernel 1138495920 M * blizz can i kill it somehow? so i can start over.. 1138495939 M * Bertl just update to one of the testing/unstable kernels 1138495949 M * Bertl make sure that the package mentions micah 1138495960 M * Bertl (those packages are working AFAIK) 1138495961 M * blizz running 2.6.15.1 1138495973 M * blizz im crux, not de(r)bian :) 1138495979 M * Bertl hmm, sorry 1138495994 M * blizz nevermind! 1138496008 M * Bertl okay, let's do the testme.sh then 1138496021 M * blizz right! 1138496045 M * blizz everything succeeded 1138496061 M * Bertl hmm, what kernel aptch? 1138496064 M * Bertl *patch 1138496074 M * blizz 2.1.0.5 1138496095 M * Bertl hmm, okay, you should definitely upgrade (for a different reason) see ML 1138496108 M * Bertl but, that doesn't explain your missing capabilities 1138496142 M * blizz maybe because i started it with vserver-utils and not util-vserver? 1138496166 M * Bertl now I'm confused ... 1138496172 M * blizz well 1138496178 M * Bertl ah, 1337 is an existing guest? 1138496182 M * blizz right 1138496193 M * blizz eww 1138496196 M * Bertl okay, then do that same test with a non existing xid 1138496201 M * blizz [root@anarchy:/etc/vservers]# chcontext --xid 42 --secure -- grep CapE /proc/self/status 1138496204 M * blizz New security context is 42 1138496205 M * blizz sorry.. didnt know that 1138496208 M * blizz CapEff: 00000000344c04ff 1138496217 M * Bertl that's fine now ... 1138496227 M * Bertl nevertheless, upgrade to 2.1.0.9 1138496233 M * blizz ill do 1138496272 M * blizz do i need the rc1 kernel for that? 1138496302 M * Bertl hmm, well, yes, but you can simply add the relevant patch or use the 2.1.0.5.1 for 2.6.15 1138496576 M * ScoobyD00 thanks for all the help tonight guys - im off to bed :o) 1138496585 P * ScoobyD00 1138496619 J * Thorsten ~Thorsten@dslb-084-058-177-167.pools.arcor-ip.net 1138496626 M * Bertl welcome Thorsten! 1138496632 M * Thorsten Hi Bertl 1138496664 M * Thorsten Remeber that I wanted to upgrade my kernel? 1138496677 M * blizz so ill use the .1 patch 1138496790 M * Thorsten I just want to report that 2.6.15.1-vs2.0.1.2 works without problems so far and I'm happy. 1138496874 M * Bertl excellent! you know our Happy Users page? 1138496886 M * Bertl (or if that applies to you, the Hosting Page) 1138496905 M * blizz who says there are no happy hosters? ;) 1138496905 Q * bwana Read error: Connection reset by peer 1138496909 J * bwana ~bwana@71-82-125-40.dhcp.roch.mn.charter.com 1138496922 M * Bertl blizz: well, that's what the Hosting page is for :) 1138496927 M * blizz sure :) 1138496943 M * blizz oops, 404 here: http://vserver.13thfloor.at/Experimental/patch-2.6.15-vs2.1.0.5.1.diff 1138496979 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.15.1-vs2.1.0.5.1.diff 1138496984 M * Thorsten Bertl, http://linux-vserver.org/VServer+Users scroll to the bottom ;-) 1138497031 M * Bertl ah, perfect .. well, what can I say more ... feel free to hang around :) 1138497834 M * blizz baking my new 2.6.16-rc1 with latest patches ;) 1138498123 M * blizz CC security/security.o 1138498124 M * blizz security/security.c:202: error: redefinition of 'vx_capable' 1138498126 M * blizz uh oh? 1138498149 M * Bertl interesting, could you upload your .config ? 1138498165 M * blizz sure 1138498189 M * blizz http://ehn.shacknet.nu/kernel-config.txt 1138498247 M * blizz include/linux/sched.h:985:static inline int vx_capable(int cap, int ccap) 1138498256 M * blizz security/security.c:201:int vx_capable(int cap, int ccap) 1138498273 M * blizz its a fresh 2.6.16-rc1 with 2.1.0.9 1138498313 M * Bertl k, give me a few minutes to figure and fix it 1138498327 M * blizz thank you alot 1138498370 M * Bertl you have security option enabled, but only capabilities selected 1138498388 M * Bertl do you plan to add other security frameworks? 1138498443 M * blizz i thought it was a dependency for the lcs 1138498491 M * Bertl hmm, right, nevermind 1138498511 M * Bertl gcc version sorry if I ask that again) 1138498522 M * blizz you dont :) 3.4.4 1138498763 M * Bertl okay, seems a patch went wrong there ... tracking down now 1138498803 M * blizz oki 1138499078 Q * Thorsten Quit: Leaving 1138500079 Q * derjohn Ping timeout: 480 seconds 1138500212 M * Bertl blizz: http://vserver.13thfloor.at/Devel/PAT-2.6.16/delta-capable-fix01.diff 1138500328 M * Bertl (jsut apply and do 'make') 1138500536 M * blizz okidoki 1138501119 M * blizz rebooting 1138501459 J * DoM_ ~DoM@151.56.248.215 1138501475 M * Bertl welcome DoM_! 1138501476 M * DoM_ hi all 1138501487 N * DoM_ `DoM` 1138501555 J * derjohn ~derjohn@80.69.37.19 1138501742 J * cohan ~cohan@koniczek.de 1138501753 M * Bertl wb derjohn! cohan! 1138501768 M * cohan re 1138501796 M * derjohn Bertl, he he ..... wb ?? well, my DSL carrier had service cycle ;) 1138501833 M * `DoM` a simple question: how many vserver can i run on a p4 3.20GHz + 2GB DDR ? 1138501848 M * cohan i just joined with my irssi i also use for freenode 1138501864 M * derjohn `DoM`, well, 2^16-2 .... 1138501869 M * derjohn or so 1138501876 M * Bertl `DoM`: roughly between 5 and 200 1138501882 M * cohan so i should not often need to reconnect ;) but i'll go to bed now... cu 2morrow 1138501903 M * derjohn `DoM`, what purpose are they used for? 1138501915 M * `DoM` i ask you this cause i talked with kir of openvz 1138501938 M * Bertl and? what can OpenVZ do there? 1138501943 M * derjohn `DoM`, what does kir say? 1138501946 M * `DoM` and he told me that for good perfomance on openvz 10 vs was around good for any performance 1138501979 M * `DoM` of course depends on user load 1138501981 M * Bertl yes, that's probably a realistic answer ... 1138501999 M * Bertl (usually the VZ guys always talk of hundreds of guests) 1138502003 M * `DoM` but just to know which could be best solution to divide vserver 1138502015 M * derjohn `DoM`, well, you can neglect the vserver overhead, just count what the processes need. thats it. 1138502041 M * Bertl `DoM`: OVZ and Linux-VServer are very similar regarding virtualization 1138502057 M * Bertl plus minus a few features on both sides 1138502065 M * `DoM` i am not sure about which is more performant between openvz and vserver but i see that vserver looklike more a great community .. more opensource style while openvz not.. 1138502079 M * `DoM` i read a big thread on forums about openvz VS vserver 1138502079 M * Bertl that's the main difference 1138502104 M * Bertl OpenVZ is basically an SWSoft project 1138502116 M * `DoM` but at least i didnt understand which is more good and performant :P cause each one same that own was better for some kernel functions and so on 1138502130 M * `DoM` yes same as virtuozzo 1138502177 M * `DoM` anyway i had to adapt some init script to let openvz works on debian while vserver looklike more confortable 1138502213 M * `DoM` openvz works on 2.6.8 while vserver 2.6.14.3 if i am not wrong 1138502226 M * `DoM` and vserver works with grsec too right? 1138502232 M * derjohn `DoM`, on Debian it runs 100% smooth. You can also find the DD responsible for the debs here on this channel. 1138502256 M * blizz what the hell? my new kernel says it has no NAT builtin? 1138502292 M * derjohn `DoM`, but take care to use sid as host (at least if you dont want to compile the tools yourself) 1138502299 M * Bertl `DoM`: latest versions are for 2.6.16-rc1 :) 1138502313 N * cohan cohan_zzz 1138502315 M * derjohn blizz, then use iptables :) 1138502334 M * `DoM` but i read on faq that vserver does not support ipv6 yet :( ... openvz too but i hope that vserver was the solution for my problem :\ 1138502342 M * blizz derjohn, i do :) 1138502354 M * `DoM` Bertl, right :) 1138502356 M * Bertl `DoM`: what's your problem? 1138502359 M * blizz [root@anarchy:/home/blizz]# iptables -vnL 1138502359 M * blizz FATAL: Module ip_tables not found. 1138502366 M * blizz but its exactly the same kernel config as last time 1138502383 M * `DoM` Bertl, i dream about vserver and ipv6 to implement for some services that i offer for free 1138502389 M * derjohn blizz, thats anarchy on the kernel modules, 1138502400 M * Bertl `DoM`: ipv6 will come pretty soon ... 1138502415 M * Bertl (but of course, it will need testing) 1138502418 M * blizz derjohn... :P 1138502453 M * derjohn blizz, it looks like whole ip_tables is not there .. not NAT is missing .... 1138502460 M * blizz funny thing 1138502466 M * Bertl `DoM`: but yes, for now there is no ipv6 inside the guests 1138502466 M * `DoM` derjohn, you mean debian sid? Why?Sarge packages are not good to work or what O_o ? 1138502506 M * derjohn Bertl, is there a way to "NAT" a hosts v6 adress to a rfc1918 IP of a guest? 1138502521 M * Bertl regardless what debian version, make sure that you have _recent_ packages, the maintainer (ola) did a lousy job 1138502537 M * `DoM` Bertl, ah 1138502554 M * derjohn `DoM`, I didnt check sarge for quiter a time, but the big cleanup in sarge is only a few weeks old. 1138502556 M * `DoM` Bertl, anyway i can compile from source + checkinstall 1138502571 M * `DoM` just for a clean installation 1138502572 M * derjohn `DoM`, so expect sid to be cool, etch maybe .... sarge .. i doubt 1138502576 M * Bertl `DoM`: yep, that's the distro independant way 1138502613 M * Bertl `DoM`: but recent packages (which mention Micah) should be fine 1138502621 M * derjohn blizz, modprobe ip_tables ? 1138502639 M * blizz no, its builtin 1138502642 M * blizz it *should be* 1138502648 M * blizz i just took the old config from 2.6.15.1 1138502649 M * `DoM` well .. you guys are much friendly :) thanks for all answers and i try to install now vserver :) 1138502654 M * blizz to 2.6.16-rc1, ran make oldconfig 1138502659 M * blizz and now iptables is gone out of my kernel 1138502666 M * Bertl derjohn: I have no idea (regarding the nat) 1138502686 M * Bertl `DoM`: you're welcome! that's why it is a community project! 1138502688 M * derjohn Bertl, then we share somethign common ;) 1138502703 M * Bertl derjohn: but maybe `DoM` knows? 1138502767 M * derjohn Bertl, goog Question. I dont have much experience with v6 - But I should have, it will come sooner than we expect. 1138502786 M * derjohn Bertl, cellphones, PDA etc. 1138502787 M * Bertl `DoM`: have you found the getting started stuff on the wiki? 1138502843 M * derjohn silence ..... ? 1138502877 M * Bertl patience :) 1138502883 A * derjohn leaves the office now. going to sleep. 1138502892 M * Bertl derjohn: good night then! 1138502901 M * derjohn Bertl, namaste ;) 1138503840 M * marl anyone able to tell me, if i have a drive that runs debain normally, and i install a new h/d and install a very basic system with vserver, so long as i set up /etc/vservers/old-drive should i be able to mount the old drive in /vserver/old-drive and enter it and run everything as normal? 1138503857 M * Bertl yes 1138503873 M * Bertl you will get a bunch of errors, when the image tries to access hardware 1138503881 M * Bertl but that is fine, it will start anyway 1138503915 M * marl :) one other question, im sure ive seen someware ont he website a page explaining problems when using x-windows within a vserver, but i cant find it, am i getting confused or has it been removed? 1138503982 M * Bertl google says it's still there 1138503983 M * Bertl http://linux-vserver.org/Vservers+and+X 1138504082 M * marl thanks Bertl, the sites search option didnt find anything, but thats the page :) 1138504178 M * Bertl you're welcome! 1138504310 M * marl btw, just for info, u remember i was having problems the other day with a new kernel booting up but getting nothing on the screen after some of the boot messages, althow i could still type etc? 1138504357 M * marl it was down to me having enables some of the older hurcules video card compatibility stuff int he kernel config wihtout relising it, system now boots fine :) 1138504408 M * Bertl ah, cool! 1138505718 J * marl_ ~matt@84.92.193.226 1138505719 Q * marl Read error: Connection reset by peer 1138506912 Q * duplex Quit: Leaving 1138507239 N * ebiedermOo ebiederm 1138507353 M * Bertl wb ebiederm! 1138507360 M * ebiederm Hey Bertl. 1138507422 M * ebiederm I have been having an argument with myself today after looking and seeing that references to processes don't map well at all to tty_io.c 1138507465 M * ebiederm The problem is there wasn't a member of function. 1138507516 M * ebiederm However controlling tty's have a one to one mapping to processes so their problems can be solve in any manner of ways. 1138507532 M * Bertl andttys are ancient 1138507563 M * ebiederm Weak references to processes still seem a good solution to other parts of the kernel. 1138507581 M * ebiederm Bertl: How many terminals do you have open :) 1138507628 M * Bertl ptys or ttys? 1138507650 M * ebiederm tty_io.c is the common code for both. 1138507702 M * Bertl okay 1138507753 M * ebiederm The nice thing is that tty's aren't susceptible to session id rollover (pgrps maybe but that is fixable). 1138507884 M * ebiederm For fcntl F_SETOWN and friends fixing the pid rollover problem is over unnecessary (as nothing really bad can happen) and userspace is given enough control to prevent it. 1138507892 M * ebiederm But it also seems harmless to fix. 1138507903 M * Bertl sounds good 1138507937 Q * bwana Read error: Connection reset by peer 1138508078 M * ebiederm I also discovered that put_user is not as type safe as it should be. 1138508149 M * ebiederm It was really annoying changing the type from pid_t to a pointer and put_user did not even complain!. 1138508327 M * ebiederm The good news is that after just a little more testing and merging of a couple of patches I think I am ready to post code. 1138508367 M * Bertl excellent .. but I guess the networking stuff is still on the todo list, yes? 1138508406 M * ebiederm Yes. But that should come fairly quickly. 1138508410 M * ebiederm Once I get to it. 1138508415 M * ebiederm My apologies for keeping you waiting. 1138508430 M * Bertl np, it's your time ... 1138508441 M * Bertl you have to decide what to do when ... 1138508563 M * ebiederm Thanks for understanding. 1138508601 J * bwana ~bwana@71-82-125-40.dhcp.roch.mn.charter.com 1138508772 M * Bertl ebiederm: IMHO that goes without saying ... 1138508801 M * bwana what goes without saying 1138508804 M * ebiederm But saying it is polite. 1138508805 M * bwana that I ROCK 1138508821 M * Bertl bwana: of course :) 1138508823 M * bwana heh 1138508833 M * ebiederm bwana: Is that the distro you use? 1138508874 M * bwana no, that is just my general state of being 1138508894 M * ebiederm Good for you. 1138508900 M * bwana heh 1138508964 M * ebiederm I think I am tired of changing header files though. 1138508972 M * ebiederm So much code rebuilds. 1138509059 M * Bertl yeah, a lot of headers could use cleanups or a good splitting 1138509100 M * ebiederm There is some of that. But when you are changing struct file.... It is hard to fix that. 1138509154 M * Bertl that's right ... 1138509171 M * ebiederm I did figure out an alternative technique to reference counting to deal pid wrap around if we ever need it. 1138509199 M * ebiederm Keep a list of things to look at and walk it when the pid allocator hits the wrap around point. 1138509219 M * Bertl eeek 1138509222 M * ebiederm That's terrible from a latency point of view though. 1138509243 M * Bertl that reminds me of unique inode numbers :) 1138509275 M * ebiederm Bertl what about uniuqe inode numbers? 1138509290 M * Bertl the wrap around should not be a problem, just use an unique identifier 1138509374 M * ebiederm The larger and more sparsely populated your address space the easier it is to find a unique identifier. 1138509378 M * `DoM` Berti looklike works really fine 1138509389 M * `DoM` ^^ 1138509392 M * Bertl `DoM`: good to hear! 1138509409 M * `DoM` i just need to understand a thing 1138509418 M * `DoM` if i do image of vserver1 for example 1138509454 M * Bertl yes? 1138509471 M * `DoM` after i need only to copy/modify settings that i find in /usr/local/etc/vservers/vserver1 to create vserver2 ? 1138509488 M * `DoM` and of course extract image on right path about vserver2 1138509497 M * Bertl well, yes that's an option 1138509508 M * Bertl you can do that easier with the skeleton build method 1138509516 M * `DoM` ah 1138509531 M * `DoM` there is already an utility that do this :d 1138509538 M * Bertl http://linux-vserver.org/alpha+util-vserver 1138509560 M * Bertl basically you build the skeleton, which is empty except for proc and /dev 1138509575 M * Bertl but you already specify all the options at build time 1138509601 M * Bertl then you simply throw away the directory and replace it by a copy of your template 1138509644 M * Bertl if you use unification, you can reduce the disk space and resource usage by unifying common files between guests 1138509794 M * `DoM` i try immediately 1138509812 M * `DoM` mm not much immediately it still downloading packages :P 1138509817 M * Bertl do you use the devel branch (i.e. vs2.1.x) 1138509854 M * Bertl just because devel supports CoW link breaking which allows you to unify the entire guest 1138509899 M * `DoM` i use util-vserver [0.30.210] 1138509900 M * Bertl with the stable branch (i.e. vs2.0.x) you should only unify binaries and libraries (i.e. files which do not change except on package upgrade) 1138509908 M * `DoM` and vserver 2.01 1138509943 M * Bertl okay, so read the unification part on the http://linux-vserver.org/alpha+util-vserver page 1138509948 M * ebiederm Bertl: How are cow links implemented? 1138509967 M * Bertl ebiederm: you know the concept of unified files? 1138510006 M * ebiederm Bertl: Yes. Is this simply inode sharing or can you share at the block level. 1138510014 M * Bertl they are hard links, made immutable with an additional flag which inverts the immutaility on unlink 1138510033 M * Bertl *immutability 1138510058 M * ebiederm Good for small files sucks for large files. 1138510059 M * Bertl now the CoW link breaking comes into play when you write to such an inode 1138510068 M * Bertl ebiederm: why? 1138510105 M * ebiederm Largely it depends on how much of a large file you are writing to, and how often you are breaking cow links on it. 1138510126 M * Bertl ebiederm: it mostly about common files in identical installations 1138510132 M * ebiederm But in general if you have a block level cow breaking it can be much more efficient if you don't change most of the blocks. 1138510141 M * Bertl ebiederm: we are not trying to save disk space or do filesystem work :) 1138510147 M * ebiederm Right for root installs it is perfectly fine. 1138510173 M * Bertl IMHO block level tricks should happen inside or below the filesystem layer 1138510174 M * ebiederm If you aren't trying to safe disk space why the cow links? Very quick copies? 1138510196 M * Bertl ebiederm: it saves resources and disk space when the files are unified 1138510206 M * ebiederm Bertl: I agree with where block level tricks need to happen. 1138510214 M * Bertl ebiederm: it would not save resources with a block level approach 1138510254 M * ebiederm Ok. Different use model. You create than unify. I think of creating and then breaking the cow link. 1138510256 M * Bertl ebiederm: consider 20 instances of bash 2.0.1 in 20 different guests 1138510276 M * Bertl we have them once in the inode cache :) 1138510311 M * Bertl the only catch is, what if somebody wants to upgrade his bash 1138510315 M * ebiederm Bertl: Agreed, and the single copy in the file cache is probably one of the biggest wins. 1138510330 M * Bertl and that's where the iunlink comes in 1138510339 M * Bertl and on devel the CoW link break 1138510356 M * Bertl the iunlink allows you to remove the bash, and install a new one 1138510371 M * Bertl (which obviously doesn't work for writing or appending to files) 1138510392 M * Bertl this is covered by the CoW link breaking, which uses sendfile to make a copy of that file 1138510604 M * ebiederm I keep hoping to hear some has down the filesystem work. But it is good to at least know that cow links are being used in some way. 1138510662 M * Bertl well, we built our code ontop of Jörn Engls work .. with some modifications ... 1138510690 M * Bertl IMHO it's a shame that there is no sane way to 'copy' an inode in the kernel 1138510919 M * ebiederm Yes letting filesystems know a copy is happening could be interesting. 1138510934 M * `DoM` mmm Bertl 1138510952 M * Bertl ebiederm: it is the basis for efficient block sharing algorithms 1138510973 M * `DoM` i only copied image of vserver1 on vserver2 path and i find vs2 config on /usr/local/etc/vs ... it's possible ? 1138510996 M * `DoM` no i am wrong somewhere 1138511018 M * Bertl if you 'copied' the config tree too, 1138511027 M * Bertl you should check the symlinks inside 1138511035 M * Bertl which will still point to the 'old' guest 1138511044 M * `DoM` no no point to new one 1138511051 M * `DoM` name server2 ip new and so on 1138511067 M * `DoM` maybe i when i try skeleton option it recreate cfgs 1138511092 M * Bertl the skeleton method builds a config for you, yes 1138511122 M * Bertl it will also move the old guest out of the way (if forced) 1138511153 M * `DoM` with skel option i have new cfgs and on new vs dir only dev proc ecc 1138511158 M * `DoM` and etc :P 1138511162 M * Bertl yep 1138511167 M * Bertl precisely, 1138511177 M * `DoM` so i just need to untar old image to new path 1138511183 M * Bertl now you remove the /vservers/ 1138511193 M * Bertl and replace it with your template copy 1138511207 M * `DoM` ok that's really really really fantastic :D 1138511217 M * Bertl we thought so too :) 1138511456 M * Bertl btw, you should make sure to have a unique id for each of your guests, to avoid issues in the future 1138511500 M * Bertl i.e. when using disk limits on a shared partition or so 1138511518 M * `DoM` yes but listen 1138511521 M * Bertl (basically virtual limits for the disk space) 1138511531 M * `DoM` config files inside vs like hostname for example 1138511542 M * Bertl yup? 1138511546 M * `DoM` hostname indicate not guest vs but real vs server 1138511562 M * `DoM` i need to change values like this or need to leave same on each vs guest ? 1138511578 M * Bertl that entirely depends on your setup 1138511586 M * Bertl your options are: 1138511611 M * Bertl - set the hostname on guest startup and forbid changing inside the guest 1138511631 M * Bertl - allow the guest to change the hostname (default) and configure it inside 1138511674 M * Bertl - give a default and do not change it inside (but leave the options open) 1138511754 M * `DoM` mm where can i find build option ? on man there arent 1138511795 M * Bertl which one? i.e. what do you want to change? 1138511811 M * `DoM` i want forbbiden to change hostname 1138511813 M * Bertl ah, general build options?, just use 1138511821 M * Bertl vserver - build --help 1138511822 M * `DoM` vserver build --help ?^ 1138511824 M * `DoM` k 1138511872 M * Bertl http://linux-vserver.org/Caps+And+Flags 1138511883 M * Bertl describes the capabilities and flags 1138511929 M * Bertl in your case the utsname is relevant 1138512059 M * Bertl okay folks, I'm off to bed ... will be back tomorrow ... 1138512073 M * `DoM` cu Bertl and thanks for all! 1138512095 M * Bertl `DoM`: if you have further questions which can not be answered by the wiki, just ask here, somebody will answer them sooner or later 1138512116 M * Bertl night everyone, have fun! 1138512121 N * Bertl Bertl_zZ 1138512130 M * `DoM` ok if i not find on wiki i will ask like crazy here :D nn! 1138512632 M * ebiederm Good night Bertl 1138513340 J * undefined ~undefined@adsl-68-93-109-94.dsl.rcsntx.swbell.net 1138514437 J * Smutje ~Smutje@xdsl-84-44-245-233.netcologne.de 1138514564 Q * Smutje_ Ping timeout: 480 seconds 1138515000 J * bwana2oo ~bwana@71-82-125-40.dhcp.roch.mn.charter.com 1138515000 Q * bwana Read error: Connection reset by peer 1138516357 N * ebiederm ebiedermZz 1138521021 Q * undefined Remote host closed the connection 1138522791 M * Hollow morning 1138522801 M * blizz moin hollow 1138522812 M * Hollow moin blizz, sshd bug fixed? 1138522849 M * blizz no *lol*.. i compiled 2.6.16-rc1 yesterday and well, the .config was missing IP_TABLES 1138522858 M * blizz so i went to bed too.. :P 1138522879 M * Hollow heh 1138523579 M * blizz thats really strange, i couldnt find any nat option in the generic 2.6.16-rc1 1138523767 Q * lonewolff Read error: Connection reset by peer 1138523792 J * lonewolff ~lonewolff@host-84-9-143-159.bulldogdsl.com 1138524273 N * cohan_zzz cohan 1138524539 Q * lonewolff Ping timeout: 480 seconds 1138525882 Q * daniel_hozac Quit: upgrades 1138527589 Q * Hollow Quit: SIGTERM 1138527769 J * Hollow ~hollow@home.xnull.de 1138527940 Q * Hollow Remote host closed the connection 1138528068 J * Hollow ~hollow@home.xnull.de 1138528666 J * meandtheshell ~markus@85-125-229-82.dynamic.xdsl-line.inode.at 1138528865 Q * michal_ Ping timeout: 480 seconds 1138529356 M * blizz Hollow, problem solved.. 1138529372 M * blizz # CONFIG_NETFILTER_XTABLES is not set 1138529401 M * blizz thats the new iptables option 1138529407 M * Hollow to do what? 1138529460 M * blizz 2.6.15.1 CONFIG_IP_NF_IPTABLES == 2.6.16-rc1 CONFIG_NETFILTER_XTABLES 1138529464 J * michal_ ~michal@www.rsbac.org 1138529486 M * Hollow and this solves your sshd problem? 1138529501 J * Viper0482 ~Viper0482@p54975EC0.dip.t-dialin.net 1138529514 M * blizz no :) but it cost me some headache *lol* 1138529561 M * Hollow ic ;) 1138530905 J * bubulak ~bubulak@cicka.wnet.sk 1138535243 J * lonewolf1 ~lonewolff@host-84-9-143-159.bulldogdsl.com 1138535403 Q * bwana2oo Quit: adios 1138535470 N * lonewolf1 lonewolff 1138535484 J * Doener doener@i5387F88A.versanet.de 1138535770 N * lonewolff Guest1857 1138535841 Q * Guest1857 Read error: Connection reset by peer 1138535884 J * lonewolf1 ~lonewolff@host-84-9-143-159.bulldogdsl.com 1138536059 Q * Falle Quit: leaving 1138536933 Q * Doener Quit: Leaving 1138536942 J * Doener doener@i5387F88A.versanet.de 1138537202 N * lonewolf1 lonewolff 1138538303 J * daniel_hozac ~daniel@c-2d1472d5.010-230-73746f22.cust.bredbandsbolaget.se 1138538381 Q * Viper0482 Remote host closed the connection 1138539411 J * lilalinux ~plasma@80.69.35.186 1138540077 N * Bertl_zZ Bertl 1138540081 M * Bertl morning folks! 1138540158 Q * RoadRunnR Quit: Leaving 1138540178 M * daniel_hozac morning! 1138540608 M * blizz hey ho 1138540713 M * Bertl hey daniel_hozac! blizz! 1138540738 M * blizz greetings bertl 1138540853 M * `DoM` lo Bertl 1138540861 M * `DoM` i am going sleep now 1138540863 M * `DoM` :P 1138541065 M * Bertl hey `DoM`! everything fine? 1138543304 J * ScoobyD00 ~foo@80-195-186-201.cable.ubr08.newm.blueyonder.co.uk 1138543889 M * ScoobyD00 hi - i am trying to run this command "chxid -c test1 -R /vservers/test1/" 1138543913 M * ScoobyD00 but it keeps saying "invalid argument" for every file it finds 1138543927 M * daniel_hozac did you mount /vservers with the tagxid mount option? 1138543940 M * ScoobyD00 /dev/md3 on /var/lib/vservers type reiserfs (rw,noatime,attrs,tagxid) 1138543948 M * ScoobyD00 (with /vservers symlinked to that) 1138543979 M * daniel_hozac what kernel? 1138543993 M * ScoobyD00 is debian stable (2.6.8) 1138544020 M * daniel_hozac does testfs succeed on that kernel? 1138544080 M * ScoobyD00 do you have a link to that please? 1138544083 M * daniel_hozac http://vserver.13thfloor.at/Stuff/SCRIPT/testfs.sh-0.11 1138544096 M * Bertl IIRC, 12 should be last one no? 1138544096 M * daniel_hozac (IIRC, my connection is only semi-working so i can't check it right now...) 1138544099 M * daniel_hozac oh, ok. 1138544111 M * Bertl ScoobyD00: 2.6.8 has probably issues with reiserfs 1138544132 M * ScoobyD00 :o( 1138544141 M * Bertl ScoobyD00: but the testfs.sh will show, but be careful, you need an empty device 1138544152 M * Bertl i.e. it will reformat that device to different filesystems 1138544162 M * Bertl you can use a loopback device for that 1138544166 M * ScoobyD00 doh 1138544193 M * ScoobyD00 so when it says "testing ext2 filesystem ..." it was formatting my root ;o) 1138544229 M * Bertl well, no it uses /dev/null or zero by default 1138544239 M * Bertl so no harm done, abort it with CTRL-C 1138544246 M * ScoobyD00 :o) 1138544257 M * Bertl then use -h to see the defaults and options 1138544276 M * Bertl you want something like -D /dev/loop0 -M /mnt -F reiserfs or so 1138544292 M * Bertl but setup the loop first, roughly 256MB should be enough for reiser 1138544588 M * ScoobyD00 do you know how long this should run for? 1138544613 M * Bertl as I said, it uses /dev/null which is pretty large ... 1138544641 M * Bertl so I'd suggest to abort it (CTRL-C) 1138544652 M * ScoobyD00 sorry, i am running it on a loopback device now 1138544664 M * ScoobyD00 - /dev/loop0 on /mnt type reiserfs (rw) 1138544666 M * Bertl ah, okay, there it takes about 70 seconds 1138544671 M * ScoobyD00 ./testfs.sh-0.11 -D /dev/loop0 -M /mnt -F reiserfs 1138544693 M * Bertl did you mount the loop? 1138544697 M * ScoobyD00 yeah 1138544707 M * Bertl well I guess I have to add some checks 1138544711 M * Bertl i.e. don't do that :) 1138544715 M * ScoobyD00 ok :o) 1138544719 M * Bertl just do the losetup 1138544731 M * Bertl the test script will format the device and mount it 1138544740 M * ScoobyD00 ahh that looks betetr 1138544745 M * ScoobyD00 doing stuff now 1138544750 M * ScoobyD00 just sat at "testing reiserfs filesystem ..." before 1138544762 M * Bertl okay, you might want -t -x a s options 1138544770 M * Bertl 't -x' as options 1138544775 M * Bertl *argh* 1138544780 M * ScoobyD00 gotcha :o) 1138544781 M * Bertl -t and -x as options 1138544810 M * ScoobyD00 ok, its done 1138544823 M * ScoobyD00 can i paste output here? 1138544824 M * Bertl if anything failed, then it's a problem (with the -tx) 1138544833 M * Bertl better uplaod it somewhere 1138544838 M * Bertl e.g. pastebin.com 1138544877 J * bonbons ~bonbons@83.222.38.150 1138544905 M * ScoobyD00 neat site :o) http://pastebin.com/528723 1138544917 M * ScoobyD00 the -t -x one is at the bottom 1138545007 M * Bertl okay, could you retest with the 0.12 version? 1138545022 M * Bertl I assume most of the xid tests will fail anyway 1138545031 M * Bertl but the xattr tests should succeed 1138545188 M * ScoobyD00 ok: http://pastebin.com/528728 1138545210 M * Bertl ah, yes, as expected ... 1138545219 M * Bertl well, two options there 1138545232 M * Bertl a) do not use reiserfs (use ext2 or ext3) 1138545240 M * Bertl b) upgrade to a more recent kernel 1138545249 M * Bertl basically you also have option 1138545262 M * Bertl c) add the missing patches to fix that issue 1138545273 M * ScoobyD00 i just tried ext3 which also seems to have lots of failures? http://pastebin.com/528731 1138545309 M * Bertl that is at least gross ... 1138545324 M * Bertl ah, but I remember, it was partially an issue with the tools 1138545330 M * Bertl you are using 0.30.204 ? 1138545366 M * ScoobyD00 yeah 1138545389 M * Bertl well, debian is _so_ stable :) 1138545398 M * ScoobyD00 :o) 1138545419 M * Bertl okay, do you have any problems using packages from testing/unstable? 1138545438 M * ScoobyD00 yeah, i really want the base system to track stable 1138545452 M * Bertl okay, what about compiling the tools yourself? 1138545453 M * ScoobyD00 once you let something through it upgrades half the software on the box 1138545471 M * Bertl i.e. not using a package or building your own? 1138545512 M * Bertl if that's a problem too, then I can only refer you to the debian maintainer of the stable packages ... 1138545529 M * ScoobyD00 will i have to install just new tools, or new kernel patches? 1138545541 M * Bertl we are currently at 0.30.210 and kernel 2.6.14/15 1138545555 M * Bertl IIRC, working tools fix the issues on ext2 and ext3 1138545574 M * Bertl there are patches (for the debian kernels) which fix it for reiserfs too 1138545611 M * Bertl there 'should' be updated packages somewhere on the net, but don't ask me where, you have to ask micah for that 1138545658 M * Bertl btw, micah is doing the newer packages, Ola is maintaining the broken stuff :/ 1138545664 M * daniel_hozac deb.riseup.net/vserver/ ? 1138545838 M * ScoobyD00 there is 0.30.209-1 on that server - will that have the fix (or is it .210) ? 1138545931 M * ScoobyD00 the other thing is my production boxes are amd64 and there is no amd64 deb on that server 1138545980 M * ScoobyD00 although oddly enough there are references to it in files 1138546439 J * njekto ~slackware@213.151.236.164 1138546455 M * Bertl 0.30.209 should be fine 1138546473 M * ScoobyD00 wish i had used ubuntu server now :o( 1138546485 M * Bertl ScoobyD00: you might use the x86 packages on amd too, userspace is backwards compatible 1138546489 M * Bertl welcome njekto! 1138546511 M * njekto Bertl - HI 1138546527 M * ScoobyD00 bertl: good point about x86 :o) 1138546589 M * ScoobyD00 anyone know how to upgrade just 1 package in debian from "testing" without it upgrading half the system? 1138546622 M * njekto ScoobyD00 - try slackware ;] 1138546647 M * bubulak jop :) 1138546651 M * ScoobyD00 but "apt-get install slackware" doesnt seem to work :op 1138546663 M * bubulak ScoobyD00: compile that package :) 1138546664 M * daniel_hozac do people still use slackware? :) 1138546674 M * bubulak of sure 1138546676 M * bubulak :) 1138546692 M * bubulak people still use windows ? :) 1138546699 M * Bertl http://ars.userfriendly.org/cartoons/?id=20040111 1138546703 M * njekto bubulak - ;] 1138546704 M * ScoobyD00 yeah i have one on the side of my house to let the light in 1138546716 M * bubulak njekto: czech polak slovak ? 1138546724 M * njekto bubulak slovak 1138546727 M * njekto you ? 1138546731 M * bubulak tak to sme dvaja :) 1138546737 M * njekto no super 1138546739 M * njekto odkialze ? 1138546743 M * bubulak BA 1138546747 M * Bertl #offtopic :) 1138546752 M * njekto hm .. 1138546775 M * njekto tuzsimze nam nikto iny nerozumie :) hehe 1138546789 M * bubulak hmm nove mesto nad vahom :) 1138546797 M * njekto tesne vedla. 1138546799 M * njekto Stara Tura 1138546804 M * bubulak 164.236.151.213.in-addr.arpa name = ZSPS-Nove-Mesto-nV-04.orange.sk. 1138546806 M * bubulak :) 1138546807 M * Bertl folks, please 1138546817 M * bubulak ok 1138546821 M * njekto Bertl - sorry 1138546837 M * njekto ;] 1138546850 M * njekto so .. people still uses windows .. 1138546863 M * bubulak thats good ... good for us 1138546864 M * bubulak :) 1138546869 M * njekto i dont known why .. 1138546890 M * bubulak :) 1138546891 M * njekto someone know, why is windows better ? 1138546905 M * ScoobyD00 otherwise rooms get very dark 1138546918 M * bubulak is working ... games run perfectly :) 1138546930 M * njekto yes .. 1138546964 M * daniel_hozac ScoobyD00: that's why computers have LEDs ;) 1138546980 M * ScoobyD00 :o) 1138546983 M * bubulak for the light show ... i meant 1138546993 M * njekto hm .. 1138546995 M * njekto may be . 1138547032 M * ScoobyD00 right, well i cant seem to upgrade util-vserver from testing packages without it pulling in gcc4 and a whole load of other goodies which I dont want 1138547037 M * ScoobyD00 so compile from source it is 1138547075 M * Bertl okay, watch out for dietlibc and the other requirements the package has ... 1138547093 M * Bertl ScoobyD00: you can ignore the beecrypt part if you do not plan to use vhashify 1138547259 M * ScoobyD00 i wonder if i should just find a way to move to Ubuntu Server as they have 0.30.208-1ubuntu1 and i would get better updates 1138547287 M * daniel_hozac isn't that the one with the broken PIC patch? 1138547850 Q * njekto Quit: Leaving 1138548109 M * ScoobyD00 bertl: i installed dietlibc but the configure script cant find it, any ideas? 1138548119 M * daniel_hozac ScoobyD00: what version? 1138548122 M * Bertl install dietlibc-dev too 1138548160 M * ScoobyD00 thx 1138548202 M * Bertl if it still can't find any usable, it's too old 1138548227 M * Bertl micah: ping? 1138548240 M * ScoobyD00 it found it with the dev package :o) 1138548419 M * bonbons Bertl: do you know what can cause a "Permission denied" when opening /dev/pts/0 (sshd gets this in a guest, using vserver-utils) 1138548458 M * derjohn Hey, is there a (real) performance adavntage to install a host as "pure 64" linux when all guests are in 32 bit mode? 1138548477 M * Bertl bonbons: a bunch of causes come to my mind 1138548492 M * Bertl bonbons: first, check inside the guest, how the devpts is mounted 1138548506 M * Bertl (and if it is mounted at all) 1138548510 M * derjohn and: are there any issues with linux-vserver in an amd64 kernel? 1138548514 M * bonbons yep, it's mounted 1138548525 M * Bertl bonbons: do so by entering the context and get the line from /proc/mounts 1138548533 M * bonbons how, I need to check... 1138548639 M * bonbons none /dev/pts devpts rw,nodev 0 0 1138548679 M * bonbons hmm, that nodev looks bad 1138548738 M * bonbons and don't know where it comes from either, not in the config for the guest 1138548739 M * Bertl yeah, that could be tricky :) 1138548754 M * Bertl the tools probably do that _inside_ the guest 1138548775 M * Bertl but you perfectly identified the issue 1138548789 M * Bertl I guess hollow will need to fix that ... 1138548821 M * bonbons so I will need to find where that option comes from! all mounts except / in the guests are nodev 1138549015 M * daniel_hozac derjohn: only the usual performance gains, and it should work quite fine. 1138549060 M * Hollow will do.. 1138549094 M * daniel_hozac does mounting filesystems without CAP_MKNOD automatically make it nodev? 1138549118 M * Bertl no, but mounting with secure mount does 1138549173 M * derjohn daniel_hozac, hm, I meant if the guest will have more 'boom' ... if a guest is in 32bit mode, then the kernel has to switch to 32bit emulation (?) anyway. So I am asking mayelf if it's woth the effort to go to debian amd64. 1138549197 M * derjohn daniel_hozac, so the system will be in 32bits most time? 1138549203 M * Bertl derjohn: actually 32 bit guests are faster on amd64 ekrnels 1138549226 M * Bertl derjohn: because kernel stuff wills till be done in 64 bit 1138549242 M * Bertl but of course, 64bit guests will be faster than 3bit ones 1138549246 M * Bertl *32bit 1138549301 M * derjohn Bertl, kernel != "pure 64" linux. so we are talking about kernels only? The libs on the host wont bring any gain when in 32bit mode? 1138549330 M * derjohn I mean, using an amd64 complied kernel is no problem, reisntallimg debian an my hosts is effort ... 1138549351 M * derjohn *compiled 1138549352 M * Bertl it's like this: 1138549368 M * Bertl K64/U64 >= K64/U32 >= K32/U32 1138549407 M * derjohn Does this partial order counts for guest performancem, too ? 1138549426 M * daniel_hozac the host's userspace won't matter for guests, i guess. 1138549452 M * derjohn daniel_hozac, yes, this affirms my assumption ;) 1138549479 M * derjohn daniel_hozac, I was worried about libs ... but the kernel does not rely on any libc-foo, only the tools? 1138549486 M * Bertl well, there is a 'tiny' dependancy from host userspace to guest userspace 1138549504 M * Bertl IIRC, 32bit host space cannot successfully start 64bit guests :) 1138549544 M * derjohn Bertl, ahhhh .. so I should go for 64U and then I can choose lightning fast 64Gs ? 1138549552 M * derjohn G = Guest :) 1138549833 M * Bertl yeah, probably .. 1138549846 M * derjohn BTW: Someone told me, that reiserfs may have a problem with 64Bit kernels (partitions suddenly vanishing), so I go with XFS now. Can anyone confirm the problem with reiser on 64K ? 1138550423 M * bonbons Hollow: the patched vattr can be found on http://homepage.internet.lu/brunop/vattr.c (or http://homepage.internet.lu/brunop/vattr.diff as diff against HEAD) 1138550771 Q * Doener Ping timeout: 480 seconds 1138550802 J * Doener doener@i5387E9E5.versanet.de 1138550884 M * bonbons Hollow: how are you fixing the nodev issue? Just kick the automatic addition of nodev? (probably better this way, and adding nodev to fstab where it's sensible) 1138550921 M * Bertl if it really messes up things, we could make an exception for proc and devpts 1138550929 M * Hollow hm... 1138550932 M * Hollow if (strcmp(fsent->vfstype, "devpts") == 0) fsent->flags |= MS_NODEV; 1138550959 M * bonbons Hollow: I saw that line, but it does the opposite of what we want! 1138550970 M * Hollow hm 1138550974 M * Bertl heh, yeah :) 1138550985 M * Hollow &= ~(MS_NODEV)? 1138551004 M * bonbons flags = flags | MS_NODEV ... inversion of MS_NODEV is reauired 1138551027 M * Hollow in theory it should ignore nodev in fstab for devpts 1138551083 M * bonbons but even with that we block users from mounting something special to /dev or it's subfolders (unless we add option dev to vmount to disable nodev) 1138551171 M * Hollow ehm.. whe would you mount something in /dev/pts? 1138551212 M * Hollow but in general i agree, we can skip this line completely 1138551212 M * Bertl first, just to clarify that for me :) 1138551234 M * Bertl - does the devpts mount happen within the context or not? 1138551250 M * Bertl - does it happen with or without the typical caps 1138551262 M * bonbons I thought of things like mounting /dev/snd to the guests (for automatic management by host's udev or other dir that provides devices for guests and is managed by host) 1138551265 M * Hollow the mount happens outside the context 1138551269 M * Hollow just like in util-vserver 1138551286 M * Bertl okay, so a pure userspace problem then .. (please ignore me :) 1138551293 M * Hollow bonbons: yeah, /dev is alreight.. but we're talking about /dev/pts no? 1138551360 M * bonbons /dev is alright as long as it's on guest's root filesystem, if we mount something to /dev in guest (like /dev/pts) it won't work either 1138551420 M * Hollow hm, can't follow you, sorry 1138551446 M * Hollow but if it's ok by removing that line, we'll do it ;) 1138551447 M * Doener bonbons: hm, tell me if i missed something, but the above code line only affects devpts fs.. 1138551448 M * bonbons currently on a guest / allows devices, any other mount does not 1138551465 M * Doener and that is usually mounted to /dev/pts and nowhere else... 1138551503 M * bonbons that meanst that /dev must reside on guest's root filesystem, and may not be an additional mount 1138551518 M * Doener bonbons: what has devpts to do with /dev? 1138551550 M * bonbons Doener: devpts is mounted on /dev/pts, and /dev is a more general cas than only /dev/pts 1138551585 M * Doener and where is the code that sets nodev for /dev? the above line only concerns devpts 1138551621 M * bonbons Currently we have a fstab for guests that mounts / (dev,...); /dev/pts (nodev,...); /tmp (nodev,...), assume we would insert /dev (nodev) to this... 1138551644 M * bonbons note, the dev/nodev in parenthesis is what is finally executed, not what is written in the fstab 1138551675 M * Doener hm, did i miss some code then? 1138551678 M * Doener 17:08:52 Hollow if (strcmp(fsent->vfstype, "devpts") == 0) fsent->flags |= MS_NODEV; 1138551678 M * Doener 17:09:19 bonbons Hollow: I saw that line, but it does the opposite of what we want! 1138551682 M * bonbons Hollow: I think it's better not forcing anything to nodev and adjust the (default) fstab file instead 1138551705 M * daniel_hozac forcing nodev is good. it saves admins from themselves. 1138551708 M * Doener that doesn't do anything to non-devpts mounts. 1138551732 M * bonbons Doener: a few lines before there is fsent->flags = MS_NODEV; 1138551770 M * bonbons the line you saw should have reset the MS_NODEV bit 1138551844 M * bonbons To this /dev/pts issue I added the general case of anything that is intended to contain devices and is configured by host's admin for the guests (he can say nodev, but not the oposite!) 1138551903 M * Doener bonbons: ok, should've checked the code and not relied on irc pasting ;) 1138551925 M * Doener cause of the bug in the devpts line I assumed that it was intended to add NODEV there ;) 1138552024 M * Doener bonbons: and i think that nodev default is fine, cause it helps keeping the vserver safe, for any dev mount, the fstab should explicitly state that 1138552112 M * bonbons The fstab should set nodev, not vmount even when admin really doesn't want it (not possible to negate any of nodev, nosuid, no...) 1138552158 M * Doener well, mount accepts a 'dev' option, vmount could do so as well 1138552308 M * bonbons looks like there are quite a few options that mount handles, but vmount does not. I'm wondering if there isn't a way to get the complete mount operations done by standard mount command 1138552365 M * bonbons vmount whould only serve moving/copying mountpoints around (inside a namespace, or even across namespaces) 1138553105 M * Bertl okay, off for dinner now ... back alter 1138553109 M * Bertl *later 1138553114 N * Bertl Bertl_oO 1138554983 N * Bertl_oO Bertl 1138554988 M * Bertl back now :) 1138555328 Q * gerrit Quit: Client exiting 1138557026 J * oliwel ~mail-at-o@host-62-245-151-178.customer.m-online.net 1138557030 A * oliwel waves hello to the crowd 1138557064 M * oliwel anybody in ? 1138557081 M * daniel_hozac always. 1138557105 M * Bertl welcome oliwel! 1138557121 M * oliwel Hi daniel_hozac , hi Bertl 1138557146 M * oliwel short question...the 2.1.0.4 is "functional" ? 1138557146 M * FaUl hm, it's hardly oftopic here but i'll ask anyway: 1138557166 M * FaUl when trying to link make against dietlibc 1138557168 M * FaUl /usr/local/lib-sparc64/libc.a(strtod.o)(.text+0x368): In function `strtod': 1138557168 M * FaUl : undefined reference to `_Qp_qtod' 1138557176 M * oliwel Bertl: The bug with the namespaces we found in .3 is fixed already ? 1138557192 M * FaUl in lib/strtod.c is no hint of _Qp_qtod 1138557209 M * FaUl as well as of course in include there is no hint 1138557217 M * FaUl -there 1138557277 M * daniel_hozac oliwel: you should go with 2.1.0.5.1 or 2.1.0.9. 1138557281 M * FaUl someone with any idea where this _Qp_qtod may originate from? 1138557313 M * daniel_hozac FaUl: is /usr/local/lib-sparc64/libc.a dietlibc? 1138557314 M * Bertl FaUl: looks like gcc 1138557371 M * oliwel daniel_hozac: 2.1.0.4 is the latest one in the gentoo portage.... 1138557382 M * oliwel so I will push hollow ;) 1138557444 M * FaUl daniel_hozac: yes 1138557448 M * FaUl Bertl: any idea to work around? 1138557552 M * Bertl compile a proper gcc? 1138557739 M * FaUl it's gcc-3.3.5 1138558085 M * derjohn Bertl, was was about the devel release? Topic still says: devel 2.1.0. Were there objections? 1138558121 M * Bertl derjohn: hmm? 1138558267 M * derjohn Bertl, two days ago you told that I may be time for a new devel release? Or did you mean exp ? 1138558292 M * Bertl no,devel, but hey, that's nothing you do over night ... 1138558299 M * Bertl (also requires testing and stuff) 1138558341 M * derjohn Bertl, it sounded like "I am moving current exp to devel". 1138558382 M * derjohn Bertl, no offense, I was just wondering what version I will build for my new amd64 system now .. so if there was a devel before the door I may have waited 1138558416 M * Bertl I'd say, not before tuesday 1138558437 M * derjohn Bertl, thats the kind of info I was looking for :) 1138558441 M * derjohn Bertl, txh ! 1138558448 M * derjohn Bertl, I mean .. THX ;) 1138558457 M * Bertl certified? 1138558490 M * FaUl Bertl: tried 3.4.4 also, but same problem - any idea? 1138558490 M * derjohn Bertl, certified? You mean /me ? 1138558501 M * Bertl derjohn: the THX :) 1138558553 M * derjohn Bertl, on guest per loudspeaker ;) 1138558561 M * derjohn *one 1138558741 M * Bertl okay, I guess I need a nap ... (nap attack) 1138558746 M * Bertl back later ... 1138558766 N * Bertl Bertl_zZ 1138559931 J * Smutje_ ~Smutje@xdsl-87-78-40-154.netcologne.de 1138560044 Q * Smutje Ping timeout: 480 seconds 1138560123 Q * oliwel Quit: Chatzilla 0.9.69.1 [Firefox 1.5/2005111116] 1138560201 J * DoM_ ~DoM@151.56.246.70 1138560429 J * DoM__ ~DoM@151.56.206.189 1138560579 Q * `DoM` Ping timeout: 480 seconds 1138560789 Q * DoM_ Ping timeout: 480 seconds 1138562834 M * michal_ Bertl_zZ: when you are back... do you have some .config file for qemu hosts ? (ppc or i386). 1138562902 M * daniel_hozac qemu host or guest? 1138562934 M * daniel_hozac (either way, i wasn't aware you needed a special .config) 1138563224 J * Viper0482 ~Viper0482@p54976CDA.dip.t-dialin.net 1138563237 M * marl_ hi anyone tell me (or point me in the rite direction!!!!) asto what caps need to be enabled for X to run in a vserver? the site page on X says CAP_SYS_RAWIO but im still getting fdailures for APM not be acessable etc. 1138563324 M * daniel_hozac did you try to just give it all the caps to see if it's really a capability problem? 1138563348 M * marl_ is that just cap_sys_admin ? 1138563356 M * daniel_hozac no, ALL. 1138563358 M * marl_ or do i have to specify each individually? 1138563500 M * marl_ still failing with 'open APM failed (/dev/apm_biod) (No such device) 1138563526 M * marl_ this machine was copyied from a non-vserver working install into a vserver system, with a newer kernel :( 1138563533 M * marl_ is that likly to be my problem? 1138563603 M * daniel_hozac do you have APM on that machine? 1138563611 M * marl_ ah, if i try : cat /dev/apm_bios from the command linje inside the vserver i get no such device althow the device is shown within the dev directory 1138563613 M * daniel_hozac why is X trying to access APM anyway? 1138563627 M * bonbons marl_: does your new kernel still have support for APM? Might be it's now using ACPI. What feature of X does not work? 1138563639 M * marl_ not sure, but in the host i dont have the /dev/apm entry :( 1138563661 M * marl_ x wont start at all, or rather gdm wont start 1138563674 M * marl_ and now that u mention it yes it is using acpi 1138563676 M * marl_ :( 1138563906 M * bonbons who old is you X, maybe you should also remove APM-options from your X-config, if you can post your X config and X log it would help 1138563943 M * bonbons put it somewhere one the web of course, not pasting to irc 1138564256 M * marl_ ill try and copy and paste them someware in a wee bit thanks :) 1138566195 Q * Viper0482 Ping timeout: 480 seconds 1138567633 J * Aiken ~james@tooax6-178.dialup.optusnet.com.au 1138567738 M * blizz back in town... 1138568070 J * shed| ~siggi@inferno.lhi.is 1138568354 J * s4edi ~siggi@inferno.lhi.is 1138568407 Q * s4edi Quit: 1138568485 Q * shedi Ping timeout: 480 seconds 1138568755 Q * shed| Ping timeout: 480 seconds 1138569888 N * Bertl_zZ Bertl 1138569893 M * Bertl evening folks! 1138569906 M * aba evening, Bertl 1138569907 M * bonbons Evening Bertl 1138569918 M * Bertl hey aba! bonbons! 1138569976 M * bonbons Bertl, do you know internals of mount(...) call? 1138569996 M * Bertl user or kernel side? 1138570065 M * Bertl well, to answer that anyway: kernel side I should know 1138570131 M * bonbons ok, I'm trying to find out why vmount (vserver-utils) failes on the fstab-entry for /tmp if I add nodev, noatime or so, not tried on devpts, but should be same issue 1138570160 M * bonbons I would guess it's because of the flags (bits) and the data arguement with overlapping information 1138570194 M * Bertl how do you call the sys_mount()? 1138570219 M * bonbons is this allowed, or must userspace filter out options that are in flags so they don't appear in data? [man 2 mount()] 1138570284 M * Bertl the sys_mount() gets the following 'arguments' 1138570295 M * Bertl dev,dir,type,flags,data 1138570308 M * Bertl dev,dir and type ar char* 1138570321 M * Bertl flags are long and data is a void* 1138570345 M * bonbons the two args that probably cause "problems" are flags and data 1138570366 M * bonbons can e.g. nodev appear in flags AND in data? 1138570389 M * Bertl no, that is wrong, IIRC 1138570404 M * Bertl the data is only used for binary type of mounts 1138570441 M * bonbons Ok, then vmount must be patched to filter out those items it puts into flags 1138570470 M * bonbons some work to get the nodev issue cleanly fixed 1138570482 M * Bertl the data only makes sense for remount and new mounts 1138570499 M * Bertl and it is basically parsed by the filesystem 1138570549 M * bonbons does userspace have to do anything special to the options that got to data? (','-separated list of items as for mount's -o option?) 1138570745 M * Bertl basically that depends on the filesystem 1138570781 M * Bertl get_sb(type, flags, name, data) 1138570798 M * Bertl where type,name and data is passed unmodified 1138570807 M * Bertl only the flags have got pre-processes 1138570812 M * Bertl *processed 1138570813 M * bonbons will assume that for now, exceptions may be added later on :) 1138570832 M * Bertl best way is to check with strace no? 1138570911 M * bonbons or look at mount's code (will trust the man page where it says that for most filesystems it's a ','-separated list) 1138570978 M * marl_ hi again, ive disabled all referances to apm in the X11 directory, im still getting an apm error, but im also getting a /dev/mem error (operation not permitted) in the X page on the site is the follow line about /dev/mem : (note: Xorg on debian sid require a /dev/mem -- NebuchadnezzaR?) anyone tell me how to allow a vserver access to /dev/mem, or if there is an alternative to this that can by sym linked or something? 1138570984 M * Bertl bonbons: mount("none", "/test", "tmpfs", MS_NODEV|MS_POSIXACL|MS_ACTIVE|MS_NOUSER|0xec0000, 0x517290) = -1 EINVAL (Invalid argument) 1138571001 M * Bertl for example is called by: mount -t tmpfs -o nodev,direct none /test 1138571068 M * bonbons and the 0x517290 is for the 'direct' option? 1138571072 M * Bertl marl_: you need to make it visible via the proc hiding 1138571085 M * Bertl bonbons: I'd assume that ... 1138571108 M * Bertl bonbons: my strace doesn't show the data page 1138571160 J * shedi ~siggi@inferno.lhi.is 1138571162 M * bonbons I'm will check in mount's code, I have it open for reading :) 1138571282 M * bonbons but I'm wondering if it isn't better to mostly wrap a call to real mount command by changing the source/target paths 1138571310 M * Bertl would not be the worst idea, as there potentially is a mount for each filesystem 1138571379 M * Bertl /sbin/mount. 1138571385 M * bonbons that would avoid any problems related to interpreting options, we could trap some special items on fs-type (like future cross-namespace binds/moves) 1138571464 M * bonbons not all filesystems have a /sbin/mount., only those needing special attention 1138571867 Q * shedi Quit: Leaving 1138571950 J * vex ~vdjerek@student.fizika.org 1138571999 M * vex hi there! i have a question, if you don't mind :) 1138572049 M * vex my vserver build (debootstrap) freezes on setting up sysvinit (when creating /dev/initctl) 1138572401 M * Bertl welcome vex! 1138572425 M * vex hi there! nice to see you 1138572429 M * Bertl vex: a few details, like tool and ekrnel aptch versions? 1138572453 M * Bertl (if something in my sentences looks wrong, try to swap a few letters :) 1138572488 M * Bertl well, that is truly ancient :) 1138572507 M * Bertl you are on debian, right :) 1138572517 M * vex yup, but it worked. and i was able to create quite a few vservers 1138572545 M * vex yup. but this isn't default debian vserver package, i patched the kernel a while ago 1138572549 N * ebiedermZz ebiederm 1138572551 M * Bertl yes, I'd say it should work .. at least the creation, how do you create it? 1138572562 M * vex ok, here it goes :) 1138572564 M * Bertl morning ebiederm! 1138572579 M * ebiederm afternoon Bertl. 1138572598 M * vex i have a dumb shell script... 1138572601 M * vex --context $5 -- -d $6 -m http://ftp.hr.debian.org/debian 1138572603 M * vex sorry :) 1138572628 M * vex /usr/sbin/vserver $1 build -m debootstrap --hostname $2 --netdev $3 --interface $4 --context $5 -- -d $6 -m http://ftp.hr.debian.org/debian 1138572640 M * vex i use sarge 1138572647 M * Bertl okay, looks fine ... 1138572665 M * vex the thing is that it sometimes sucessfully creates a vesrver, and sometimes fails spectacularly 1138572668 M * Bertl so I'd assume something in debootstrap has changed 1138572673 M * vex pretty much by random :) 1138572695 M * vex i created a vserver 5 days ago 1138572696 M * Bertl nevertheless I'd suggest to move on to 0.30.210 (or at least 209) 1138572709 M * vex now i can, and a mont ago i had the same problem as now 1138572722 M * Bertl just to verify that this isn't a tool issue 1138572737 M * vex ok, i will. just thought that there was a quick fix for that 1138572751 M * Bertl it's the first time I heard about that ... 1138572752 M * vex thanks! 1138572764 M * Bertl maybe ask aba, if he is still aroun 1138572784 M * vex :) try http://www.sculpturedlife.com/vserver/irc-logs/index.php?date=2004-09-03 1138572820 J * shedi ~siggi@inferno.lhi.is 1138572853 M * vex btw, all of my working vservers don't have /dev/initctl 1138572868 M * Bertl vex: wow, your google foo is obviously better than my memory 1138572871 M * vex and the build fails when trying to make one for the new vserver 1138572944 M * vex :) i tried fixing the thing myself, but thought, what the hell, let's try asking the masters :) 1138572995 M * Bertl well, let's give the tool update a try, if the issue persists, we ask the debian masters :) 1138573020 M * Bertl btw, updating to (at least) the stable 2.0 release would be a good idea too 1138573030 M * blizz Bertl, btw, everything works fine now with util-vserver! finally ;) 1138573041 M * blizz and it works great :) 1138573046 M * Bertl excellent! 1138573063 M * vex do i have to rebuild/patch a kernel to install new util-vserver? 1138573084 M * Bertl no, the tools will even work with 2.4 kernels 1138573103 M * Bertl (just a matter of configuration) 1138573107 J * Johnnie ~jdlewis@dynamic-acs-24-154-53-16.zoominternet.net 1138573109 M * vex ok, thanks! will give it a try. if it doesn't work, i'll buiild a new kernel and let you know 1138573133 M * Bertl vex: if you build 210, make sure to specify the NOLEGACY API 1138573139 M * vex btw, vserver kix ass :) thanks 1138573153 M * vex Bertl: ok 1138573159 M * Bertl yes it does! you're welcome! 1138573170 M * Bertl feel free to hang around ... 1138573434 Q * ScoobyD00 Quit: 1138573446 M * bonbons Bertl: The BME patch, what mount-flags does it enforce for binds-mounts? ro/rw, also (no)dev, (no)suid, (no)exec, (no)atime, ...? 1138573468 M * Bertl yes, but a lot of them are already there 1138573506 M * Bertl i.e. vanilla already supports (no)exec, no(atime) 1138573525 M * Bertl not sure about (no)suid and (no)dev 1138573550 M * bonbons so the various man-pages are already out of date/incomblete? (mount(2), mount(8)) 1138573609 M * Bertl hmm, why? 1138573644 M * Bertl I don't see anything obviously wrong in mine ... 1138573685 M * bonbons it states that flags and data are ignored for bind mounts (2.4 onwards) 1138573701 M * bonbons mount(2), MS_BIND 1138573705 M * blizz should i delete all the halt/reboot stuff from my distro's init scripts? 1138573731 M * Bertl hmm, mine do not state that ... 1138573751 M * Bertl blizz: guest side I assume? 1138573755 M * blizz yepa 1138573772 M * Bertl blizz: well, really depends on how you setup the guest 1138573781 M * blizz mhh, which means? 1138573802 M * Bertl blizz: if you want it to shut down (using init for example), you have to keep some shutdown scripts 1138573815 M * Bertl blizz: in general you want to remove the hardware related parts 1138573898 M * bonbons Bertl: my man pages are from man-pages-2.20 (Gentoo), with mention of 2.6.12, May 2004 at the bottom of mount(2) 1138573931 M * Bertl man-pages-1.48-1mdk 1138573966 M * Bertl so they were broken later :) 1138574011 M * Bertl especially --bind uses the flags for different things 1138574036 M * blizz i c 1138574039 M * Bertl e.g. to tell if it is a recursive bind or not 1138574072 M * Bertl but it's correct that the other flags are 'ignored' without BME (in some way) 1138574093 M * vex Bertl: i just updated util-vserver to 0.30.209...still doesn't work :( 1138574105 M * bonbons and even to say it's a bind mount, unless the fstype "bind" was used for this... 1138574123 M * Bertl vex: okay, so no tool issue, which doesn't sound too bad (you are using dietlibc, yes?) 1138574203 M * vex Bertl: uhm, no...should I? 1138574214 M * Bertl well, the tools warned you, no? 1138574231 M * Bertl (check vserver-info - SYSINFO) 1138574232 M * vex Bertl: no, i used the precompiled version :) 1138574275 M * vex Bertl: but it seems that whoever compiled it didn't use dietlibc 1138574281 M * vex Bertl: Use dietlibc: no (you have been warned) 1138574288 M * Bertl well, send a complaint :) 1138574302 M * vex Bertl: :) 1138574315 M * Bertl and please retry with proper tools 1138574348 M * vex Berrtl: sure, sorry 1138574367 M * Bertl nothing to be sorry about, but you should really contact the packager 1138574492 M * vex i followed a link from http://linux-vserver.org/ to www.backports.org 1138574505 M * bonbons Bertl: that reminds me the bad syscall issue with omitfp thingie :) 1138574560 J * bwana ~bwana@71-82-125-40.dhcp.roch.mn.charter.com 1138574569 Q * Johnnie Quit: G'bye! 1138574622 M * Bertl vex: if you think it's inappropriate, please remove the link there :) 1138574711 M * vex Bertl: ok, found it...Ola Lundqvist is the maintainer..should I contact him, or is he also hanging arround here? :) 1138574763 M * Bertl no he is not hanging around here, he jsut produces broken packages :) 1138574773 M * vex hehe 1138574773 M * Bertl yes, please contact him :) 1138574868 M * derjohn Bertl, vex, util-vserver in etch and sid work. 1138574868 M * bwana hmm, so what is the difference between util-vserver and vserver-utils 1138574894 M * derjohn I check etch today ... the -209 hit etch (copied from sid) 1138574909 M * derjohn bwana, different things. 1138574930 M * derjohn bwana, stay with util-vserver and it runs. 1138574963 M * derjohn bwana, vserver-utils are debian centric perl based scripts ... I dunno if they run with newstyle config at all 1138574977 M * derjohn bwana, but if you find out: tell us ;) 1138575022 M * bwana ahh ok i see, thanks derjohn 1138575023 M * bonbons derjohn, which vserver-utils do you talk about? Not hollow's, those are c+bash 1138575151 M * derjohn bonbons, ... eh .... uh .... when I started with vserver there were those .deb packages. I dont think hollows stuff run on debian at all? I thought they are gentooish ... and besides that debian wouldnt package unmature-stuff. The packahe is simpley called "vserver-utils" 1138575244 M * Doener derjohn: you mean vserver-debiantools? 1138575253 M * bonbons there are some Gentoo traces in there, but thoses are fading away (distro-specific code is either in samples or just being removed for packagers to do) 1138575348 M * derjohn Doener, ?rcks ... mom ... 1138575472 M * derjohn I have objections to my own utterings .... Doener is right, I was talking about 'vserver-debiantools'. But nevertheless you should get happy with util-vserver only. 1138575555 M * Doener bwana: util-vserver is the current toolset, vserver-utils is a replacement being developed by Hollow, which is not ready for real usage yet AFAIK 1138575565 M * bonbons yep, vserver-utils is still too much in development for thinking about production use. In addition it is targetted at newest development kernel patches 1138575638 M * Doener bonbons: hm, wasn't it targetted at anything >=2.0? i.e. only the recent syscall api? (v13 or whatever it is called) 1138575657 M * Hollow libvserver is for >=2.0, but vserver-utils needs at least 2.1.0.4 1138575736 M * Doener ah ok 1138575751 M * Hollow due to persistant nx 1138575777 M * derjohn tools utils vserver ... I agree that all this may be confusing. Or I am confused. 1138575793 M * bonbons Hollow is back :) for the case you missed/skipped part of my discussion with Bertl, I'm changing vmount to call real mount (doesn't handle options too well) 1138575822 M * Hollow ok 1138575884 M * Doener derjohn: don't forget that the first toolset was called vserver-tools ;) the fourth incarnation is doomed to be called tool-vservers ;) 1138575890 M * bonbons this way we don't need to care about any option that may appear for special filesystems 1138575943 M * Hollow Doener: the next rewrite will be named wrzlbrmpft 1138575963 M * Doener written in brainf...? 1138575973 M * Hollow e.g. 1138575974 M * Hollow :D 1138575990 M * Doener sounds good :) 1138576230 M * michal_ lol 1138576244 M * michal_ you guys go and rest a bit mayby ;] 1138576273 J * ataraxis ~ataraxis@p54AC20BD.dip0.t-ipconnect.de 1138576344 M * Bertl welcome ataraxis! 1138576354 M * ataraxis Hi Bertl 1138576398 M * ataraxis i'm probably in the wrong place when i have a problem as a user on a vserver (from a vserver company), am I? 1138576414 M * Bertl well, not really ... 1138576429 M * Bertl you might have a good chance the provider is here somewhere :) 1138576452 M * ataraxis hehe :) 1138576490 M * ataraxis after a kernel upgrade my postfix wants to use ipv6: fatal: lmtp_virtual_bind: getaddrinfo("::1"): Address family for hostname not supported 1138576521 M * ataraxis is this due to a postfix error or because inet Adresse:127.0.0.1 Maske:255.0.0.0 is missing in ifconfig lo 1138576533 M * blizz g'nite 1138576620 M * Bertl ataraxis: well, neither nor, actually ipv6 is not supported inside a guest 1138576645 M * Bertl ataraxis: now, the fault is by the provider (to some extent) as he should not ahve compiled in ipv6 in the first place 1138576665 M * ataraxis then I better send him a trouble ticket :) 1138576669 M * Bertl ataraxis: OTOH, your postfix should not use ipv6 addresses 1138576674 M * bonbons ataraxis: you should tell postfix explicitly on which addresses to listen, I have mine in a guest with just a single non-lo address 1138576697 M * ataraxis inet_protocols = ipv4 <- that didn't work 1138576705 M * Bertl well, at least it should not guess ipv6 addresses, did you check your /etc/hosts, imaybe that contains some entries 1138576731 M * ataraxis /etc/hosts does contain both ipv4 and ipv6 entries 1138576762 M * Bertl okay, remove the ipv6 entries there ... 1138576766 Q * mugwumpjism Quit: leaving 1138576768 M * ataraxis bonbons: myhostname = m29s02.xxx.de <- that should do it? 1138576788 M * bonbons if it resolves to something correct 1138576862 M * ataraxis yes, it resolves from the inside and the outside 1138576891 M * bonbons I just told it inet_interface = 123.456.78.9, you should not need anything more (unless your guest gets dynamic IP address) 1138577032 P * meandtheshell 1138577077 M * ataraxis bonbons: thank you, it flushes now :) 1138577098 M * ataraxis and i keep the ipv6 entries commented out 1138577158 M * vex Bertl: I've just build 0.30.210 with dietlibc, still doesn't work 1138577171 M * vex Bertl: probably debootstrap is fscked 1138577203 M * Bertl could be .. wanna try a newer kernel too? 1138577229 M * bonbons ataraxis: I have the IPv6 entries in my hosts file and it doesn't cause trouble, resolving the hostname may show the compiled-in support for IPv6 though (so better work with fixed IP's when possible) 1138577271 M * Bertl vex: check with testme.sh and testfs.sh for good reasons to upgrade .) 1138577272 M * vex Bertl: tomorrow probably. but i have a feeling this is a debian issue, since debootstrap basically dies... 1138577356 M * vex vex: old bugs? 1138577372 M * vex Bertl: i meant Bertl : 1138577378 M * vex Bertl: :) 1138577414 M * vex Bertl: btw, this is what it look when it freezes: 1138577423 M * vex Bertl: Setting up sysvinit (2.86.ds1-1) ... 1138577435 M * vex Bertl: sysvinit: creating /dev/initctl 1138577443 M * vex Bertl: and then nothing :) 1138577522 M * Bertl hmm, yeah, looks like debootstrap 1138577561 M * Bertl funny that I never encountered that myself 1138577577 M * Bertl I mean, all debian guests I created didn't show that issue 1138577577 J * monrad_ ~mikkel@213083190131.sonofon.dk 1138577586 Q * monrad Ping timeout: 480 seconds 1138577763 M * vex It's probably something incredibly stupid :) 1138577841 M * vex btw , my working vservers don't have /dev/initctl, and sysvinit is installed 1138577858 M * Bertl that could have different reasons 1138577869 M * Bertl could be that the tools removed it afterwards 1138577896 M * bonbons or even some guests just use a dummy init 1138577960 M * derjohn vex, I did a debootstrap with yesterday's sid debootstrap without problems (installed an ubuntu guest ... works like charm) 1138578054 M * derjohn vex, something is strange with you system. i re-read chatlog ... could it be that you have partially installed local util-vserver? what does `which vserver`tell you? 1138578069 M * vex derjohn, i understand that. my setup was working just fine when suddenly one day it decided not to work anymore...as far as i know, nothing was changed on server in the meantime 1138578070 M * derjohn vex, do the testme scritps run? 1138578092 M * derjohn vex, so I could only be apt ? 1138578122 M * Bertl it could be a kernel issue too 1138578151 M * vex well, it worked before with the same kernel... 1138578152 M * derjohn vex, did you upgrade the kernel ? i mean you said nothing was changed 1138578168 M * vex no, i didn't touch the kernel 1138578171 Q * michal_ Ping timeout: 480 seconds 1138578179 M * derjohn well, when I worked last, which kernel version was it? 1138578193 M * derjohn *it 1138578210 M * vex it worked 5 days ago :)2.6.10-vs1.9.4 1138578230 M * vex i know i should upgrade and then talk to you... 1138578266 M * vex :) 1138578275 M * derjohn vex, yes, 1.9.5 is end-of-life ;) 1138578301 M * derjohn but anyway, the machines doesnt stop by random just to tease you ... 1138578325 M * vex i know that, but you know unix/linux servers...they sometimes seem just to do that :) 1138578346 M * derjohn but to get you problem solver I would install a new kernel (well, at least you could try my ready made .deb ... just dpkg it an reboot) 1138578370 M * derjohn vex, yes, they are sometimes hit by nap attacks 1138578391 M * vex yup, i saw your debs, will try them tomorrow 1138578433 M * vex until then, good night...it's late, have to get up early :( 1138578435 M * derjohn well, it PIII only and 2.6.14 (only) but with devel 2.1.0 wich is current. This week I will produce a new series with new kernels. 1138578441 M * derjohn yes 1138578461 M * vex thanks, bye! 1138578463 M * Bertl vex: good night! 1138578463 M * derjohn and dont forget to apt 'yaird' t get your bootdisk running 1138578466 M * derjohn vex, n8 1138578484 Q * vex Quit: [BX] Just do it like Nike... BEEATCH! 1138578648 Q * ataraxis Remote host closed the connection 1138578749 J * michal_ ~michal@www.rsbac.org 1138578758 Q * bonbons Quit: good night everyone