1162426130 J * lilalinux_ ~plasma@dslb-084-058-196-255.pools.arcor-ip.net 1162426562 Q * lilalinux Ping timeout: 480 seconds 1162427110 Q * ntrs Quit: Leaving 1162427326 M * derjohn2 daniel_hozac, hi! Hm, the permission denied ... i cannot remeber what that caused. I'll try some possibilties ... 1162427344 P * schMA Konversation terminated! 1162427345 M * daniel_hozac ok. 1162427388 M * daniel_hozac btw, ltrace might be able to tell us which apr call is failing, the is_directory or the pathmerge. 1162427620 M * derjohn2 i tried an ltrace already ... but didnt find the switch to output to a file (no joke ... even if I am not blonde) 1162427781 M * ray6 -o? 1162427850 M * daniel_hozac ltrace -f -o ..., most options are similar to strace. 1162427943 M * ray6 BTW, any x86/kernel pros here? I wonder why my machine went SLOW after I upgraded to 4G RAM, but is OK again when I boot with mem=3980M? I don't use PAE, but not just I/O got slow but everything (openssl speed by a factor of 10-100...) 1162428025 M * daniel_hozac so 4G HIGHMEM? 1162428273 M * ray6 hmm. I'm not sure I must admit. Tried RH stock kernel and xen %-) (all my vservers run in Xen, but those kernels only have 256M-1G or so RAM). 1162428296 M * daniel_hozac stock should be 4G HIGHMEM, IIRC. 1162428313 M * daniel_hozac (since PAE breaks on some hardware, and 960 MiB isn't enough) 1162428347 M * ray6 yeah, but then they build xen as PAE only (installed FC6 on my Centrino Notebook and couln't boot it, haha :) 1162428371 M * daniel_hozac ah? that's new in FC6, i think. 1162428381 M * daniel_hozac FC5 has a separate PAE kernel. 1162428390 M * ray6 dan: I installed directly to Xen - the xen kernel is PAE only 1162428434 M * ray6 (the FC6 one. And when you check "virtualization" during install, only Xen kernels are installed. bad idea, IMHO) 1162428451 M * daniel_hozac really? yeah, that sounds bad. 1162428510 M * ray6 but back to the 4G thing, I found a LKML thread which reported 30x slowdown with HIGHMEM 4G in 2.4, could be the same problem? But why does sparing just a few MB out help? 1162428542 M * ray6 I mean, I have no problem running with mem=3980 and lose about 32 MB, but I'd like to understand why :) 1162428555 M * daniel_hozac yeah, that seems odd. i don't really know much of anything about how it works though... 1162429872 Q * sladen Ping timeout: 480 seconds 1162429938 J * sladen paul@193.28.45.41 1162429963 Q * s0undt3ch Ping timeout: 480 seconds 1162429979 M * doener ray6: probably a mtrr problem 1162430005 M * doener ray6: could you provide the contents of /proc/mtrr 1162430333 M * ray6 hm, would have to boot back to the stock kernel then, but that could be caused by a buggy or old BIOS when I understand this correctly/ 1162430373 M * doener it should be ok to provide the contents of /proc/mtrr of your currently running kernel 1162430445 M * ray6 that's with mem= option on boot 1162430553 J * s0undt3ch ~s0undt3ch@81.193.56.172 1162430599 M * doener so the area from 4000MB-4096MB is not accessible via mtrr 1162430648 M * doener the dmesg output might have further information on the why (maybe eg. graphic memory is mapped there?) 1162430657 M * ray6 might be 1162430674 M * cehteh not a 64 bit capable cpu? 1162430676 M * ray6 and what would the machine do then to get that slow? 1162430722 J * ntrs ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1162430731 M * doener (heavily) access memory in that area. the mtrr stuff helps to speed up memory accesses in some way, about which I have no clue ;) 1162430742 M * ray6 I currently don't want to use 64Bit there... neither do I want to use PAE. 1162430811 M * cehteh ray6: you can use a 64 bit kernel and still running 32 bit userland 1162430831 M * cehteh that would quite likely solve your problems 1162430838 M * ray6 I know, but with a few problems IIRC 1162430848 M * cehteh i have none by that 1162430860 M * ray6 some structures are incompatible between 32bit userspace tools and 64bit kernel IIRC 1162430866 M * cehteh well i am running a real 64 bit system and 32 bit only in a vserver 1162430880 M * ray6 iptables or something for example iirc 1162430902 M * cehteh maybe 1162430943 M * cehteh well 64 bit host and 32 bit vserver really works for sure 1162430988 M * cehteh having anything above 3GB on a 32 bit machine is insane imo .. x86 isnt meant for that 1162431006 M * cehteh and all fixes for that are ugly 1162431019 M * cehteh (prolly even more than 2GB) 1162431029 M * cehteh it works .. but it is no fun 1162431041 M * ray6 ceh: I use xen on the host and vserver only in the xen domUs, no domU has more than 2GB so I have no issues with the memory split 1162431074 M * cehteh yes i am not talking about memory split .. thats a pure virtual memory thing anyways 1162431116 M * ray6 but about what? DMA? 1162431141 M * cehteh but a 32 bit kernel cant easily handle 4GB physical memory ... afaik you have to use PAE since some PCI registers are mapped above 3GB 1162431180 M * cehteh or you loose some physical memory .. and physical memory is slightly fragemented dunno exactly 1162431182 M * ray6 the kernel which handles the PCI runs in xen, hm, but I have to admit I'm not really sure how xen handles this 1162431210 M * cehteh the dom0 kernel has to handle 4GB .. and thats the problem 1162431222 M * cehteh oe xen maybe .. dunno 1162431236 M * ray6 ce: no, Xen has to handle the 4G. The domU has just 128MB but handles the PCI 1162431241 M * cehteh but anyways .. you actually see that it doesnt work very well ;) 1162431247 M * ray6 s/domU/dom0/ 1162431249 M * cehteh yes 1162431271 M * cehteh can you boot Xen with mem=3.5GB or such and try it 1162431277 M * ray6 cet: it works well when I specify mem=39XXM on boot. I'm trying 3999 now as that would be the end of the mtrr table :) 1162431281 A * cehteh has no xen experience 1162431297 M * cehteh then do that 1162431299 M * doener ray6: mem=4000M even ;) 1162431321 M * cehteh loosing a few mb is likely the simplest solution 1162431322 M * ray6 ceh: I also have the issue when using a stock RH kernel. And I already found out that mem=xxxx helps yesterday, I'm just in the process of understanding WHY it helps 1162431418 M * ray6 and where the limit is. But that mtrr thing looks reasonable 1162431636 J * eyck_ eyck@ghost.anime.pl 1162433571 Q * FireEgl Quit: Bye... 1162434969 J * adrien43654765 ~vserver@modemcable190.209-201-24.mc.videotron.ca 1162435006 M * adrien43654765 Ok I've just created my first vserver using the yum build method 1162435017 M * adrien43654765 my server has a local ip 1162435023 M * adrien43654765 but when I ssh to it 1162435034 M * doener you need to limit the host's sshd 1162435035 M * adrien43654765 it works but I end up inside the host and not the guest 1162435046 M * adrien43654765 ok 1162435054 M * doener eg ListenAddress 192.168.0.101 1162435059 M * adrien43654765 I see 1162435059 M * doener in /etc/ssh/sshd_config 1162435063 M * adrien43654765 thanks 1162435070 M * doener vservers are limited automatically, but the host is not 1162435074 M * doener you're welcome! 1162435079 M * adrien43654765 do I have to do the same for all the applications running on the host ? 1162435102 M * adrien43654765 anyway it's not good to have alot running on it 1162435133 M * doener yes. most folks run only few things (often almost nothing) 1162435138 M * doener ... on the host 1162435192 M * adrien43654765 can I ask you what do you use vserver for ? 1162435195 M * Aiken Bertl daniel_hozac that dietlibc patch did not work with the version of dietlibc I was using (0.29) but extacting the source tar ball from the mandrake rpm did work, patch not needed 1162435287 M * doener adrien43654765: I use it to mostly to create separate guests as development testing environments 1162435346 M * doener eg. my host runs Debian Sid, but programs need to work on Debian Sarge, so I use the vserver to test that. Or create a copy of the working environment of the target server 1162435437 M * adrien43654765 ok 1162435448 M * doener or to create separate areas on a production server, to lock away services or provide different vservers for different tasks the box is serbing 1162435454 M * doener s/serbing/serving/ 1162435484 M * adrien43654765 ok, I want to use it to do voip with asterisk 1162435488 M * adrien43654765 I've been using xen 1162435502 M * adrien43654765 it works but not completely 1162435519 M * adrien43654765 and with vserver I should be able to have more servers inside a box 1162435554 M * doener IIRC someone has built a complete appliance that also supports voip/asterisk where services are separated with vservers 1162435593 M * adrien43654765 do you have a name ? 1162435681 M * doener on irc it's pflanze, not sure when he showed up the last time 1162435717 M * adrien43654765 ok thanks ! 1162436140 Q * ensc Killed (NickServ (GHOST command used by ensc_)) 1162436151 J * ensc ~irc-ensc@p54B4D6FC.dip.t-dialin.net 1162436239 J * s0undt3ch_ ~s0undt3ch@81.193.56.89 1162436353 M * adrien43654765 fatal: PAM: pam_open_session(): Cannot make/remove an entry for the specified session 1162436372 M * adrien43654765 now, I can't login through ssh, I think I missed a step 1162436401 M * doener hm, might be a wrongly set barrier 1162436415 M * doener are your vservers in /vservers? 1162436448 M * adrien43654765 yes 1162436528 M * doener ok, "showattr -d /vservers" should output a line where the fields at the beginning of the line contain a "B" (upper case) 1162436553 M * doener while "showattr -d /vservers/yourvserver" should have a "b" (lower case) there 1162436575 M * doener also, permissions on /vservers/yourvserver should be 0755 1162436675 Q * s0undt3ch Ping timeout: 480 seconds 1162436675 N * s0undt3ch_ s0undt3ch 1162437222 M * doener 4am, good time to go to bed.... g'night! 1162437365 M * Skram peace 1162437712 M * adrien43654765 good night thx 1162438362 Q * Rusty`` Ping timeout: 480 seconds 1162438840 Q * duckx Quit: Client exiting 1162438957 J * duckx ~Duck@tox.dyndns.org 1162439192 Q * Rusty` Ping timeout: 480 seconds 1162439667 Q * adrien43654765 Ping timeout: 480 seconds 1162443934 J * FireEgl FireEgl@Sebastian.Atlantica.US 1162447543 J * node_ ~dwindsor@c-69-143-154-220.hsd1.md.comcast.net 1162447630 M * Hollow and still no 2.1.1 ... 1162447810 Q * bronson Quit: Ex-Chat 1162449658 J * ^mBtlHTb\ ~hollow@styx.xnull.de 1162449788 Q * ^mBtlHTb\ Remote host closed the connection 1162449929 J * OGHnqoltH ~hollow@styx.xnull.de 1162449964 Q * Hollow Read error: Connection reset by peer 1162449975 N * OGHnqoltH Hollow 1162450162 J * WOvcXiWfh ~hollow@styx.xnull.de 1162450163 Q * Hollow Read error: Connection reset by peer 1162450729 J * Piet_ hiddenserv@tor.noreply.org 1162450856 Q * node_ Remote host closed the connection 1162450871 J * node_ ~dwindsor@c-69-143-154-220.hsd1.md.comcast.net 1162451148 Q * Piet Ping timeout: 480 seconds 1162451726 J * meandtheshell ~markus@85-125-231-128.dynamic.xdsl-line.inode.at 1162451918 J * dna_ ~naucki@58-205-dsl.kielnet.net 1162451965 J * bronson ~bronson@c-71-198-75-160.hsd1.ca.comcast.net 1162452063 Q * eyck_ Quit: leaving 1162452066 J * eyck eyck@ghost.anime.pl 1162452781 N * WOvcXiWfh Hollow 1162453246 M * Borg- moin 1162454890 Q * dna_ Quit: Verlassend 1162455101 Q * ruskie Remote host closed the connection 1162455296 J * ruskie ~ruskie@ruskie.user.oftc.net 1162455946 Q * ij Quit: screen update 1162456100 J * bonbons ~bonbons@83.222.36.111 1162456444 Q * Piet_ Quit: Piet_ 1162456622 Q * node_ Ping timeout: 480 seconds 1162456962 J * SoftIce ~newbie@vc-196-207-45-253.3g.vodacom.co.za 1162456975 M * SoftIce howdie, having an issue 1162456993 M * SoftIce [root@blackhole /]# vserver gast build -m yum --context 42 --hostname=ns1.security.web.za --interface gast0=eth0:65.75.175.41/24 -- -d fc5 1162456993 M * SoftIce /etc/vservers/.defaults/vdirbase/gast: Function not implemented 1162457068 J * Alfred ~Alfred@83-215-36-12.bruck.dyn.salzburg-online.at 1162457358 P * Alfred Kopete 0.11.1 : http://kopete.kde.org 1162457937 J * Alfred ~Alfred@83-215-36-12.bruck.dyn.salzburg-online.at 1162457941 P * Alfred Kopete 0.11.1 : http://kopete.kde.org 1162458027 Q * lylix Ping timeout: 480 seconds 1162458036 J * SoftIce_ ~newbie@vc-196-207-45-253.3g.vodacom.co.za 1162458041 M * SoftIce_ disconnected 1162458047 M * SoftIce_ any reply back on why its doing that? 1162458212 Q * SoftIce Ping timeout: 480 seconds 1162460025 J * dna_ ~naucki@22-231-dsl.kielnet.net 1162460400 Q * SoftIce_ 1162461655 M * renihs nope 1162461778 J * nayco_work ~nayco@proxy2.laroche.univ-nantes.fr 1162461785 M * nayco_work helo, all :-) ! 1162461791 J * comfrey ~comfrey@87-194-46-50.bethere.co.uk 1162462271 M * nayco_work Bertl_zZ: I get an error while building util-vserver-0.30.210-5mdk.src.rpm on Mandriva 2007... : 1162462280 M * nayco_work http://paste.linux-vserver.org/614 1162462569 Q * michal` Ping timeout: 480 seconds 1162462737 M * nayco_work Bertl_zZ: It seems (thanks to google) that there is a "-lutil" missing somewhere... 1162462766 Q * sebastian 1162463063 M * nayco_work Bertl_zZ: I tried with 4mdv : The build succeeds ! (LC_ALL=C rpm --rebuild util-vserver-0.30.210-4mdk.src.rpm) 1162463357 M * nayco_work Bertl_zZ: Another thing : doing "rpm -ivh util-vserver-0.30.210-5mdk.src.rpm" created a /usr/src/rpm/ tree, with the source tarballs in /usr/src/rpm/SOURCES. I think this is an error, since the usual directory should is named /usr/src/RPM ... 1162463421 M * nayco_work Bertl_zZ: Another : I too get a /usr/src/debug/util-vserver-0.30.210/ tree :-O ! 1162463463 M * nayco_work What is the purpose of this directory (well... debug, ok ;-)), and is it the good path ? 1162463758 J * michal` ~michal@www.rsbac.org 1162463808 J * kliknet ~linux@202.149.93.73 1162463816 P * kliknet 1162466687 Q * Aiken Ping timeout: 480 seconds 1162469047 Q * nayco_work Ping timeout: 480 seconds 1162470383 Q * mugwump Quit: Reconnecting 1162470389 J * mugwump ~samv@watts.utsl.gen.nz 1162470682 J * nayco_work ~nayco@proxy2.laroche.univ-nantes.fr 1162471212 Q * nayco_work Ping timeout: 480 seconds 1162472531 Q * s0undt3ch Quit: leaving 1162472547 J * s0undt3ch ~s0undt3ch@bl4-56-89.dsl.telepac.pt 1162472661 Q * comfrey Read error: Operation timed out 1162473192 J * comfrey ~comfrey@87-194-46-50.bethere.co.uk 1162473429 Q * comfrey 1162473450 J * comfrey ~comfrey@87-194-46-50.bethere.co.uk 1162473590 J * anonc ~anonc@staffnet.internode.com.au 1162474337 J * nayco_work ~nayco@proxy2.laroche.univ-nantes.fr 1162474447 M * daniel_hozac nayco_work: you're aware that 0.30.211 is the latest util-vserver release, right? 1162474476 M * daniel_hozac and installing a source RPM will put the files in whatever directories you have defined, it's not influenced by the source RPM at all. 1162474545 M * daniel_hozac and /usr/src/debug looks like the directory automatically generated by the RPM scripts creating debuginfo packages. 1162474598 J * marcfiu ~mef@targe.CS.Princeton.EDU 1162474617 M * marcfiu hello 1162474641 M * daniel_hozac hi 1162474649 M * marcfiu daniel_hozac: have you tried patch vs2.0.2.2-rc4 against kernel-2.6.18-1.2200.fc5 1162474682 M * daniel_hozac well, i have kernel-2.6.18-1.2209.fc5.vs2.0.2.1 in the repo, IIRC. 1162474698 M * daniel_hozac kernel-2.6.18-1.2209_FC5.vs2.0.2.1.i686.rpm 1162474705 M * daniel_hozac but that's just -rc3. 1162474732 M * daniel_hozac (-rc4 coming soon...) 1162474876 M * marcfiu hmph.... my local mirror must not be up to date, as the latest FC5 kernel is 1.2200. 1162474885 M * daniel_hozac it is, AFAIK. 1162474889 M * daniel_hozac i'm building from CVS ;) 1162474946 M * marcfiu I'm running into merge problems under fs/nfs/super.c and fs/nfs/dir.c 1162474950 M * marcfiu Did you resolve those by hand. 1162474952 M * daniel_hozac yep. 1162475065 M * daniel_hozac http://cvs.hozac.com/viewvc/rpms/kernel/fedora-5/linux-2.6-vserver-2.0.patch?view=markup is my patch, if you're interested. 1162475068 M * daniel_hozac let me know if i missed anything. 1162475215 J * node_ ~dwindsor@c-69-143-154-220.hsd1.md.comcast.net 1162475238 M * marcfiu that's your fixed up patch to the 2209 kernel based on vs2.0.2.1? 1162475259 M * daniel_hozac that's the patch used to build the 2.6.18-1.2209_FC5.vs2.0.2.1 kernel in my repo, yes. 1162475266 M * daniel_hozac (but it's based on 2.0.2.2-rc3 1162475291 M * marcfiu why? 1162475298 M * daniel_hozac why what? 1162475314 M * marcfiu sorry... misread it. 1162475322 M * marcfiu 2.0.2.2.-rc3 is fine. 1162475336 M * daniel_hozac not really, i ought to update to -rc4. 1162475349 M * marcfiu I'll use your patch to see how to best resolve the conflicts I've run into now. 1162475366 M * daniel_hozac but i'm waiting for some stabilization in the FC kernels before i build new ones. 1162475495 M * marcfiu what's the delta from rc3 -> rc4? 1162475518 Q * comfrey Quit: leaving 1162475545 M * marcfiu who do I just grab the above patch in txt mode. 1162475549 M * marcfiu s/who/how 1162475646 M * marcfiu got it 1162475673 M * daniel_hozac mips ptrace (shouldn't be relevant for FC5 IIRC, with utrace and such), some proc fixes (that need further fixing), some compile fixes for ppc64/ia64, a scheduler fix (missing vx_activate_task, so accounting would be wrong), some signalling fixes, config.h removals, alpha/v850 IRQ fixes, and some sparse warnings fixed. 1162475681 M * daniel_hozac (-rc3 to -rc4) 1162475699 M * marcfiu I'll take the scheduler fix... will go to rc4 1162475748 M * daniel_hozac you probably shouldn't go with 2200 though. i've heard some pretty bad things about it. 1162475755 M * marcfiu oh 1162475756 M * marcfiu ok 1162475765 M * marcfiu I've had some odd problems with it, too. 1162475798 M * marcfiu where can I grab the 2209 version? 1162475831 M * daniel_hozac you should probably get the latest. 1162475841 J * tchan1 ~tchan@doppio.foo-projects.org 1162475845 M * daniel_hozac 2224 1162475861 M * daniel_hozac cvs -z3 -d:pserver:anonymous@cvs.fedoraproject.org:/cvs/dist co kernel 1162475887 Q * node_ Ping timeout: 480 seconds 1162475890 P * tchan1 1162475904 M * marcfiu that's the latest fc5 ? 1162475942 M * daniel_hozac that's what i got when i just cvs -z3 update -d'd, at least ;) 1162475953 M * marcfiu oh 1162475971 M * daniel_hozac if you want, i could commit what i have to my CVS. 1162476007 M * daniel_hozac (2224 with -rc4) 1162476113 M * daniel_hozac btw, any particular reason you don't want to use my RPM? 1162476120 Q * duckx Quit: Client exiting 1162476516 Q * bonbons Read error: Operation timed out 1162476593 M * nayco_work daniel_hozac: Hello, and thanks for your answer. The fact is that I got used to use Bertl_zZ 's SRPMS for Mandriva... And his last packages seem to be 0.30.210(5mdv) 1162476612 M * daniel_hozac nayco_work: what is it about the specfile included in the tarball that fails on mandriva? 1162476696 M * daniel_hozac i.e. why doesn't rpmbuild -tb util-vserver-0.30.211.tar.bz2 work? 1162476760 M * nayco_work daniel_hozac: Er... What do you mean ? The spec file in the tarball is known to fail ? Well, I don't know|remember, I haven't used the tarball for ages. It's got to many dependencies, where I can simply use the rpms made by Bertl_zZ ;-) Anyway, as his rpms fail to install on any machine I've got, I got used to use his SRPMS :D 1162476798 M * nayco_work daniel_hozac: I gonna try this. I hope there wont be too many dependancies... I do not like to load my host servers with devel packages ;-) 1162476802 M * daniel_hozac the srpm should have just as many dependencies as the tarball... you're doing the exact same thing. 1162476814 M * daniel_hozac and why are you building packages on your hosts? 1162476817 M * daniel_hozac use a guest for it? 1162476850 M * daniel_hozac (and then distribute the packages to all of your hosts) 1162476925 M * nayco_work Well, I thought about this, tried it, but... Well, as it it is better to build a package on the given machine, and as you need the utils in the first place to be able to build a guest... Well, I didn't thought long of it, and abandonned it ;-) I should reconsider it too ! 1162476934 P * adrien 1162476963 M * nayco_work Ok, gonna try the tarball. 1162476970 M * daniel_hozac why would it be better to build the package on the given machine? 1162476974 M * daniel_hozac that's Gentoo talk! 1162476977 M * daniel_hozac ;) 1162476985 M * sid3windr :) 1162477051 A * trippeh is currently fighting portage upgrading a ooold gentoo install to a newer profile 1162477164 M * trippeh I had to do a portage 'recovery', edited 2 ebuilds and 1 eclass to remove depends (some of them documented in official upgrade guide - DIRTY!), and manually fixed 2 header files behind the back of the packaging system to even get glibc and gcc in. 1162477246 J * bonbons ~bonbons@83.222.36.111 1162477268 M * daniel_hozac marcfiu: FYI, i just committed the latest. 1162477275 J * novice ~joe@host21.2037745.gcn.net.tw 1162477338 P * novice 1162477362 M * nayco_work daniel_hozac: Oh, yes, I now understand : I thought you were talking about another command... But rpmbuild too resolves dependancies, so that's great :-) ! 1162477548 J * adrien ~mirc@ppp08-89.dsl.citenet.net 1162477678 M * nayco_work Oh, anyway, all, I've got a problem with patch-2.6.18.1-vs2.1.1-rc45.diff : I get a kernel panic at boot, and a message like "could not open /dev/console". I'd like to paste the full boot message, but I dunno how 1162477710 M * daniel_hozac did you forget to make an initrd? 1162477716 M * adrien Hi - I'm still can't ssh to my first guest; I did the listen trick for the host; can someone help me ? 1162477717 M * daniel_hozac and -rc46 is the latest ;) 1162477729 M * adrien the error I have from var/log/secure in the guest is 1162477736 M * adrien localhost sshd[4316]: fatal: PAM: pam_open_session(): Cannot make/remove an entry for the specified session 1162477744 M * daniel_hozac adrien: nothing before that? 1162477751 M * adrien the guest accept my password 1162477753 M * adrien the die 1162477754 M * adrien yes 1162477763 M * adrien localhost sshd[4314]: Accepted password for root from 192.168.3. 82 port 59197 ssh2 1162477816 M * adrien I installed my guest using vserver name build -m yum... 1162477864 M * daniel_hozac which distribution? 1162477875 M * adrien centos 4.4 both host/guest 1162477902 M * daniel_hozac can you use su - if you do vserver ... enter? 1162477978 M * adrien yes that works 1162477991 M * nayco_work daniel_hozac: For the initrd, I gonna look, I dunno. (make install should do it...) For -rc46 : Well, I made the error to use the wiki homepage to get the last version, and it's not always up-to-date ;-) 1162478027 M * daniel_hozac indeed, it requires someone to update it ;) 1162478056 M * nayco_work daniel_hozac: Anyway, is there a means to copy the kernel boot message after a kernel panic ? Or the redirect them real-time to another machine ? 1162478085 M * daniel_hozac sure, serial/net console should let you do that. 1162478205 M * nayco_work net console ? Is there a kernel command line option to achieve this (append=, no ?) 1162478225 M * daniel_hozac if you have it compiled into the kernel, yes. 1162478230 M * nayco_work anyway, I just verified : I got an initrd. 1162478244 M * nayco_work Oh, and which option is it ? 1162478257 M * daniel_hozac see Documentation/networking/netconsole.txt 1162478261 M * nayco_work K 1162478318 M * daniel_hozac adrien: that's odd. what does your /etc/pam.d/sshd contain for the session? 1162478367 M * adrien #%PAM-1.0 1162478367 M * adrien auth required pam_stack.so service=system-auth 1162478367 M * adrien auth required pam_nologin.so 1162478367 M * adrien account required pam_stack.so service=system-auth 1162478367 M * adrien password required pam_stack.so service=system-auth 1162478368 M * adrien session required pam_stack.so service=system-auth 1162478368 M * adrien session required pam_loginuid.so 1162478379 M * daniel_hozac try getting rid of loginuid. 1162478407 M * adrien it works !!! 1162478417 M * adrien god I would never have found it 1162478470 M * adrien I'm going to google PAM because I have no idea what it is. 1162478484 M * adrien I think I should add this in the centos wiki 1162478510 M * daniel_hozac hmm? should go in the Linux-VServer wiki, IHMO. 1162478539 M * adrien yes in the centos page 1162478549 M * nayco_work daniel_hozac: this netconsole stuff is great. Ok, rebooting ;-) 1162478570 M * daniel_hozac nayco_work: you already had it compiled in? 1162478669 M * nayco_work daniel_hozac: well# grep -i CONSOLE .config 1162478669 M * nayco_work CONFIG_NETCONSOLE=m 1162478679 M * daniel_hozac =m won't work for early panics at all. 1162478684 M * nayco_work I hope it'll be enough 1162478687 M * daniel_hozac you'll want =y 1162478699 M * nayco_work oh, dear.... :-( 1162478716 M * nayco_work So the kernel won't load alone... 1162478739 M * nayco_work Ok, BRB, I've got a kernel panic message to learn by heart ;-) 1162478764 M * marcfiu daniel_hozac: had to leave my office for a moment. 1162478792 M * marcfiu I've not used FC cvs repository. How do I generate a 2.6.18 FC5 kernel from their patch sets? 1162478820 M * daniel_hozac after checking it out, make i386/x86_64/etc. should do it. 1162478826 M * daniel_hozac make prep should get you a tree. 1162478847 M * marcfiu nice 1162479158 M * daniel_hozac so, what's wrong with my kernels? :) 1162479369 M * daniel_hozac if you don't mind me asking. 1162479456 J * node_ ~dwindsor@stanford.columbia.tresys.com 1162479461 M * marcfiu nothing wrong with your kernels... we just have some additional changes on top of the FC5 + vs2.0.2.x stuff that you put together. 1162479485 M * marcfiu So we can't just pick up yours. 1162479504 M * marcfiu dumb question: after the "make prep" how do I know which FC5 kernel the thing put together? 1162479573 M * daniel_hozac well, the version you ran make prep in :) 1162479583 M * daniel_hozac i don't think it puts the version anywhere though. 1162479594 M * daniel_hozac in the built tree, i mean. 1162479605 M * marcfiu I ran "make prep" in the FC-5 directory. 1162479643 M * marcfiu I'd like to know whether this is 1.2209 or something else. 1162479661 M * daniel_hozac make verrel 1162479670 M * daniel_hozac (or just vi kernel-2.6.spec) 1162479672 M * marcfiu well, it is in the cvs tag. 1162479684 M * marcfiu kernel-2_6_18-1_2224_fc5 (revision: 1.1) 1162479701 M * daniel_hozac yep. 1162479722 M * marcfiu How do I know whether that's a good one or not? 1162479727 M * daniel_hozac you don't ;) 1162479734 M * marcfiu that's not good. 1162479766 M * daniel_hozac kernels get fixes all the time. you'll have to decide if you want them or not. 1162479795 M * marcfiu but at some point in time someone decides to release one of them, presumably after having done some sort of additional testing. 1162479826 M * marcfiu what's the relationship between the FC-5 and FC-6 version numbers? 1162479836 M * daniel_hozac none, AFAIK. 1162479841 M * marcfiu oh 1162479871 M * marcfiu Am I better off moving straight to FC6 or sticking to whatever is at the head of FC5? 1162479905 M * daniel_hozac there aren't too many differences between the trees. 1162479916 M * daniel_hozac fixes get applied to both of them, when relevant. 1162479928 M * marcfiu ok 1162480225 J * sebastian ~sebastian@office.star-hosting.de 1162480315 M * marcfiu However, I need to pick a stable kernel that we can roll out onto hundreds of machines and then use for, say, 6 months or so. 1162480334 M * marcfiu It would be nice to know whether something between 2200 -- 2224 was considered good. 1162480352 M * daniel_hozac wouldn't there have been another release then? ;) 1162480440 M * daniel_hozac but really, i don't think it works like that. it's a never-ending process, every kernel will have bugs. 1162480462 M * daniel_hozac the question is just if they affect you enough to warrant an upgrade. 1162480492 M * marcfiu fair enough. 1162480498 M * marcfiu well... not really. 1162480507 J * dlezcano ~dlezcano@cimai.net4.nerim.net 1162480516 M * dlezcano hi all ! 1162480517 M * marcfiu maybe I'll switch to Oracle's unbreakable Linux. 1162480521 M * marcfiu just kidding. 1162480533 M * daniel_hozac i still haven't read anything about that :( 1162480545 A * daniel_hozac is a bit behind on the news front. 1162480548 M * daniel_hozac hello dlezcano 1162480586 Q * dlezcano Remote host closed the connection 1162480619 J * dlezcano ~dlezcano@cimai.net4.nerim.net 1162480663 M * marcfiu Its Oracle flexing its enterprise muscle, selling support on RHEL at 1/2 the support price. Took about 25% off of RedHats stock price. I wouldn't be surprised if Oracle simply deflated RH's stock price and then just bought them. 1162480683 M * marcfiu anyway... we are way off vserver topic. 1162480689 M * daniel_hozac heh, indeed. 1162480726 M * daniel_hozac davej has been hinting towards a new kernel release RSN though. 1162480759 M * daniel_hozac which is supposed to fix most of the issues with the 2200 kernel. 1162481152 M * marcfiu its weird how the vserver fix to fs/nfs/super.c on the 2.6.18 FC5 kernel is split across super.c and client.c. 1162481159 Q * node_ Read error: Operation timed out 1162481184 M * daniel_hozac the FC kernels restructured NFS quite a bit. 1162481189 M * marcfiu yeah 1162481191 M * marcfiu why? 1162481219 M * waldi because they pulled changes which are pending for inclusion into mainline? 1162481236 M * daniel_hozac superblock sharing and such. 1162481332 J * node_ ~dwindsor@209.60.7.66 1162481355 M * marcfiu - if (IS_RDONLY(dir) && (nd->intent.open.flags & (O_CREAT|O_TRUNC|FMODE_WRITE))) |} 1162481355 M * marcfiu + if ((IS_RDONLY(dir) || MNT_IS_RDONLY(nd->mnt)) && | 1162481355 M * marcfiu + (nd->intent.open.flags & (O_CREAT|O_TRUNC|FMODE_WRITE))) 1162481364 M * marcfiu sorry about that. 1162481403 M * marcfiu looks like you have a fix in fs/nfs/dir.c that's not strictly vserver... that is, your fix already is in the 2224 kernel. 1162481424 M * marcfiu ignore me... 1162481434 M * marcfiu going to get some coffee. 1162481743 P * marcfiu 1162482004 M * derjohn daniel_hozac: connect from xxxx to unset(nfs): request from unprivileged port - might that be the cause of the problem? (I mounted now inside the guest) 1162482086 M * daniel_hozac derjohn: hmm, i guess so. 1162482103 M * daniel_hozac derjohn: i'm not terribly familiar with NFS though, so i can't say for sure. 1162482107 M * derjohn daniel_hozac, but I dont get how I can change that? 1162482114 M * derjohn ah, same problem .. 1162482156 M * daniel_hozac does httpd find the directory now? (when mounting inside the guest?) 1162482163 M * derjohn no 1162482205 M * daniel_hozac are you using a Linux box for the NFS? 1162482247 M * daniel_hozac (server) 1162482304 J * marcfiu ~mef@targe.CS.Princeton.EDU 1162482331 M * marcfiu looks like ia64/kernel/ptrace.c changed. 1162482341 M * marcfiu so that the vserver patch no longer applies. 1162482357 M * daniel_hozac my patch? 1162482376 M * marcfiu not your patch. 1162482381 M * marcfiu the rc4 version. 1162482382 M * daniel_hozac but yes, the FC kernels have utrace, so pretty much all of the ptrace hunks cause rejects. 1162482395 M * derjohn daniel_hozac, yes the filer is within a vsever guest, I used unfsd to serve 1162482410 M * daniel_hozac (and please let me know if i missed anything there. i'm still not sure i got that completely right) 1162482425 M * daniel_hozac derjohn: oh, unfsd... 1162482430 M * daniel_hozac unfs3 from Debian? 1162482440 M * derjohn debian yes 1162482445 M * derjohn vsersion: hold a sec 1162482447 M * daniel_hozac sarge? 1162482468 M * mnemoc sarg :p 1162482475 A * mnemoc hides 1162482489 J * weasel_ weasel@asteria.debian.or.at 1162482507 M * derjohn unfs3, Version: 0.9.13+dfsg-1 1162482509 N * weasel_ Guest2218 1162482544 Q * weasel Quit: Reconnecting 1162482545 N * Guest2218 weasel 1162482545 M * daniel_hozac looks like only 0.9.15 is still available. 1162482605 M * daniel_hozac derjohn: so it's running etch? 1162482621 M * derjohn yes, I just upgraded to 0.9.15+dfsg-1 1162482687 M * daniel_hozac ok, i'll try to reproduce it here. 1162482699 M * daniel_hozac do you know if it also happens with the kernel NFS server? 1162482738 M * derjohn daniel_hozac, no. I tried arounf with several nfs servers, but to run one in a guest, I *guessed* that I need a userspace daemon 1162482768 M * derjohn unfortunately te box is in production , means: with a devel rc35 the system runs 1162482774 M * derjohn I 1162482791 M * derjohn 'll be a little scared to replace the daemon in production 1162482801 M * daniel_hozac yeah, you'll need userspace daemons to run them in guests. 1162482911 M * derjohn but i doubt that it is a problem with the filer - as the 2nd boxen runs fine. I even switched to waldis debian kernel (2.6.17-2) on the host, same problem 1162482921 M * daniel_hozac dumb question: how do you tell dpkg to not execute scripts? 1162482936 M * derjohn not to ? 1162482943 M * derjohn does one fail? 1162482945 M * daniel_hozac yes. 1162482949 M * derjohn which? 1162482967 M * daniel_hozac pre-removal script. 1162482993 M * derjohn hm, you can put an exit 0 into it 1162483009 J * Piet hiddenserv@tor.noreply.org 1162483012 M * marcfiu daniel_hozac: so the only patch rejects that I cannot easily resolve are those related to s390 and ia64 ptrace.c 1162483025 M * derjohn or simply try a dpkg --forace-all -i /var/cache/apt/archives/...deb 1162483029 M * derjohn *force 1162483031 M * marcfiu has the logic in that file just moved somewhere else? 1162483037 M * daniel_hozac marcfiu: AFAICT, yes. 1162483043 M * daniel_hozac that's part of the point of utrace, i think. 1162483054 M * daniel_hozac rather than spreading out the logic into arch-specific files, put it in some generic ones. 1162483083 M * daniel_hozac derjohn: where are the scripts located? i'm trying to remove an installed package. 1162483088 Q * ntrs Ping timeout: 480 seconds 1162483088 M * daniel_hozac derjohn: --force-all doesn't help. 1162483131 M * derjohn /var/lib/dpkg/info/ AFAIR 1162483155 M * daniel_hozac thanks. 1162483168 M * derjohn you could also tell dpkg to forget about the packlage - but that leaves old fragments there 1162483189 M * daniel_hozac yeah, just adding exit 0 to the start of the scripts worked fine. 1162483437 M * marcfiu daniel_hozac: what do you suggest that I do for now? Ignore this? Your patch doesn't address this, right? 1162483469 M * daniel_hozac marcfiu: address what? i believe my patch has everything needed for ptrace, but let me know if it doesn't. 1162483567 M * marcfiu I am only eyeballing your patch. 1162483580 M * marcfiu I am trying to apply the rc4 vs2.0.2.2 patch. 1162483596 M * marcfiu it has two rejects: 1162483596 M * marcfiu % ptrace.c.rej<2> 430 Diff ~/projects/kernel-2.6.18-1.2224.fc5/arch/ia64/kernel/ptrace.c.rej 1162483596 M * marcfiu % ptrace.c.rej 412 Diff ~/projects/kernel-2.6.18-1.2224.fc5/arch/s390/kernel/ptrace.c.rej 1162483645 M * marcfiu Both are failing to add the following snippet: 1162483645 M * marcfiu *************** 1162483645 M * marcfiu *** 723,729 **** 1162483645 M * marcfiu goto out; 1162483645 M * marcfiu } 1162483646 M * marcfiu 1162483646 M * marcfiu ret = do_ptrace(child, request, addr, data); 1162483648 M * marcfiu put_task_struct(child); 1162483648 M * marcfiu out: 1162483650 M * marcfiu unlock_kernel(); 1162483650 M * marcfiu --- 723,735 ---- 1162483652 M * marcfiu goto out; 1162483652 M * marcfiu } 1162483654 M * marcfiu 1162483654 M * marcfiu + if (!vx_check(vx_task_xid(child), VX_WATCH|VX_IDENT)) { 1162483656 M * marcfiu + ret = -EPERM; 1162483656 M * marcfiu + goto out_tsk; 1162483658 M * marcfiu + } 1162483658 M * marcfiu + 1162483660 M * marcfiu ret = do_ptrace(child, request, addr, data); 1162483660 M * marcfiu + out_tsk: 1162483662 M * marcfiu put_task_struct(child); 1162483662 M * marcfiu out: 1162483664 M * marcfiu unlock_kernel(); 1162483664 M * marcfiu 1162483675 M * daniel_hozac wow 1162483680 M * daniel_hozac paste.linux-vserver.org, please. 1162483691 M * marcfiu sorry. 1162483702 Q * ruskie Remote host closed the connection 1162483704 M * marcfiu what does paste.linux-vserver.org mean. 1162483712 M * ms_ it's a url 1162483716 M * daniel_hozac it's a pastebin. 1162483716 M * marcfiu oh 1162483759 M * marcfiu ok... 1162483762 M * marcfiu just did that. 1162483767 M * marcfiu ahem... what does it do? 1162483775 M * marcfiu oh ok 1162483791 M * derjohn lart marcfiu !:) 1162483826 M * daniel_hozac but yes, i'm aware of the conflicts. 1162483835 M * daniel_hozac the code it's patching is removed by utrace though. 1162483844 M * marcfiu that's what I figured. 1162483852 M * marcfiu so can I just ignore it? 1162483856 J * ruskie ~ruskie@ruskie.user.oftc.net 1162483868 M * daniel_hozac that's what i did. 1162483872 M * marcfiu that is, is it handled in a non-arch specific way already? 1162483891 M * daniel_hozac AFAICT. 1162483905 M * daniel_hozac but again, let me know if i missed it. 1162483918 M * marcfiu I don't have s390 or ia64 to validate this myself. 1162483930 Q * node_ Read error: Connection reset by peer 1162483932 M * marcfiu So I am not sure how I'd let you know if you messed up. :) 1162483969 M * daniel_hozac well, if you find a place where we don't check the xid of the processes, for instance. 1162484111 M * daniel_hozac would be rather serious if you could strace a process from another guest or the host ;) 1162484163 M * marcfiu oh... well, I suspect you got it right for i386. 1162484180 M * marcfiu Its more a matter of completeness. 1162484194 M * marcfiu For the other architectures, that is. 1162484348 Q * bonbons Read error: Operation timed out 1162484366 Q * Piet Remote host closed the connection 1162484367 M * daniel_hozac derjohn: hmm, i must be stupid. i can't even get it to mount the NFS. 1162484386 M * daniel_hozac i just get "NFS version 3 is not supported." 1162484413 J * Piet hiddenserv@tor.noreply.org 1162484601 J * bonbons ~bonbons@83.222.36.111 1162484671 J * the_hydra ~a_mulyadi@202.59.168.29 1162484858 M * derjohn daniel_hozac, kill "stoo pid" :) 1162484863 M * Hollow meh... 1162484865 M * Hollow zeus ~ # vserver baselayout113 delete 1162484865 M * Hollow Are you sure you want to delete the vserver baselayout113 (y/N) y 1162484865 M * Hollow pkgcfgbase-dir '/var/vservers/.pkg' does not exist or is invalid 1162484883 M * daniel_hozac looks like incomplete packages :) 1162484893 M * derjohn daniel_hozac, I mounted now the same dir as samba mount from the host into the guest. Invredible: that works. 1162484901 M * derjohn *Incredible 1162484907 M * daniel_hozac hehe. 1162484909 M * Hollow i don't need pkg support from util-vserver ;) 1162484952 M * daniel_hozac well, the scripts tend to expect the directories they create to be there ;) 1162484998 M * derjohn daniel_hozac, maybe hosts allow or portmapper prob ? 1162485027 M * daniel_hozac hmm, running portmap -l in the host seems to have gotten it working from the same guest, at least. 1162485027 N * Bertl_zZ Bertl 1162485032 M * Bertl morning folks! 1162485041 M * daniel_hozac FC6 to etch though complains about differing versions. 1162485084 N * [PUPPETS]Gonzo _[PUPPETS]Gonzo 1162485086 M * the_hydra morning Bertl 1162485099 M * Hollow daniel_hozac: well, teh template build method does not create these..? 1162485118 M * dlezcano hi Bertl 1162485127 M * marcfiu hello there bertl 1162485134 M * the_hydra Bertl: good sleep? 1162485179 M * Bertl yeah, was fine .. no complaints ... 1162485206 M * derjohn daniel_hozac, i p-msg you the conf files. 1162485228 M * Bertl marcfiu: to what kernel are you adapting the patches? 1162485236 M * derjohn daniel_hozac, now i put the samba mount line into /etc/vserver//fstab ... that also mounts and works fine. 1162485281 M * Hollow daniel_hozac: and neither does make install 1162485289 M * nayco_work Hello, Bertl ! 1162485363 M * daniel_hozac Hollow: the distribution-install should create it, IIRC. 1162485385 M * daniel_hozac derjohn: yeah, samba isn't quite as problematic as NFS. 1162485390 M * nayco_work Sniffle : I just rm -f'ed my first VServer kernel, vmlinuz-2.6.9-vs1.9.3 (replaced by -2.6.18.1-vs2.1.1-rc45). 1162485408 M * marcfiu bertl: 2.6.18 FC5 1.2224 1162485429 M * Hollow daniel_hozac: ah, we don't run that... 1162485449 M * Bertl marcfiu: ah, ic, and they modified some things? 1162485451 M * daniel_hozac Hollow: there you go then ;) 1162485470 M * nayco_work daniel_hozac: Ok, I managed to make -2.6.18.1-vs2.1.1-rc45 start : Was an issue with initrd not properly built, by lack of space on disk :-O. Why the hell the "make install" succeedeed, so ??? Well.. 1162485485 M * derjohn daniel_hozac : I dont know if i dislike nfs more or windows oriented fs-es ... 1162485488 M * daniel_hozac nayco_work: hehe, that does sound strange. 1162485518 M * derjohn daniel_hozac, i asked myself is fuse might be an alternative ... but i think that wont be high performqnce 1162485524 M * daniel_hozac derjohn: yeah... plague or cholera. 1162485539 M * derjohn daniel_hozac, *g* 1162485574 M * nayco_work Well, I had an error message about "device full" during "make install", but in the middle of all the output... I did not see it. Anyway, I think the command should have failed... Well, now the kernel boots, so... 1162485618 M * derjohn daniel_hozac, I think i'll use samba as solution now .. ( as it works and my customer gets slowly p...ed ) 1162485638 M * daniel_hozac yeah, samba is sort-of-okish. 1162485639 M * nayco_work Bertl: Have you seen my message here this morning ? About util-vserver SRPMS (-5mdv) not building ? 1162485674 M * daniel_hozac nayco_work: btw, that problem is fixed in 0.30.211 ;) 1162485696 J * Rich_Estill ~restill@c-24-11-195-139.hsd1.mi.comcast.net 1162485705 M * marcfiu bertl: they've modified the way the fs/nfs code is structured and apparently the ia64 and s390 kernel/ptrace.c has changed such that the vs2.0.2.2-rc4 patch results in rejects. 1162485706 M * nayco_work daniel_hozac: The "-lutil" thing ? 1162485736 M * Bertl nayco_work: nope, but I should updated those to 0.30.211 1162485752 M * daniel_hozac nayco_work: yes. 1162485754 M * marcfiu bertl: daniel_hozac figured out the fixes to fs/nfs and also mentioned that for now he thinks it is ok to ignore the ia64/s390 ptrace changes, as they're moving to utrace. 1162485784 M * Bertl yeah, I wondered if we actually need the ptrace checks there 1162485814 M * derjohn daniel_hozac, okish ? orkish ;) 1162485848 A * the_hydra found interesting word above....utrace? 1162485876 M * nayco_work Bertl: Ok. Oh, by the way, do patch-....-rc45 includes modifications related to quotas ? Or maybe -rc46 ? 1162485881 M * the_hydra a new kind of ptrace? 1162485882 M * derjohn what about vtrace ? no one made one yet ? 1162485900 M * derjohn nayco_work, dont use rc45 ! 1162485954 M * Bertl nayco_work: nope, you mean the xfs quota stuff, yes? 1162485958 M * Borg- Bertl: exacly same problem.. as yesterday... I can kill or vkill process.. or even killall5 from shell.. and vserver stops 1162485975 M * nayco_work Bertl: yep ;-) 1162485989 M * nayco_work derjohn: oops :-( 1162485991 M * Borg- I dunno really what can be wrong.. very weird 1162485999 M * nayco_work derjohn: Broken ? 1162486009 M * Bertl Borg-: what Linux-VServer patch do you use atm? 1162486035 M * Borg- vs2.0.2.1 1162486064 M * nayco_work Oh, BRB. 1162486064 M * Borg- the only difference today is that. it dont show processes :) 1162486068 M * Borg- just empty line.. 1162486106 M * Borg- it just complains that process didnt stoped.. but dont say with one.. I added ps -axc to rc script to see what processes are running 1162486183 M * Bertl Borg-: could you, just for the fun of doing it, try with 2.1.1-rc46? 1162486238 Q * ruskie Remote host closed the connection 1162486252 M * derjohn nayco_work, with rc45 pids in guest might vanish without cause, IIUC 1162486264 M * Borg- Bertl: sure.. when I will get some free time.. I will setup additional box.. exacly same instalation.. to check things out 1162486279 M * Borg- if problem will be reproductable.. I will upgrade to anything latest :) 1162486293 M * Borg- I dont want to touch this box :) its production server.. 1162486304 Q * sebastian 1162486305 M * Borg- and stoping vservers is not so big issue anyway.. 1162486320 M * Bertl Borg-: okay, please keep us updated on this ... 1162486345 M * Borg- oki :) 1162486416 J * ruskie ~ruskie@ruskie.user.oftc.net 1162486540 M * Bertl welcome ruskie! 1162486589 M * Bertl marcfiu: you should probably check out the devel kernels instead (IIRC that's what you planned :) 1162486638 J * Mr_rayman_2006 ~Mr_rayman@212.183.136.194 1162486844 P * Mr_rayman_2006 1162486903 M * daniel_hozac Bertl: how do they help? i got at least twice as many rejects for the devel patch ;) 1162486913 M * daniel_hozac (which is why i dropped it for now, i don't think anyone is using it anyway) 1162487004 J * s0undt3ch_ ~s0undt3ch@81.193.61.37 1162487012 M * Bertl daniel_hozac: well, the features marcfiu is looking for _are_ in devel :) 1162487026 M * daniel_hozac oh, what features are those? 1162487036 M * daniel_hozac i must've missed something. 1162487080 M * daniel_hozac Bertl: and why isn't 2.1.1 released yet? :) 1162487096 M * daniel_hozac just finishing the changelog? 1162487115 M * Bertl changelog is almost done, patch breakdown has to be done 1162487130 M * Bertl daniel_hozac: the idle time scheduler, for example 1162487150 M * daniel_hozac ah, ok. 1162487171 M * daniel_hozac how about a 2.0.2.2 release then? ;) 1162487391 J * ruskie_ ~ruskie@84.20.228.4 1162487445 Q * s0undt3ch Ping timeout: 480 seconds 1162487470 Q * ruskie Ping timeout: 480 seconds 1162487486 M * daniel_hozac i'm just kidding, but other than the put_task_struct in find_proc_task_by_pid, what's 2.0.2.2 waiting for? 1162487527 M * Bertl have to check the changes we did in 2.1.1 so far 1162487541 N * ruskie_ ruskie 1162487590 Q * ruskie 1162487604 J * ruskie ~ruskie@ruskie.user.oftc.net 1162487606 J * |yang| ~yang@cpe-213-157-253-172.dynamic.amis.net 1162487642 M * Bertl okay, off for dinner, back shortly 1162487647 N * Bertl Bertl_oO 1162487656 M * |yang| I keep getting strange errors on apt-get upgrade and on /etc/init.d/package restart inside a guest, also some segmentation faults :( ... http://paste.debian.net/15997 1162487677 Q * s0undt3ch_ Ping timeout: 480 seconds 1162487718 M * marcfiu bertl: dinner? I thought you just woke up. :) 1162487737 M * daniel_hozac |yang|: looks like problems with your software. file bugs. 1162487790 Q * ruskie Read error: Connection reset by peer 1162487854 M * |yang| daniel_hozac: damnit, it was a clean install 1162487982 J * ruskie ~ruskie@ruskie.user.oftc.net 1162488034 J * s0undt3ch ~s0undt3ch@bl4-61-37.dsl.telepac.pt 1162488402 Q * |yang| Ping timeout: 480 seconds 1162488402 Q * ruskie Read error: Connection reset by peer 1162488471 J * ruskie ~ruskie@ruskie.user.oftc.net 1162488627 J * ntrs ~ntrs@207-119-222-143.dyn.centurytel.net 1162488813 J * ntrs_ ~ntrs@d16-241.rt-bras.wnvl.centurytel.net 1162488946 Q * bronson Read error: Operation timed out 1162489093 N * Bertl_oO Bertl 1162489123 M * Bertl marcfiu: I alway have dinner shortly after I wake up ... and breakfast in the morning right before I go to bed :) 1162489149 M * Hollow hehe. 1162489173 M * Hollow don't you know that you should not eat right before going to bed? ;) 1162489183 Q * ruskie Read error: Connection reset by peer 1162489220 Q * ntrs Ping timeout: 480 seconds 1162489223 M * Bertl Hollow: my brain is doing quite some work when I sleep, so it needs some kind of nutrition ... 1162489243 M * Hollow dreaming of kernel fixes, eh? ;) 1162489295 Q * ntrs_ Ping timeout: 480 seconds 1162489377 J * ruskie ~ruskie@ruskie.user.oftc.net 1162489400 Q * the_hydra 1162489411 M * Bertl Hollow: sometimes :) 1162490338 Q * ruskie Read error: Connection reset by peer 1162490436 J * ruskie ~ruskie@ruskie.user.oftc.net 1162490493 J * duckx ~Duck@tox.dyndns.org 1162491466 Q * ruskie Remote host closed the connection 1162491598 J * ruskie ~ruskie@ruskie.user.oftc.net 1162491777 M * Hollow daniel_hozac: why does vsomething execute the command on the host if VSOMETHING_PKGMGMT=0? 1162491793 M * daniel_hozac hmm? 1162491816 M * Hollow vsomething, line 59 1162491840 M * daniel_hozac while true; do? 1162491840 M * Hollow ok.. not =0, but empty 1162491875 M * Hollow no..: else "$cmd" "$i" "$@" 1162491887 M * daniel_hozac that's line 119. 1162491897 M * Hollow ah yeah.. sorry 1162491910 M * daniel_hozac i guess it's mostly meant for scripts. 1162491949 M * daniel_hozac vsomething vserver ... -- exec ... works fine though ;) 1162491962 M * Hollow what improvement :P 1162491987 M * daniel_hozac over? 1162491992 M * Hollow well, i see if i can add all those pkgmgmt options for gentoo 1162492020 M * daniel_hozac you just need VSOMETHING_TITLE and VSOMETHING_PKGMGMT, no? 1162492032 M * daniel_hozac no VSOMETHING_WORKER since external package management isn't supported. 1162492063 M * Hollow i see 1162492330 J * ruskie_ ~ruskie@84.20.228.4 1162492340 Q * derjohn2 Ping timeout: 480 seconds 1162492342 M * daniel_hozac or are you also working on support for unification? 1162492378 J * derjohn2 ~aj@dslb-084-058-232-082.pools.arcor-ip.net 1162492441 M * Hollow daniel_hozac: at least i will take a look at it ;) 1162492449 M * daniel_hozac ok, cool! 1162492472 Q * ruskie Ping timeout: 480 seconds 1162492481 N * ruskie_ ruskie 1162492487 M * Hollow daniel_hozac: can you give me a hint where do i need to add things for external pkg management? 1162492577 M * daniel_hozac scripts/vpkg, scripts/pkgmgmt, at least. 1162492589 M * Hollow ok.. 1162492605 M * daniel_hozac and then you would need a VSOMETHING_WORKER ;) 1162492624 M * daniel_hozac how do you install a Gentoo-guest without emerge and such though? 1162492629 M * Hollow yeah, figured that already :) 1162492630 J * bronson ~bronson@66.160.177.201 1162492656 M * Hollow daniel_hozac: well, for yum/apt/etc you need the approriate binaries on the host, no? 1162492663 M * daniel_hozac yes. 1162492677 M * Hollow so, if you have emerge on the host, you can use ROOT=$vdir emerge 1162492695 M * Hollow implies gcc & friends of course 1162492723 M * Hollow but i doubt any distribution has portage packages :o 1162492829 M * daniel_hozac well, yum/apt/etc. weren't very common before either ;) 1162492840 M * daniel_hozac if people want it, they'll have to install it. 1162492900 M * daniel_hozac so it is possible to have a Gentoo-guest without compilers, emerge, etc.? 1162492900 Q * ruskie Read error: Connection reset by peer 1162492919 M * Hollow daniel_hozac: yep 1162492929 M * Hollow in theory 1162492931 M * Hollow :) 1162492937 M * daniel_hozac ok, cool. 1162492998 M * derjohn daniel_hozac: again, smb works nice and is even pw protected. my host is up again, but I can test with NFS again, if there are any substential new ideas :) 1162493010 J * ruskie ~ruskie@ruskie.user.oftc.net 1162493020 M * daniel_hozac derjohn: not yet... i'll try to reproduce the issue locally. 1162493020 M * Hollow and i probably need a vserver-build.functions.portage then... 1162493026 M * Hollow or *.emerge 1162493032 M * daniel_hozac indeed. 1162493037 M * phreak`` Hollow: -K (well you will need emerge on the host *g*) 1162493039 M * daniel_hozac well, if there are any functions. 1162493064 M * Hollow phreak``: -K? 1162493066 M * phreak`` s/-K/ROOT=/vservers/$VPS emerge -K / 1162493074 A * phreak`` slaps head 1162493092 M * Hollow uhm? 1162493099 M * Hollow i know.. usebinpkgonly 1162493108 M * phreak`` Hollow: yup :) 1162493111 M * Hollow but you can also compile it inot $ROOT, no? 1162493115 Q * Piet Remote host closed the connection 1162493116 M * Hollow *into 1162493125 M * phreak`` Hollow: sure you can .. 1162493143 M * phreak`` *err, only merge into $ROOT 1162493146 M * Hollow i don't trust binpkgs in portage ;)) 1162493151 M * phreak`` (merge/qmerge) 1162493181 M * phreak`` Hollow: yeah, they ain't trustworthy yet 1162493204 J * Piet hiddenserv@tor.noreply.org 1162493834 J * ruskie_ ~ruskie@84.20.228.4 1162493890 Q * shedi Remote host closed the connection 1162493944 Q * dlezcano Quit: Leaving 1162493992 Q * ruskie Ping timeout: 480 seconds 1162494046 N * ruskie_ ruskie 1162494103 M * Hollow daniel_hozac: btw, what about the vschedcalc script we have in gentoo? do you want to have that in util-vserver? 1162494119 M * Bertl what does it do? 1162494139 M * Bertl ah, speaking of the scheduler .... 1162494149 M * Hollow yep.. 1162494162 M * Hollow http://dev.croup.de/repos/gentoo-vps/util-vserver/tools/vschedcalc 1162494169 M * Bertl daniel_hozac: does 211+ support the new scheduler parameters? 1162494172 M * daniel_hozac wouldn't it be better to have some sort of page explaining the values with input boxes or so? 1162494188 M * daniel_hozac Bertl: no, that's one of the things i have planned for 212. 1162494196 M * daniel_hozac (and splitting out the scheduler configuration into multiple files) 1162494203 M * Bertl excellent! please do that 1162494209 J * rgl ~Rui@84.90.9.66 1162494215 M * rgl hello 1162494218 M * daniel_hozac hi 1162494220 M * Bertl welcome rgl! 1162494249 M * Bertl daniel_hozac: from the tool PoV be careful, you need two syscalls to configure the new scheduler (and even more for per cpu configs) 1162494329 M * daniel_hozac are there any further changes to be expected to the available parameters? 1162494354 M * daniel_hozac or is it "done"? :) 1162494356 M * Bertl no, not for the scheduler 1162494369 M * Bertl we might use token buckets for other things in the future 1162494377 M * daniel_hozac yeah, i think it's pretty much as capable as it can be right now. 1162494378 M * Bertl but I think the scheduler is quite finished 1162494527 Q * ruskie Ping timeout: 480 seconds 1162494667 J * ruskie ~ruskie@ruskie.user.oftc.net 1162494849 M * Bertl http://plm.testing.osdl.org/patches/show/Linux-VServer-2.6.18.1-vs2.1.1-rc46 (still running) 1162494876 M * daniel_hozac oooh, you can do them by name now? nice. 1162494896 M * Bertl yeah, didn't know that either, just found it out ... 1162494922 M * Hollow daniel_hozac: is it possible to do some sanity check for --all to just pick gentoo guests? based on /etc/gentoo-release inside? 1162494985 M * daniel_hozac humm, no... 1162495034 M * Hollow maybe implement sth like VSOMETHING_SANITYCHECK? 1162495052 M * daniel_hozac that would make sense. 1162495062 M * Hollow ok, will try that ;) 1162495124 M * Hollow though.. vsomething is called with exec.. that will render all previous functions useless.. 1162495139 M * daniel_hozac yeah, so it'd have to be a separate script. 1162495147 M * daniel_hozac and that just seems like overhead... 1162495150 M * Hollow indeed 1162495181 M * daniel_hozac but maybe it's ok if things fail. 1162495187 M * Bertl btw, it seems we are inconsistant in the ptrace error handling 1162495203 M * daniel_hozac i mean, if you tell it to run in emerge in all of your guests, and you have 10 Debian guests, you have to expect 10 errors. 1162495205 M * Bertl some archs/cases return EPERM, others ESRCH 1162495212 M * daniel_hozac Bertl: yeah, i just noticed that. 1162495231 M * Bertl I also wonder if we aren't doing some checks twice 1162495242 M * daniel_hozac i imagine we're doing them at least twice. 1162495249 M * Bertl but I guess we'll fix that up after 2.1.1 1162495255 M * Bertl I WANT IT OUT :) 1162495257 M * daniel_hozac all of the ptrace functions use find_task_by_pid or so, no? 1162495261 M * daniel_hozac hehe, we all do! 1162495273 M * Hollow daniel_hozac: what about adding these functions to $_LIB_FUNCTIONS? vsomething sources it.. just pass the function name in VSOMETHING_SANITYFUNC ..? 1162495289 M * daniel_hozac Hollow: but that seems wrong too. 1162495308 A * phreak`` goes building 0.30.211-r1 1162495313 M * Hollow yeah, quite hacky 1162495321 M * Hollow phreak``: please add make install-distribution 1162495326 M * Hollow and remove the fperms 000 1162495473 M * Hollow daniel_hozac: well, at least vsomething does not set -e, so it should be ok.. 1162495535 M * daniel_hozac yeah, it's ok. 1162495542 M * daniel_hozac it prints a nice warning though. 1162495608 J * shedi ~siggi@inferno.lhi.is 1162495616 M * daniel_hozac pizza just got here... i'll bbiab. 1162495910 M * phreak`` Hollow: daniel_hozac: is /etc/vservers.conf even needed nowadays (with =0.30.211 that is) ? 1162495960 A * Hollow smells legacy 1162496047 A * phreak`` just pruned legacy 1162496059 M * Hollow USE="legacy"? 1162496127 M * phreak`` nope, no legacy at all ;) 1162496128 M * phreak`` # CONFIG_VSERVER_LEGACY is not set 1162496128 M * phreak`` # CONFIG_VSERVER_LEGACYNET is not set 1162496140 M * Hollow ah.. but please leave USE="legacy" in place... 1162496150 M * phreak`` Hollow: sure, within the ebuild :) 1162496161 M * Hollow phreak``, daniel_hozac: http://people.linux-vserver.org/~hollow/util-vserver/delta-gentoo-vsomething-feat01.diff 1162496184 M * Hollow not sure how to handle vesync... 1162496190 M * Hollow regarding the overlay thing 1162496441 J * |yang| ~yang@cpe-213-157-253-172.dynamic.amis.net 1162496642 Q * hardwire Quit: Coyote finally caught me 1162496669 J * hardwire ~hardwire@89-208-58-66.gci.net 1162497142 M * daniel_hozac Hollow: is it necessary to hardcode the path to emerge? 1162497242 M * daniel_hozac and what's yum_opts doing there? ;) 1162497251 M * daniel_hozac (that looks like something i missed though... 1162497295 M * daniel_hozac hmm, no, looks fine to me. 1162498052 Q * bronson Ping timeout: 480 seconds 1162498157 A * phreak`` pokes daniel_hozac with /etc/vservers.conf 1162498257 M * daniel_hozac it is legacy. 1162498272 M * daniel_hozac contrib/manifest.dat.pathsubst can tell you about that ;) 1162498292 J * DreamerC_ ~dreamerc@61-217-226-119.dynamic.hinet.net 1162498438 M * phreak`` daniel_hozac: legacy == older tools or legacy support within current tools ? 1162498444 M * daniel_hozac both. 1162498453 M * daniel_hozac legacy support within current tools == older tools.- 1162498456 M * daniel_hozac basically. 1162498465 M * daniel_hozac well, legacy config. 1162498474 M * daniel_hozac legacy kernels with new config don't use it. 1162498485 J * bronson ~bronson@66.160.177.201 1162498516 M * Hollow daniel_hozac: ah yep.. one ref to yum_opts remained.. ;) 1162498535 M * Hollow and well... does not need to have a hardcoded path, bt doesnÄt hurt either, does it? 1162498548 M * daniel_hozac what if it's installed in /usr/local? 1162498577 M * Hollow beside that it will probably never happen, i can still change it ;) 1162498583 J * stefani ~stefani@tsipoor.banerian.org 1162498665 Q * DreamerC Ping timeout: 480 seconds 1162498812 M * Bertl wb stefani! 1162498819 M * daniel_hozac so about the scheduler configuration, what do you think about a sched directory, with optional subdirectories for CPU-specific settings, and each of those potentially containing a directory called idle, each directory containing fill_rate, interval, tokens_min, tokens_max, [prio_bias]? 1162498868 M * Bertl hmm, I'd prefer to put the idle stuff in fill_rate2 or fill_rate_idle 1162498882 M * Bertl isntead of having a subdirectory with two entries 1162498898 M * daniel_hozac ah, so it is just fill_rate and interval for idle time? ok. 1162498901 M * daniel_hozac yeah, that makes more sense. 1162498924 M * Bertl yes 1162498939 M * daniel_hozac for some reason i had gotten the impression the min/max changed as well. 1162498966 Q * bronson Ping timeout: 480 seconds 1162498980 M * Bertl nah, only the rate/interval is used, but we need a flag for idle time skipping 1162499000 M * Bertl but that can be implicit when one of the idle_* values is given 1162499081 M * stefani Bertl: IBM has something they call a 'director' which apparently manages hypervisors of different virtualizing methods, such as VM, xen, but they knew nothing about vservers, so i told them. 1162499108 M * Bertl good, but Linux-VServer is not hypervisor based ... 1162499122 M * stefani i know. 1162499145 M * stefani they had not even heard of this. i gave them a copy of an article to read. 1162499169 M * Bertl excellent .. well some IBM folks definitely heard about Linux-VServer before :) 1162499195 M * stefani the article was the "container-based operating system.." 1162499990 J * bronson ~bronson@adsl-64-161-106-11.dsl.snfc21.pacbell.net 1162501133 J * Aiken ~james@tooax6-009.dialup.optusnet.com.au 1162501718 M * Bertl wb bronson! morning Aiken! 1162501742 M * bronson Howdy Bertl. 1162501910 Q * stefani Quit: I Quit! 1162501991 M * Aiken good morning 1162502087 J * kir_home ~kir@213.152.157.70 1162502116 M * Bertl wb kir_home! 1162503121 J * s0undt3ch_ ~s0undt3ch@bl9-224-199.dsl.telepac.pt 1162503499 Q * s0undt3ch Read error: Operation timed out 1162503499 N * s0undt3ch_ s0undt3ch 1162506767 Q * dna_ Quit: Verlassend 1162507611 Q * kir_home Ping timeout: 480 seconds 1162507731 Q * rgl Quit: Fui embora 1162508599 J * Aiken_ ~james@tooax6-009.dialup.optusnet.com.au 1162508731 Q * Aiken Ping timeout: 480 seconds 1162508879 J * muallin ~me@79.Red-217-127-197.staticIP.rima-tde.net 1162508892 M * muallin hi all 1162508902 Q * muallin 1162508912 M * Bertl that was quick ... 1162510104 M * Bertl *phew* finally got the breakdown done 1162510204 M * Aiken_ Bertl the dietlibc src rpm you pointed me to yesterday, was the tar ball just a std dietlibc? I never got around to applying any of the patches in the rpm 1162510221 M * Bertl yes, the tar ball is the source from fefe 1162510235 M * Aiken_ ok, that was enough 1162510236 M * Bertl i.e. all the patches are ontop of upstream 1162510578 Q * FireEgl Quit: Bye... 1162510640 M * doener http://www.novell.com/linux/microsoft/openletter.html -- even the url alone scares me ;) 1162510656 M * Bertl hehe 1162510878 M * Bertl sounds like M$ is facing serious issues :) 1162510921 M * doener or novell... from what I heard, the latest SuSE Versions weren't exactly great 1162510949 M * Bertl well, seriously, who is using suse nowadays? 1162510987 M * Bertl (except for the novell folks, which have to) 1162511046 M * Bertl note: I'm not saying that SuSE is bad or so, just it isn't used very much 1162511053 M * Bertl s/much/often/ 1162511132 M * doener our robotic labs linux boxes at university are almost all SuSE (some are RedHat but most are SuSE IIRC). (OpenSUSE, not sure if it's RHEL or pre-Fedora RedHat community thingy) 1162511262 M * doener But I never understood why someone would want to use it either, at least after the ISDN support in other distros improved and DSL came into play 1162511290 M * doener I've always known SuSE as point-and-click Linux with great ISDN support (which was pretty important in Germany back then) 1162511342 M * doener My actual personal usage was limited to about 30 minutes after the installation completed though *g* 1162511362 M * Bertl lol 1162511581 M * daniel_hozac Germany is probably the only place SuSE is used at all;) 1162511611 M * Bertl nah, recently some US folks tolkd me about the 'susi' distro :) 1162511655 M * daniel_hozac hehe, really? 1162511661 J * ntrs ~ntrs@68-188-55-120.dhcp.stls.mo.charter.com 1162511703 M * Bertl wb ntrs!