1170547204 M * Aiken version: 0002:0102 1170547209 M * Aiken vserver: ret = 0x00020102 (131330) 1170547209 M * Bertl hehe 1170547345 Q * comfrey_1 Server closed connection 1170547356 J * comfrey_ ~comfrey@70.91.185.84 1170547398 M * jjpacheco hola everyone 1170547404 M * Bertl hey jjpacheco! 1170547428 M * Bertl Aiken: now this one should produce better code, but the same results: http://vserver.13thfloor.at/Experimental/SYSCALL/syscall_shiny14d.h 1170547501 J * shuri ~shuri@hq01.electronicbox.net 1170547638 M * Aiken not good 1170547642 M * Aiken (root@h31-106) ./vcmd -V 1170547643 M * Aiken vc_get_version: Function not implemented 1170547643 M * Aiken vserver: Function not implemented 1170547790 M * Bertl ah, I forgot the clobber list 1170547806 M * Bertl try adding the following line after #define__sysc_regs 1170547814 M * Bertl (again in the ALPHA section) 1170547843 M * Bertl #define __sysc_clbrs __sysc_regs 1170547872 M * Bertl nah, won't work 1170547904 M * Bertl let's make that 1170547908 M * Bertl #define __sysc_clbrs "$16", "$17", "$18", "memory", "$20", "$21" 1170548000 M * Bertl if that doesn't work out either, we switch back the old version :) 1170548082 M * Aiken vc_get_version: Function not implemented 1170548082 M * Aiken vserver: Function not implemented 1170548105 M * Bertl strange, but so be it ... 1170548115 M * Bertl no optimized code for alpha then :) 1170548129 M * Aiken in 14d should the 2 mov be commented? 1170548135 M * Bertl yep 1170548146 M * Bertl the compiler should take care of proper assignment 1170548155 M * Bertl but obviously it fails on doing that 1170548201 M * Aiken currently 3.3.6 1170548224 M * Bertl yeah, I think gcc is not up to inline asm after all 1170548243 M * Bertl too many 'internal' assumptions and issues with register clobbering 1170548366 M * Aiken when I try getting a set of working tools again on that machine which syscall header should I use? 1170548371 M * Bertl so let's replace the alpha section with: http://paste.linux-vserver.org/1034 1170548374 M * Aiken at the moment vcmd is working with the shiny 14c 1170548402 M * Bertl try the 14d with this change (i.e. it should revert to 14c behaviour) 1170548414 M * Bertl without affecting other archs 1170548428 M * Bertl if not, we are dealing with a different issue 1170548593 M * Aiken vc_get_version: Function not implemented 1170548594 M * Aiken vserver: Function not implemented 1170548603 M * Bertl that is bad 1170548639 M * Aiken that is the entire alpha section in 14d http://paste.linux-vserver.org/1035 1170548670 M * Bertl looks good to me 1170548688 M * Bertl so something else must have changed (which actually breaks it) 1170548698 M * Bertl the 14c with the register switch works, yes? 1170548770 M * Aiken thought so, compiling again 1170548797 M * Aiken yes -> version: 0002:0102 1170548819 M * Bertl okay, let's compare the rsulting asm code then 1170548823 M * Bertl sec 1170548857 M * Bertl you want to do 1170548877 M * Bertl gcc -g -Wall -O2 -Ivserver -S -o vmcd.s vcmd.c 1170548885 M * Bertl gcc -g -Wall -O2 -Ivserver -E -o vmcd.E vcmd.c 1170548915 M * Bertl for both versions, i.e. 14c+switch and 14d+change 1170549077 Q * shuri Remote host closed the connection 1170549357 M * Aiken uploading 1170549371 M * Bertl ok, great, tx! 1170549453 M * jjpacheco question 1170549487 M * jjpacheco if we used secure-mount to mount bind (using --chroot and --mtab) how do we unmount? 1170549514 M * jjpacheco vnamespace -e XID umount $MOUNT_POINT ? 1170549518 M * Aiken http://www.users.on.net/~boddingt/Bertl/ 1170549563 M * Bertl jjpacheco: should work 1170549566 M * Bertl Aiken: tx 1170549592 M * jjpacheco i keep getting a message saying that $MOUNT_POINT is not mounted 1170549655 M * Bertl well, probably mount is again too smart ... 1170549680 M * Bertl but you might as well make a secure-umount which does what you want from outside 1170549699 M * jjpacheco nm, i got it 1170549701 M * jjpacheco thx bertl 1170549772 M * Bertl Aiken: oops, seems I forgot something important ... checking ... 1170549781 M * jjpacheco ther is a secure-umount ? 1170549790 M * jjpacheco i only seem to have secure-mount 1170550601 M * Bertl no, but you might write one :) 1170550621 M * jjpacheco ohh, gotcha 1170550660 M * Bertl Aiken: okay, let's give that a try ... http://vserver.13thfloor.at/Experimental/SYSCALL/syscall_shiny14e.h 1170550676 M * Bertl (we lost the syscall number in the previous versions :/) 1170550741 M * Bertl which pretty much explains the unexpected results ... 1170551063 M * Aiken 14e -> version: 0002:0102 1170551069 M * Bertl excellent! 1170551081 M * Bertl let's give that a try with util-vserver then please :) 1170551321 Q * rpc Ping timeout: 480 seconds 1170551506 M * Aiken shiny gets copied over syscall-alternative.h ? 1170551516 M * Bertl yep, that should work 1170552028 Q * Johnnie Quit: G'bye! 1170552355 M * Aiken hoppy lives 1170552367 M * Aiken I can start/stop a guest with 0.30.213-rc1 1170552386 J * Johnnie ~jdlewis@jdlewis.org 1170552451 M * Bertl excellent! 1170552472 M * Bertl so I guess alpha is working at least for 3.3.6 1170552491 M * Bertl btw, you did force the alternative syscall on config, yes? 1170552514 M * Bertl (or at least it got enabled somehow) 1170552515 M * Aiken syscall(2) invocation: alternative 1170552515 M * Aiken vserver(2) syscall#: 428/glibc 1170552524 M * Bertl great! 1170552534 M * Aiken I should go 1170552542 M * Aiken maybe be back later today 1170552549 M * Bertl okay, np, thanks a lot! 1170552560 Q * Aiken Quit: Leaving 1170553619 M * vasko hi, any progress on this issue? http://www.mail-archive.com/vserver%40list.linux-vserver.org/msg10058.html 1170553632 M * Bertl sec 1170553673 M * Bertl yes, that is already part of newer util-vserver 1170553689 M * Bertl i.e. 0.30.213-rc1 should provide the sysctl limits 1170553699 M * vasko great :) 1170553788 M * jjpacheco i got a question 1170553794 M * jjpacheco i use this command 1170553806 M * jjpacheco vserver vs_name enter 1170553810 M * jjpacheco i get this in return 1170553817 M * jjpacheco No command given; use '--help' for more information. 1170553843 M * Bertl that's unusual, what distro/tool version? 1170553883 M * jjpacheco gentoo/0.30.210 1170553922 M * jjpacheco im mounting every dir for that vs from another vs 1170553931 M * Bertl I'd try to upgrade to 212 or later 1170553932 M * jjpacheco by using its fstab 1170553946 M * jjpacheco similar to a 'thin' client 1170553964 M * jjpacheco if i type vserver vs_name exec bash 1170553968 M * jjpacheco i can get in 1170553970 M * Bertl why isn't the CoW/unifications tuff sufficient for your purpose? 1170553978 M * Bertl *stuff 1170553984 M * jjpacheco CoW? 1170553989 M * jjpacheco unifications? 1170553993 M * Bertl copy on write link breaking 1170554006 M * Hollow unification (vhasify) is not implemented for gentoo, and cow is only supported in 213 1170554008 M * jjpacheco sorry, but im not familiar with that 1170554015 M * Bertl the Linux-VServer method to save disk space for similar/identical guests 1170554051 M * jjpacheco what im doing doesn't make sense? 1170554061 M * jjpacheco im mount binding every dir in the root directory 1170554070 M * jjpacheco from another vs (master_hs) 1170554083 M * jjpacheco it starts fine 1170554090 M * jjpacheco just entering is a problem 1170554098 M * jjpacheco is there a better way to do what im trying? 1170554115 M * Bertl yes, IMHO there is :) 1170554129 M * Bertl you know how hardlinks work? 1170554141 M * jjpacheco ln src dest? 1170554161 M * Bertl yep, I mean more the unix details/internals ... 1170554171 M * jjpacheco nothing specific 1170554174 M * Bertl i.e. how they behave in regard of disk usage and such 1170554177 M * jjpacheco just that it points to the same inode 1170554199 M * jjpacheco not really 1170554220 M * Bertl well, the important thing about hardlinks is, that they do not use up diskspace 1170554232 M * jjpacheco right 1170554240 M * Bertl i.e. they only need _another_ entry in a directory inode 1170554267 M * Bertl this makes it perfect for having a bunch of 'copies' of a single guest 1170554277 M * jjpacheco ok 1170554281 M * jjpacheco but for every guest 1170554290 M * Bertl but each guest wants to change a few files, e.g. logs or so 1170554294 M * jjpacheco i have to setup a whole bunch of new hardlinks no? 1170554319 M * Bertl which makes it hard to have a 1:1 copy for each of them 1170554342 M * vasko Bertl: i do not see a tag for 0.30.213-rc1 in svn repo, could you point me to the source? 1170554364 M * Hollow vasko: http://svn.linux-vserver.org/projects/util-vserver/changeset/2458 1170554432 M * vasko thanks 1170554436 M * jjpacheco Bertl: so what should i do? 1170554474 M * Bertl vasko: also:http://people.linux-vserver.org/~dhozac/p/uv/experimental/util-vserver-0.30.213-rc1.tar.bz2 1170554516 M * Bertl jjpacheco: IMHO the easiest way would be to create a template, make a copy from that for each guest, and have them unified with CoW link breaking 1170554536 M * Bertl jjpacheco: that will save you disk space and other resources, while making the guests independant 1170554548 M * jjpacheco do you have some documentation somewhere about CoW link breaking? 1170554559 M * Bertl of course, if you want to mount specific stuff read only into them, you can do that too 1170554585 M * jjpacheco for my situation i don't need to have the guests change anything in the guest vs 1170554590 M * Hollow Bertl: which unit does uptime have in ctx_stat? 1170554595 M * jjpacheco only certain directories 1170554602 M * jjpacheco which i have skipped in my system mount 1170554621 M * jjpacheco /export is independant 1170554633 M * jjpacheco but everything else should be untouched 1170554644 M * Bertl okay, should work too 1170554655 M * jjpacheco would CoW really make much of a difference in this case? 1170554675 M * Bertl if there are no changes, it won't 1170554705 M * jjpacheco so this CoW would break the file link if that guest tried to make a change? 1170554715 M * jjpacheco and save that file in its own vs? 1170554731 M * Bertl yep 1170554741 M * jjpacheco nice 1170554751 M * jjpacheco but do you knwo why i can't use vserver vs_name enter? 1170554777 M * Bertl nope, either you messed up the guest somehow, or the tools are too old 1170554777 M * jjpacheco nm, the util-vserver update fixed the problem 1170554787 M * jjpacheco thx for the help bertl 1170554793 M * Bertl np 1170555127 M * Bertl Hollow: ctx_stat does not report any uptime? 1170555141 Q * dna Quit: Verlassend 1170555225 M * Hollow VCMD_virt_stat 1170555297 M * Bertl vc_data.uptime = timespec_to_ns(&uptime); 1170555304 M * Bertl so nanoseconds 1170555317 M * Hollow ok, thanks 1170555322 M * Bertl np 1170555336 M * Bertl offset is in ns too 1170555429 M * jjpacheco k 1170555433 M * jjpacheco getting a weird error 1170555438 M * jjpacheco vserver vs_name start 1170555446 M * jjpacheco secure-mount: mount(): No such file or directory 1170555459 M * Bertl what's your fstab line? 1170555482 M * jjpacheco sorry? 1170555493 M * jjpacheco /usr/lib/util-vserver/vserver.functions: line 682: pushd: /etc/vservers/vs_020107_0006/vdir: No such file or directory 1170555505 M * jjpacheco i checked that dir 1170555507 M * Bertl well, you have some 'mount' lines, or do you call the secure-mount yourself? 1170555508 M * jjpacheco it exists 1170555534 M * jjpacheco this is just with vserver vs_name start 1170555544 M * jjpacheco just using its own fstab 1170555547 M * Bertl so it is executing the config in /etc 1170555561 M * Bertl /etc/vservers//fstab 1170555564 M * jjpacheco yes 1170555577 M * Bertl so please upload that one to paste.linux-vserver.org 1170555660 M * jjpacheco http://paste.linux-vserver.org/1037 1170555739 M * Bertl and /vservers/master_hs exists? 1170555760 M * Bertl what does 'ls -la /etc/vservers/vs_020107_0006/vdir' give? 1170555764 M * jjpacheco yes 1170555784 M * jjpacheco /etc/vservers/vs_020107_0006/vdir -> /etc/vservers/.defaults/vdirbase/vs_020107_0006 1170555803 M * Bertl and ls -la /etc/vservers/.defaults/vdirbase 1170555821 M * jjpacheco /etc/vservers/.defaults/vdirbase -> /vservers 1170555843 M * Bertl okay, I assume /vservers/vs_020107_0006 exists and is accessible? 1170555852 M * jjpacheco yup 1170555869 M * jjpacheco ls /vservers/vs_020107_0006/ 1170555875 M * jjpacheco bin dev etc export initrd lib opt proc root sbin srv sys tmp usr var 1170555878 M * Bertl nevertheless the pushd failed, so it must have some kind of permission issue 1170555898 M * Bertl first, check the dmesg log for warnings 1170555917 M * jjpacheco no errors 1170555977 M * Bertl try to run the start with --debug, and upload the output too 1170556206 M * jjpacheco any specific parts? 1170556220 M * Bertl I'd say the one around the pushd 1170556287 M * jjpacheco http://paste.linux-vserver.org/1038 1170556374 M * Bertl you already hit a problem on the rbind for the root dir 1170556386 M * Bertl so I'm pretty sure your permissions are messed up somehow 1170556396 M * Bertl let's check them one by one ... 1170556412 M * jjpacheco ok 1170556422 M * Bertl ls -lad /etc/vservers/vs_020107_0006/vdir{,.,..} 1170556452 M * Bertl showattr -lad /etc/vservers/vs_020107_0006/vdir{,.,..} 1170556465 M * jjpacheco drwxr-xr-x 19 root root 440 Feb 3 21:17 /etc/vservers/vs_020107_0006/vdir/ 1170556466 M * jjpacheco drwxr-xr-x 19 root root 440 Feb 3 21:17 /etc/vservers/vs_020107_0006/vdir/. 1170556466 M * jjpacheco drwxr-xr-x 8 root root 248 Feb 1 19:04 /etc/vservers/vs_020107_0006/vdir/.. 1170556470 M * Bertl ahem, showattr -d /etc/vservers/vs_020107_0006/vdir{,.,..} 1170556493 M * jjpacheco ---bui- /etc/vservers/vs_020107_0006/vdir/ 1170556493 M * jjpacheco ---bui- /etc/vservers/vs_020107_0006/vdir/. 1170556493 M * jjpacheco ---bui- /etc/vservers/vs_020107_0006/vdir/.. 1170556602 M * Bertl you are missing a barrier there, but that is not your problem 1170556640 M * Bertl ls -lad / /vservers /vservers/vs_020107_0006 1170557179 M * jjpacheco one sec bertl 1170557182 M * jjpacheco i think i found the problem 1170557190 M * Bertl ah? let's hear ... 1170557202 M * jjpacheco i wrote a file in etc 1170557206 M * jjpacheco call mtab 1170557528 M * Bertl and? 1170557540 M * jjpacheco one sec, having some issues 1170557545 M * Bertl np 1170558089 M * jjpacheco ok 1170558093 M * jjpacheco still doesn't work :( 1170558107 M * jjpacheco drwxr-xr-x 20 root root 488 Aug 1 2005 / 1170558107 M * jjpacheco lrwxrwxrwx 1 root root 19 Aug 1 2005 /vservers -> /export/srv/vhosts/ 1170558107 M * jjpacheco lrwxrwxrwx 1 root root 28 Feb 3 20:37 /vservers/vs_020107_0006 -> /home/virtual/vs_020107_0006 1170558268 M * jjpacheco but im going to drop this method 1170558275 M * jjpacheco i think your way is better 1170558280 M * jjpacheco the hardlinking 1170558301 M * jjpacheco so to make it work 1170558311 M * doener maybe the mount cleanup is going wild? 1170558329 M * jjpacheco do i just issue a few ln on the root directories? 1170558432 M * Bertl ah, you have several symlinks across your fs tree? 1170558442 M * jjpacheco Bertl: yes 1170558517 M * jjpacheco well 1170558519 M * jjpacheco CoW is out 1170558526 M * jjpacheco im running 2.0.x 1170558534 M * jjpacheco can't really update on this machine 1170558535 M * jjpacheco :( 1170558580 M * Bertl I'd start with a simple filesystem setup 1170558589 M * jjpacheco got one 1170558592 M * jjpacheco master_hs 1170558597 M * Bertl (not with 20 symlinks pointing around :) 1170558599 M * jjpacheco full blown debian install 1170558603 M * jjpacheco oh that 1170558605 M * jjpacheco lol 1170558618 M * jjpacheco yeah, cause of partitioning 1170558621 M * Bertl then make sure it works, try a single dir as bind mount first 1170558631 M * Bertl something uncritical like /usr 1170558644 M * jjpacheco ic 1170558652 M * jjpacheco just step through it 1170558861 M * Bertl Hollow: could you give the new shiny a spin with vcd and friends? 1170558942 M * Bertl http://vserver.13thfloor.at/Experimental/SYSCALL/syscall_shiny14e.h 1170560427 M * jjpacheco Bertl: is there documentation on CoW linking? 1170560433 M * jjpacheco making templates, etc 1170560692 M * Bertl check out the wiki, there are some pages on the old and new wiki 1170560704 M * Bertl best search for unification/vhashify/CoW 1170560728 M * Bertl okay, I'm off to bed now ... have a good one everyone! cya! 1170560736 N * Bertl Bertl_zZ 1170561577 J * shuri ~shuri@hq01.electronicbox.net 1170561898 Q * shuri Remote host closed the connection 1170563265 Q * q\ Remote host closed the connection 1170566653 J * Aiken ~james@ppp220-70.lns2.bne1.internode.on.net 1170572455 Q * MrX Ping timeout: 480 seconds 1170572641 J * MrX ~chaos@179.13.95.219.kmr01-home.tm.net.my 1170574348 J * rpc ~no@ajl37.neoplus.adsl.tpnet.pl 1170576737 J * jmcaricand ~kvirc@d83-179-134-241.cust.tele2.fr 1170576792 Q * MrX Ping timeout: 480 seconds 1170578465 Q * jmcaricand Quit: KVIrc 3.2.4 Anomalies http://www.kvirc.net/ 1170582038 Q * thunder1 Remote host closed the connection 1170582173 J * bonbons ~bonbons@83.222.37.103 1170582318 J * thunder1 ~thu@tor-irc.dnsbl.oftc.net 1170582750 J * oleg_erm ~someone@ppp265.pool1.as05.cn.tts.khb.dial-up.kht.ru 1170582764 Q * weeze Read error: Connection reset by peer 1170583262 Q * Aiken Ping timeout: 480 seconds 1170583435 J * Aiken ~james@ppp220-70.lns2.bne1.internode.on.net 1170583794 N * Bertl_zZ Bertl 1170583798 M * Bertl morning folks! 1170583826 M * rpc morning Bertl! 1170584231 M * Aiken good evening 1170584324 M * Bertl hey rpc! Aiken! 1170584750 M * Aiken I seem to have an etch guest under a modified rh6.2 machine running 2.4.32-vs1.2.10 1170584767 M * Bertl nice 1170585117 J * someone_ someone@ppp300.pool1.as03.cn.tts.khb.dial-up.kht.ru 1170585149 M * Bertl welcome someone_! 1170585168 N * someone_ oleg 1170585465 Q * oleg Read error: Connection reset by peer 1170585542 Q * oleg_erm Ping timeout: 480 seconds 1170586039 J * dna ~naucki@95-205-dsl.kielnet.net 1170586736 N * virtuoso_ virtuoso 1170587236 Q * dmax Ping timeout: 480 seconds 1170587875 Q * dlezcano Ping timeout: 480 seconds 1170588011 Q * Aiken Quit: Leaving 1170588858 M * Hollow Bertl: shiny 14e seems to work here (x86_64 that is) 1170588883 M * Bertl okay, great! can you test with a 32bit compile too? 1170588889 M * Hollow yup 1170588905 M * Bertl is that with the PIC stuff now? 1170588962 Q * thunder1 Ping timeout: 480 seconds 1170588977 J * thunder1 ~thu@tor-irc.dnsbl.oftc.net 1170588986 M * Hollow Bertl: yep, shared library with PIC enabled 1170589000 M * Bertl good, sow e will get a test for that with x86 too then 1170589013 M * Hollow work on x86 too here 1170589025 M * Bertl excellent, gcc 4.1.x? 1170589028 M * Hollow yep 1170589045 M * Hollow though the amd64 one is gcc 3.x iirc 1170589055 M * Bertl okay, I guess that should not matter too much 1170589059 M * Hollow yep 3.4.6 1170589063 M * Bertl the x86 is the critical part for PIC 1170589082 M * Hollow ok, but the 3.4.6 one is a ahrdened toolchain, so that works too ;) 1170589085 M * Bertl so I assume we have alpha, x86 and x86_64 somewhat tested regarding shiny14 1170589087 M * Hollow *hardened 1170589093 M * Bertl good 1170589157 M * Bertl yang should have the sparc/64 ready soon, then I guess we are fine for including that instead of the current one 1170589180 M * yang yes i have sparc ready 1170589186 M * yang and sgi will be online in 30 min. 1170589207 M * Bertl perfect, the sparc has already a Linux-VServer kernel= 1170589210 M * Bertl s/=/? 1170589216 M * yang nope it doesn't 1170589228 M * Hollow did you test ppc? unfortunately my ppc is osx only currently .. 1170589236 M * Bertl yang: okay, what kernel do we have there, atm? 1170589253 M * yang Bertl: i must check, i think its 2.6.8 debian stable 1170589254 M * Bertl Hollow: not yet, but my powerbook will be available in a few hours :) 1170589259 M * Hollow ok :) 1170589272 M * yang Bertl: a 1170589282 M * yang I will put both online and check them 1170589291 M * Bertl yang: okay, tx 1170589302 M * yang Bertl: Its a sparcstation 4 or 5 - 100mhz 1170589324 M * Bertl once we have a working config, the cross compiling is no real issue 1170589345 M * yang Bertl: but i wouldn't know how to compile a proper kernel for the sparc/sgi, sometimes it doesnt even work with default debian kernel update 1170589349 M * Bertl we just want to avoid compiling the kitchensink config (i.e. deboan) 1170589378 M * Bertl we will figure that, I just need your support on site ... 1170589902 Q * duckx Read error: Connection reset by peer 1170589989 M * nebuchadnezzar hello 1170590005 M * Bertl hey nebuchadnezzar! 1170590017 M * nebuchadnezzar how are you guys ? 1170590027 M * Bertl you are sparc64/ubuntu or something like that? 1170590034 M * nebuchadnezzar spar64/debian 1170590050 M * Bertl you had some issues with the tools not working properly, right? 1170590055 M * nebuchadnezzar yes 1170590070 M * Bertl I might have a proper fix for that 1170590084 M * nebuchadnezzar good 1170590088 M * Bertl do you have some time for testing? 1170590103 M * nebuchadnezzar yes 1170590107 M * Bertl excellent! 1170590131 M * Bertl IIRC, you tried with vcmd too, yes? 1170590137 M * nebuchadnezzar yes 1170590147 M * Bertl that did fail too? 1170590153 M * nebuchadnezzar right 1170590163 M * Bertl good, try to make that fail again :) 1170590172 M * Bertl i.e. what about ./vcmd -V 1170590210 M * nebuchadnezzar vc_get_version: Unknown error 4160581024 1170590227 M * Bertl good, now let's replace the shiny_new.h with this one: 1170590247 M * Bertl http://vserver.13thfloor.at/Experimental/SYSCALL/syscall_shiny14e.h 1170590259 M * Bertl (you can jsut copy it there, and change the symlink) 1170590306 M * nebuchadnezzar I have warnings for implicit declaration of htonl htons ntohs 1170590316 M * Bertl that's probably fine 1170590370 M * nebuchadnezzar it will be better with diet installed :-) 1170590383 M * Bertl should not matter either 1170590396 M * nebuchadnezzar vc_get_version: Function not implemented 1170590404 M * nebuchadnezzar arf, need to be root 1170590413 M * Bertl yup 1170590483 M * nebuchadnezzar I'm not sure it was not working with vcmd now, 1170590500 M * nebuchadnezzar with shiny10 it works 1170590521 M * nebuchadnezzar and with shiny14e it works too 1170590527 M * Bertl well, the Unknown error 4160581024 was weird 1170590536 M * nebuchadnezzar just non root execute 1170590543 M * Bertl Function not implemented or the proper VCI result is fine 1170590570 M * Bertl let's try the syscall_shiny14e with util-vserver (with and without diet) 1170590616 M * Bertl you need to replace the lib/syscall-alternative.h 1170590619 M * nebuchadnezzar ok 1170591656 M * nebuchadnezzar it works 1170591663 M * Bertl excellent! 1170591676 M * nebuchadnezzar shiny14e with NOLEGACY 1170591688 M * Bertl is that dietlibc now? 1170591692 M * nebuchadnezzar on util-vserver 0.30-212 1170591697 M * nebuchadnezzar yes dietlibc powered 1170591710 M * Bertl okay, could you try with --disable-diet too? 1170591716 M * nebuchadnezzar ok 1170591720 M * Bertl TIA 1170591890 J * ema ~ema@lart.galliera.it 1170591898 M * Bertl welcome ema! 1170591924 M * ema Bertl: Thanks :) 1170592095 M * yang Bertl: the sparc runs a 32 bit kernel 1170592106 M * matti :) 1170592109 M * matti Bertl: :) 1170592116 M * Bertl yang: okay, did you try with a 64bit one yet? 1170592122 M * yang nope 1170592204 M * nebuchadnezzar Bertl: it works too 1170592205 M * yang and the sgi gives me some error noise on bootup, i must see what it is 1170592227 M * Bertl yang: okay, take your time, maybe nebuchadnezzar can give some hints regarding sparc64 too? 1170592233 M * Bertl nebuchadnezzar: excellent, tx! 1170592524 M * yang Bertl: hmmm, do you know what is the default baud rate for terminal access for SGI's I only see "starting diagnostics" after that there is noise and scrambled text 1170592561 M * Bertl no idea, either there is some OF which controls that, or it is an internal default 1170592566 M * Bertl try 9600 and 19200 1170593097 M * PowerKe I'm trying vserver build (used to use vserver-new) and getting: Unsupported packaging method: application/empty (application/x-bzip2) 1170593113 M * Bertl show me the line 1170593123 M * PowerKe vserver services build --context 4 --hostname services --interface dummy0:192.168.201.4/24 --initstyle gentoo -m template -- -d gentoo -t /vservers/stage3-i686-20060317.tar.bz2 1170593159 M * PowerKe Taken from Hollow's gentoo vserver-howto page and added --initstyle gentoo (would otherwise complain about that) 1170593166 M * Bertl what tools version? 1170593181 M * PowerKe 0.30.212 1170593191 M * Bertl and does /vservers/stage3-i686-20060317.tar.bz2 exist and contain a proper stage3? 1170593217 M * PowerKe from wget http://ftp.belnet.be/mirror/rsync.gentoo.org/gentoo/experimental/x86/vserver/stage3-i686-20060317.tar.bz2 1170593234 M * PowerKe tar -tjf doesn't show any problems 1170593244 M * Hollow PowerKe, Bertl: https://bugs.gentoo.org/show_bug.cgi?id=164057 1170593291 M * Bertl lol 1170593298 M * matti ;-p 1170593311 M * Bertl Hollow: how could file be wrong on that? 1170593325 M * Bertl PowerKe: file /vservers/stage3-i686-20060317.tar.bz2 1170593329 M * Hollow i have no idea 1170593359 M * PowerKe /vservers/stage3-i686-20060317.tar.bz2: bzip2 compressed data, block size = 900k 1170593364 M * matti The mystery of the universe ;] 1170593396 M * Bertl PowerKe: and with -i? 1170593406 M * PowerKe /vservers/stage3-i686-20060317.tar.bz2: application/x-bzip2 1170593419 M * Bertl so that looks fine to me, no? 1170593424 M * matti Yep. 1170593429 M * Hollow file -xi 1170593440 M * PowerKe file: invalid option -- x 1170593442 M * Hollow ar.. file -zi 1170593453 M * PowerKe /vservers/stage3-i686-20060317.tar.bz2: application/x-tar, POSIX (GNU) (application/x-bzip2) 1170593480 M * Bertl so where does the application/empty come from? 1170593481 M * Hollow hm .. that used to return application/empty 1170593491 M * matti matti@acrux ~ $ file -zi stage3-i686-20060317.tar.bz2 1170593491 M * matti stage3-i686-20060317.tar.bz2: application/x-tar, POSIX (application/x-bzip2) 1170593515 M * matti matti@acrux ~ $ file --version 1170593515 M * matti file-4.17 1170593533 M * matti I wonder, why there's no "GNU" thingie. 1170593535 M * matti :> 1170593548 M * Hollow please try: file -Nbiz -m /usr/lib/util-vserver/magic.mime /vservers/stage3-i686-20060317.tar.bz2 1170593565 M * PowerKe file: could not find any magic files! 1170593579 M * Hollow amd64 host? 1170593589 M * PowerKe intel, 32bit 1170593617 M * Hollow hm 1170593618 Q * yang Quit: KVIrc 3.2.4 Anomalies http://www.kvirc.net/ 1170593635 M * Hollow this is the command line util-vserver uses 1170593662 M * Bertl well, no wonder it fails then :) 1170593684 M * matti ;] 1170593699 M * Hollow ah, no.. the file is /usr/lib/util-vserver/magic 1170593717 M * PowerKe /usr/lib/util-vserver/magic.mime does exist 1170593733 M * Hollow thanatos ~ # file -Nbiz -m /usr/lib/util-vserver/magic /etc/passwd 1170593733 M * Hollow text/plain; charset=us-ascii 1170593736 M * PowerKe file -Nbiz -m /usr/lib/util-vserver/magic /vservers/stage3-i686-20060317.tar.bz2 1170593740 M * PowerKe application/x-tar, POSIX (GNU) (application/x-bzip2) 1170593751 M * Hollow still not empty 1170593768 M * Hollow this bug drives me nuts 1170593780 M * Bertl maybe >/vservers/stage3-i686-20060317.tar.bz2 will help there (don't try it :) 1170593803 M * Hollow :P 1170593807 M * matti ;-p 1170593861 M * matti Hollow: Try to run this script in few loops. Maybe it is only a matter of adding one extra sleep 1? 1170593967 M * PowerKe Would it help if i repackage as .tar.gz ? 1170593974 M * Hollow lol, very funny .. 1170593980 M * matti Hollow: Maybe there's a race somewhere and file indeed exists (from i-node and so on point of view) somewhere on the file system, but have no content yet? It is only a wild guess. 1170593981 J * yang ~yang@yang.sponsor.oftc.net 1170593988 M * matti PowerKe: Heheh, I don't think so ;) 1170593996 M * Hollow http://paste.linux-vserver.org/1045 1170594019 M * matti Oh. 1170594031 M * matti You get empty. 1170594032 M * matti LOL 1170594035 Q * ziga Quit: leaving 1170594038 M * matti That is insane. 1170594038 M * matti ;] 1170594048 M * Hollow indeed 1170594059 M * matti I wonder, what file is doing. 1170594064 M * matti Unpacking this? 1170594077 M * matti No... that would be too slow. 1170594106 M * yang Bertl: I think I had a broken ttyS0 connection, after reboot it worked 1170594130 M * matti Hollow: Try it again few times, but this time set strace on it. 1170594133 M * yang Bertl: going to lunch now, the server will be operational in 1/2 hours 1170594138 M * PowerKe Trying -Nbiz -m several times in a row fails on the 7th attempt 1170594155 M * PowerKe trying vserver build several times never succeeds though :( 1170594161 M * matti Hollow: I wonder, where file is unable to (I am not sure) "read" content? Whatever it is doing :] 1170594192 M * Bertl yang: np, tx! 1170594200 M * matti PowerKe: No worries mate. Keep on with positive thinking :) 1170594216 M * matti PowerKe: This is obvious file problem, probably some nasty bug. 1170594218 M * matti :> 1170594241 M * matti PowerKe: I mean "file" as an application. 1170594242 M * matti ;] 1170594248 M * matti Hollow: Any luck? 1170594253 M * Hollow sec 1170594256 M * matti Aye. 1170594275 M * Hollow have to install strace first ;) 1170594281 M * matti Oh ;p 1170594282 M * matti Shame ;p 1170594285 M * matti Mr Developer ;p 1170594286 M * matti :D 1170594295 M * Hollow well, another box .. 1170594302 M * matti Heheh, no worries :) 1170594307 A * matti hugs Hollow :) 1170594333 M * matti Hollow: Don't blame drugs this time ;p 1170594339 M * matti :> 1170594379 M * Hollow well, can't reproduce it now 1170594398 M * PowerKe I can run strace if you like ? 1170594403 M * Hollow yeah 1170594415 M * Hollow strace -ff -o file.log -fF file ... 1170594700 J * rgl ~Rui@84.90.10.107 1170594705 M * rgl hello 1170594761 M * PowerKe Hollow: http://powerpnt.dyndns.org:8080/vserver/file.log.bz2 1170594771 M * rgl is there a way to tcpdump internal traffic? I mean, I have apache that proxys to an ip address on the same vserver, how could I tcpdump that proxied traffic? 1170594814 M * matti Hollow: I have looped file ... for 100 times, and works perfectly. No, I am looping it for 1000 times. 1170594820 M * rgl oh, using the lo interface seems to work 1170594829 M * matti Hollow: Try to downgrade file maybe... 1170594843 M * matti Hollow: Mine: file-4.17 1170594850 M * PowerKe laptoptom ~ # file --version 1170594850 M * PowerKe file-4.18 1170594851 M * PowerKe magic file from /usr/share/misc/file/magic 1170594867 M * matti PowerKe: Nice hostname ;p 1170595087 M * PowerKe Maybe I could try downgrading file to 4.17-r1 ? 1170595209 M * matti Yep, give it a rey. 1170595214 M * matti s/rey/try 1170595222 M * matti 290:application/x-tar, POSIX (application/x-bzip2) 1170595244 M * matti Loop number 290 and nothing. 1170595407 M * matti 440 1170595409 M * matti ;] 1170595420 M * PowerKe works 1170595439 M * PowerKe Tried 10 in a row, no problem 1170595442 M * PowerKe Now vserver build is running 1170595449 M * PowerKe seems to go fine 1170595466 M * matti :) 1170595478 M * matti So, they break 4.18 somehow. 1170595549 M * matti 505 and still everything is as it should be ;] I will interrupt this now ;p 1170595549 M * PowerKe >>> Found baselayout-1.6.14 1170595550 M * PowerKe !!! Will not do automagic changes to baselayout < 1.13 1170595550 M * PowerKe !!! You have to take care for yourself ... 1170595594 M * PowerKe Anything special or is it just hostname / domainname stuff? 1170595598 M * matti PowerKe: Look at the https://bugs.gentoo.org/show_bug.cgi?id=164057 :> 1170595623 M * matti PowerKe: At the bottom to be more preciesly. 1170595684 M * PowerKe thanks 1170595690 M * matti :) 1170595862 M * PowerKe same message 1170595882 M * PowerKe vserver start / enter works fine though 1170595888 M * matti :> 1170595908 M * matti PowerKe: Yep, and you have message, because you still have the same baselayout ;] 1170595924 Q * yarihm Remote host closed the connection 1170595939 M * matti PowerKe: The key is to have initstyle set to plain :) That's why start/stop works fine now ;) 1170595946 M * matti PowerKe: IIRC. 1170595947 M * matti :> 1170596087 M * matti [14:34:31] Irssi uptime: 42d 23h 26m 33s 1170596090 M * matti Umh. 1170596094 M * matti Too much IRC ;p 1170596541 M * PowerKe hmm, there's a good chance that file only fails when it's running on a loaded system 1170596575 M * matti No. 1170596602 M * PowerKe between installing file-4.18 and file-4.17 I also reniced a process 1170596610 M * matti PowerKe: 4.17 and 4.17-r1 works perfectly on machine with constant medium load. 1170596623 M * matti PowerKe: On the same machine 4.18 is randomly reporting empty. 1170596644 M * matti PowerKe: But I am not sure for 100%. 1170596645 M * matti ;] 1170596722 M * PowerKe True, it should never fail, not even under load. But when the system is not loaded I think there's a better chance of vserver build succeeding 1170596780 M * matti PowerKe: Well... This is not an solution though. 1170596808 M * matti PowerKe: We cannot just warn users "please, build your vserver thingie on 0.0 load machine" :) 1170596811 M * matti Heheh. 1170596838 M * PowerKe Thinking of it, I should retry 4.17 with some load and see if still works 1170596841 M * matti PowerKe: Bug still applies - not only for vserver tarball. 1170596846 M * matti PowerKe: Aye. 1170596849 M * matti PowerKe: :) 1170596905 M * matti PowerKe: for i in $(seq 1 100); do echo -n "${i}:"; file -Nbiz -m /usr/lib/util-vserver/magic /vservers/stage3-i686-20060317.tar.bz2; done 1170596935 M * matti PowerKe: And on the second console try to do some load ;] 1170597006 M * matti PowerKe: Like cat /dev/hda > /dev/null or something. 1170597026 M * PowerKe Well, I have a vserver init process that is using up 100%, but when reniced it's not that bad :) 1170597044 M * PowerKe And how I did that? Well, cleaning up diskspace when you're only a minute awake is a bad thing... 1170597062 M * PowerKe Killed all files with only 1 hardlink in /vservers instead of /vservers/.hash 1170597073 M * PowerKe :) 1170597078 M * matti ;] 1170597106 M * matti PowerKe: Do you want some coffee? 1170597106 M * matti ;] 1170597123 M * matti Hollow: Coffee? 1170597124 M * matti Bertl: ? 1170597137 M * PowerKe I don't drink coffee... maybe that's part of the problem :p 1170597153 M * Bertl nah, have to leave now ... will be back later though ... 1170597162 N * Bertl Bertl_oO 1170597856 M * PowerKe 4.17 seems to hold up under load 1170597871 M * PowerKe 4.18 fails more than 50% now 1170598012 M * PowerKe Out of 100 runs, only 23 were correct 1170598300 M * PowerKe 4.19 seems to give the same results as 4.18 1170598667 M * matti ;] 1170598731 M * PowerKe I'll add to the bug report that for now you could use 1170598743 M * PowerKe emerge --oneshot " 1170599763 Q * derjohn Quit: by(t)e 1170599889 Q * ema Quit: leaving 1170600313 M * shedi is there a config to disable the attempt umount and ifdown when stopping a guest 1170600756 J * yarihm ~yarihm@84-74-16-225.dclient.hispeed.ch 1170601447 Q * nebuchadnezzar Quit: ERC Version 5.1.4 (IRC client for Emacs) 1170601657 M * matti shedi: None that I know of. 1170601682 M * matti shedi: Why do you want to prevent the umounting and stopping interface? 1170601704 M * shedi yes, I was hoping for a less disturbing shutdown 1170601733 M * shedi I don't want the guest to inform that it's actually trying to do those things :) 1170601803 M * matti Oh, so you want to suppress this messages but leave functionality as it was? Or kill them both! ;p 1170601944 M * shedi heh 1170601968 M * shedi no I was hoping to stop the attempt 1170602109 J * dlezcano ~dlezcano@AToulouse-252-1-110-138.w86-217.abo.wanadoo.fr 1170602174 M * PowerKe Can't kill init of a running guest :( 1170602176 M * PowerKe vps -ef | grep 9803 1170602177 M * PowerKe root 9803 2 sandbox 1 91 12:47 ? 03:08:28 init boot 1170602190 M * PowerKe chcontext --ctx 2 kill -9 9803 --> nothing happens 1170602197 M * PowerKe vkill -s 9 9803 1170602198 M * PowerKe vkill: vc_ctx_kill(): No such process 1170602245 M * PowerKe Guess I'll just remove autostart and reboot 1170602713 M * PowerKe Yup, that killed it :p 1170602934 M * matti ;p 1170602956 M * matti shedi: I have no clue then. Bug Hollow about this :) 1170602978 M * Hollow kernel version? 1170603036 M * PowerKe Hollow: mine or shedi's? 1170603054 M * Hollow yours 1170603065 M * PowerKe 2.6.19-vs2.2.0-rc8-gentoo 1170603089 M * PowerKe But I really screwed up that guest 1170603096 M * PowerKe Just deleted it 1170603112 M * PowerKe Problem was that I had cleaned out all files with only 1 hardlink in /vservers instead of /vservers/.hash 1170603153 J * nebuchadnezzar ~nebu@zion.asgardr.info 1170603327 Q * rgl Ping timeout: 480 seconds 1170603378 M * Hollow PowerKe: well, there is a vkill bug in rc8 iirc 1170603575 M * PowerKe I'll upgrade to rc9 and try it again 1170603592 M * PowerKe (seems like it fails on a "normally working" guest as well) 1170604816 J * rgl ~Rui@84.90.10.107 1170607008 J * jmcaricand ~kvirc@d83-179-131-76.cust.tele2.fr 1170607300 M * jmcaricand Hi. I want to kown if I can remove /tmp from fstab for a guest ? 1170607710 J * shuri ~shuri@hq01.electronicbox.net 1170607782 M * Hollow recently seen in berlin .. http://j03.de/stuff/2007-02-04/vista-lkw.jpg (german only) ;) 1170607952 M * FaUl lol 1170608468 J * IGAAB ~OpenBrasi@200.181.157.233 1170608698 P * IGAAB 1170608897 M * jmcaricand I want to use ACL on /tmp in a guest. ACL don't work with tmpfs. So if I can remove without problem /tmp from fstab for a guest ? 1170609164 N * Bertl_oO Bertl 1170609184 M * Bertl jmcaricand: acls work fine with tmpfs IIRC, you just have to enable them 1170609202 M * Bertl evening folks! 1170609495 M * jmcaricand Bert: I made some tests. It seems ACL don't work in guest if I have this line in fstab : "none /tmp tmpfs size=16m,mode=1777 0 0" 1170609533 M * jmcaricand Bert: ACL works on others directory. 1170609591 M * jmcaricand Bert: I want to know if it's dangerous to remove this line for my guest ? 1170609761 M * Bertl no, it is not dangerous, just bad for performance :) 1170609779 M * Bertl but as I said, if you enable acls for tmpfs, they will work too 1170609848 J * wogri ~wogri@131.130.7.148 1170609862 M * Bertl welcome wogri! 1170609869 M * wogri servas bertl! 1170609886 M * wogri it's that time again... vserver upgrade, and I need your help :) 1170609927 J * baldy ~baldy@83.223.66.130 1170609927 M * Bertl let's see what we can do for you then :) 1170609932 M * Bertl wb baldy! 1170609942 M * baldy huhu Bertl 1170609943 M * wogri i'm moving to 2.2.0-rc8 (i know rc10 is out in the meantime), using util-vserver 0.30.211-6 1170609950 M * wogri so... 1170609965 M * Bertl why an older tool version? 1170609972 M * wogri as the old config is legacy I'm going to re-create the config-files. 1170609982 M * wogri the older tool version because it's in debian testing. 1170609995 M * Bertl I hope they have 0.30.212 there too? 1170610025 M * baldy Bertl: whats the changed friom 2.2.0-rc9 to rc10? 1170610043 M * baldy chnages... 1170610044 M * baldy hrr 1170610063 M * wogri probably in unstable, which wouldn't be my way to go. but I think my problem is not util-vserver related, it's rather somehting I don't know about vserver :) I'd compile util-vserver sources if it's necessary to do so. 1170610066 M * baldy <-- tired lol 1170610089 M * Bertl two 64bit issues (longer formats, full atomic width for accounting) were ironed out 1170610106 M * wogri the isse is: I created a config-directory like this: vserver build build -m skeleton -n ldap --hostname ldap --interface eth0:192.168.13.1/24 1170610109 M * baldy oke not importent for me ;) 1170610118 M * wogri created a context-file with the content of '1'. 1170610134 M * wogri vserver ldap start gives: vnamespace: vc_set_namespace(): Operation not permitted 1170610148 M * Bertl wogri: xid=1 is used for the spectator context 1170610156 M * wogri oh. 1170610157 M * Bertl choose a number between 2 and 49151 1170610190 M * jmcaricand I change line to "none /tmp tmpfs acl,size=16m,mode=1777 0 0" but I receive this error : "secure-mount: mount(): Invalid argument" 1170610190 M * wogri okay. 1170610198 M * wogri and there it goes! 1170610204 M * wogri thanks folks :) 1170610212 M * wogri that was easy, wasn't it? :) 1170610285 M * wogri yeah, works like a charm. 1170610294 M * wogri bertl, you are - as always - my hero :) 1170610304 M * wogri thank you. 1170610312 M * Bertl you're welcome! 1170610329 M * Bertl jmcaricand: did you enable acls in the kernel= 1170610334 M * Bertl s/=/? 1170610335 M * wogri I hope you enjoyed the nice weather today in vienna! 1170610342 M * jmcaricand Yes 1170610367 M * Bertl wogri: I wasn't in vienna today, and the weather in my room was moderately fine :) 1170610383 M * wogri heh :) oh sorry I thought you're from vienna... 1170610390 Q * dna Quit: Verlassend 1170610413 M * Bertl wogri: I lived there for some time ... 1170610428 M * Bertl wogri: (and actually I was born there :) 1170610547 M * wogri but you moved out... not very common for a vienna-born-human :) [I don't know if off-topic chat is tolerated here... sorry] 1170610642 M * Bertl guess we are quite liberal about the chat topics ... 1170610671 M * wogri allright ;) 1170610720 M * wogri oh, yeah, there's one more thing: would you consider 2.2.0-rc10 production-ready? when do you plan to release the rc? (i haven't read the ml lately, sorry...) 1170610752 M * Bertl yes, it is more than production ready 1170610762 M * Bertl there will be an rc11 today 1170610771 M * Bertl (with a minor fix + cleanup) 1170610784 M * Bertl and hopefully that will become 2.2.0 next week 1170610789 M * wogri good news! hooorayyy! 1170610826 M * Bertl IMHO 2.2.0 will be one of the best releases we ever had 1170610841 M * Bertl (unless some killer bug sneaks in :) 1170610847 M * wogri okay, so I will upgrade to rc10/11 today, and will - of course - let you know if I find anything I would consider a bug. I'm on a testing-system this time, so I have time to fiddle around. 1170610903 M * Bertl please report anything which even remotely looks like a bug 1170610932 M * wogri will do. 1170610934 M * Bertl it's no problem at all, if we figure that it isn't Linux-VServer related (happens most of the time) 1170610964 M * Bertl jmcaricand: CONFIG_TMPFS_POSIX_ACL=y ? 1170611741 M * wogri bye Bertl, thanks for you help. Going to test hard now... 1170611743 Q * wogri Quit: leaving 1170612170 J * kasper ~kawsper@0x535e8019.hgnxx4.adsl-dhcp.tele.dk 1170612359 Q * shedi Quit: Leaving 1170612687 J * debugger ~Rui@84.90.10.107 1170612802 Q * shuri Remote host closed the connection 1170612893 M * jmcaricand Bert: I can't find this option (CONFIG_TMPFS_POSIX_ACL). Where is it ? 1170612983 Q * rgl Ping timeout: 480 seconds 1170613263 M * Bertl in the kernel, fs/Kconfig 1170613279 M * Bertl says, depends on TMPFS, select GENERIC_ACL 1170613659 J * alexx ~alexx@proxy.ikse.net 1170614155 J * shedi ~siggi@ftth-237-144.hive.is 1170614307 J * meandtheshel1 ~markus@85-124-232-194.work.xdsl-line.inode.at 1170614367 J * dna ~naucki@89-213-dsl.kielnet.net 1170614444 M * PowerKe is vkill supposed to require -c when killing a process running in a guest? It always returns "vkill: vc_ctx_kill(): No such process" when I try without (2.2.0-rc9). 1170614474 M * jmcaricand Bert: I found CONFIG_TMPFS=y but nothing on GENERIC_ACL or CONFIG_TMPFS_POSIX_ACL in fs/Kconfig ! 1170614626 M * Bertl well, then you are using an older (outdated?) kernel, I suppose? 1170614640 M * Bertl 2.6.19.2 has it here, and it seems to work just fine ... 1170614650 M * PowerKe jmcaricand: File Systems --> Pseudo filesystems --> Virtual memory file system support (former shm fs) --> Tmfs POSIX Access Control Lists 1170614669 M * Bertl PowerKe: yes 1170614680 M * Bertl (regarding the context id ) 1170614760 M * PowerKe Since vkill --help lists it between [] and I guess I'd use kill to kill something from the host. 1170614827 Q * kasper Ping timeout: 480 seconds 1170615510 M * jmcaricand PowerKe: I don't have submenu in <*>Virtual memory file system support 1170615589 M * PowerKe jmcaricand: What kernel version? 1170615622 M * jmcaricand PowerKe: 2.6.17.13 1170615632 M * Bertl PowerKe: best to contact daniel_hozac, in theory, most processes could be killed by asking for their context id first 1170615644 M * Bertl jmcaricand: ancient :) 1170615660 M * Bertl jmcaricand: that does indeed not support acls on tmpfs 1170615691 J * ema ~ema@lart.galliera.it 1170615697 M * Bertl wb ema! 1170615743 M * jmcaricand Bertl: Ok. Tomorrow, I migrate to the lastest. Thank and good night. 1170615877 M * Bertl have a good one! 1170615986 Q * jmcaricand Quit: KVIrc 3.2.4 Anomalies http://www.kvirc.net/ 1170616588 J * bronson ~bronson@adsl-75-36-145-166.dsl.pltn13.sbcglobal.net 1170617045 Q * alexx Quit: bye 1170617236 M * daniel_hozac hey guys. so what did i miss? 2.6.20 is out, i see... 1170617308 M * Bertl we didn't see that yet :) 1170617324 M * baldy lol 1170617341 M * Bertl means, we should probably get a 2.6.20 version up and running too :) 1170617349 M * daniel_hozac hehe 1170618274 J * dphillips ~chatzilla@74-128-38-221.dhcp.insightbb.com 1170618279 N * dphillips DSpair 1170618318 M * DSpair Good afternoon all... 1170618388 M * Bertl welcome DSpair! 1170618427 M * DSpair I have a question about networking in VServer. Can I create a "virtual" network between several virtual hosts inside of a single machine? Basic concept I am try to acheive is to have 3 virtual hosts, each has 2 interfaces. One interface is attached to a real network card, and the other is a virtual LAN which is just used to connect the 2 VPSs to each other. 1170618455 M * DSpair s/2 VPSs/3 VPSs/ 1170618645 M * Bertl there are no 'virtual' lans in Linux-VServer (yet) 1170618689 M * Bertl but of course, you can setup several network addresses 1170618700 M * Bertl and you can put a private network on dummy0 for example 1170618714 M * Bertl (which will look to the guests like a 'virtual' network) 1170618740 J * Loki|muh_ loki@satanix.de 1170618740 Q * Loki|muh Read error: Connection reset by peer 1170618750 N * Loki|muh_ Loki|muh 1170618848 M * daniel_hozac however, there's not really any point to doing so, as any traffic between guests will use the lo interface anyway. 1170618874 J * kasper ~kawsper@0x535e8019.hgnxx4.adsl-dhcp.tele.dk 1170618881 M * Bertl welcome kasper! 1170618964 M * ema Maybe it's a dumb question, but what happens if you sniff the traffic on one of the VPSs? 1170619113 M * Bertl ema: how would you do that? 1170619194 M * ema Bertl: for instance with tcpdump on the loopback interface 1170619213 M * ema Bertl: daniel_hozac said that all the traffic uses lo, that's why I'm asking 1170619214 M * daniel_hozac guests do not have the capability required to use that. 1170619224 M * ema Ah, cool :) 1170619291 M * daniel_hozac Bertl: have you run 2.2.0 through PLM yet? 1170619318 M * Bertl I knew I had forgotten something ... 1170619351 M * daniel_hozac hehe. 1170619356 M * Bertl will do so now ... 1170619360 M * daniel_hozac okay. 1170619648 M * daniel_hozac i assume you solved the problems with test-syscall on x86_64? 1170619666 M * daniel_hozac (as i saw Hollow reported success using it :)) 1170619669 M * Bertl yep, and other archs too, I hope 1170619690 M * Bertl btw, regarding the syscall test with peername and sendto ... 1170619726 M * daniel_hozac it was a bad idea to use socket things... 1170619736 M * Bertl nah, I like it ... 1170619743 M * daniel_hozac well, it's not very portable. 1170619755 M * Bertl and it seems to work (mostly) for x86_64 after some minor fixes 1170619772 M * Bertl I also made it libc independent, I think :) 1170619796 M * Bertl but I actually have a problem with the sendto using the peername addr 1170619804 M * Bertl which imho cannot work on pipes? 1170619812 M * ntrs_ daniel_hozac, is there a flower page or some doc for .213? 1170619817 M * ntrs_ for the config I mean. 1170619818 M * daniel_hozac ntrs_: doc/configuration.html 1170619822 M * meandtheshel1 daniel_hozac: PLM == Product Lifecycle Management? 1170619823 Q * dlezcano Ping timeout: 480 seconds 1170619838 M * daniel_hozac meandtheshel1: s/Product/Patch/;s/Management/Manager/, IIRC. 1170619849 M * meandtheshel1 :) 1170619871 M * daniel_hozac (http://plm.testing.osdl.org/) 1170619882 M * meandtheshel1 daniel_hozac: thx 1170619902 M * daniel_hozac Bertl: yeah, i guess not. 1170619906 M * Bertl daniel_hozac: tx, didn't find it anymore :) 1170619940 M * daniel_hozac Bertl: i was trying to take a shortcut by using socketpair, but i guess it needs to be done manually. 1170619959 M * Bertl daniel_hozac: but if you have 'better' syscalls to use, I'm always open for suggestions 1170619975 M * Bertl just check the changes I did so far, they also fix the compiler warnings 1170620341 M * baldy last bug to fix 1170620346 M * baldy then my images will mork 1170620348 M * baldy work 1170620471 M * Bertl excellent! 1170620494 M * Bertl http://plm.testing.osdl.org/patches/show/5793 1170620500 M * daniel_hozac great! 1170620507 M * Bertl (will take some while :) 1170620518 M * daniel_hozac yeah, always does, heh. 1170620563 J * dlezcano ~dlezcano@AToulouse-252-1-110-138.w86-217.abo.wanadoo.fr 1170621290 Q * yarihm Quit: Leaving 1170622167 M * baldy is there a way to kill "ghost" vps? 1170622167 M * baldy CTX PROC VSZ RSS userTIME sysTIME UPTIME NAME 1170622167 M * baldy 0 58 51.7M 14.7M 1d13h16 35m13s86 2d02h36 root server 1170622167 M * baldy 50 6 17.5M 4.1M 0m00s00 0m00s00 1m50s20 1170622192 M * nebuchadnezzar well, 2.6.20 is stable now :-) 1170622446 M * Bertl nebuchadnezzar: but it will take a few days until we have a stable patch for that 1170622458 M * Bertl (too many changes which need new workarounds) 1170622600 M * DSpair daniel_hozac: Thanks for the information. That's exactly what I was thinking of while I was away writing. Sorry I didn't respond. My associate at work and I are writing a paper on Fault Tolerant WANs to submit for the Fall LinuxWorld. 1170622645 M * baldy reboot 1170622667 M * nebuchadnezzar thanks Bertl 1170622697 M * DSpair daniel_hozac: Do you think it would be adviseable to use the 'dummy0' rather than 'lo' interface? 1170622905 M * daniel_hozac DSpair: for? 1170623078 M * Bertl DSpair: dummy0 will look nicer inside the guest 1170623084 Q * baldy Read error: No route to host 1170623090 M * Bertl DSpair: you can even rename it to eth1 or so :) 1170623109 M * Bertl (which I would not suggest for lo :) 1170623196 M * nebuchadnezzar Well, about lo interface, I tryed the auto_lback but I'm not sure to understand what it does 1170623231 M * Bertl (isn't completely working yet) 1170623233 M * nebuchadnezzar I thought that it will take the first IP assign to the vserver and use it as a 127.0.0.1 1170623243 M * daniel_hozac no. 1170623260 M * daniel_hozac there's a separate address, by default 127.x.y.1, where x.y is the xid. 1170623278 M * daniel_hozac it's not settable from util-vserver yet, you'd need to use vcmd or similar to change it. 1170623285 M * daniel_hozac (if for some reason you don't like the default) 1170623785 M * Bertl okay, I'm off to bed for now ... maybe back later, maybe tomorrow ... have fun! 1170623794 N * Bertl Bertl_zZ 1170623801 M * daniel_hozac good night! 1170623806 M * nebuchadnezzar see you Bertl_zZ 1170624513 Q * kasper 1170625310 Q * mire Ping timeout: 480 seconds 1170625359 M * daniel_hozac ensc: do you think your changes to the sigbus test would also fix https://savannah.nongnu.org/bugs/?16375 ? 1170625440 Q * bonbons Quit: Leaving 1170625469 M * daniel_hozac ensc: or is that caused by something entirely different? 1170625513 J * Aiken ~james@ppp220-70.lns2.bne1.internode.on.net 1170625519 M * ensc dunno; but does not look so. My changes are fixing only the issue that large pagesizes (64Ki) did not trigger the SIGBUS 1170625537 M * daniel_hozac yeah, that's what i figured from the commits... 1170625551 Q * DSpair Quit: Chatzilla 0.9.77 [Firefox 2.0.0.1/0000000000] 1170625574 M * ensc it's just an extension of dwmw's changes; I do not use hardcoded buffer size but PAGESIZE 1170625599 M * daniel_hozac right. 1170625682 M * ensc the reported segfaults should not happen resp. the archs are segfaulting instead of getting the SIGBUS 1170625698 M * daniel_hozac that's what i was thinking too. 1170625734 M * daniel_hozac but alpha and s/390 succeed with the diet and glibc version, respectively. 1170625742 M * ensc nevertheless, the difference between dietlibc and -gnu tests is strange 1170625784 M * daniel_hozac Aiken: you're our alpha man, right? 1170625785 M * ensc strace output would be interesting for these cases 1170625789 M * Aiken yes 1170625800 M * daniel_hozac Aiken: could you run make check in your util-vserver source tree? 1170625845 M * Aiken give me a few minutes 1170626050 J * mire ~mire@128-167-222-85.adsl.verat.net 1170626123 J * joxjox ~jox@c83-252-43-201.bredband.comhem.se 1170626296 P * joxjox Konversation terminated! 1170626534 M * daniel_hozac ensc: could you enlighten me to what the sigbus test is actually testing? i mean, where do we rely on it? 1170626613 M * ensc it writes into a file, mmap() it, truncates it, and reads it --> reading after the truncation should trigger a SIGBUS 1170626638 M * ensc the test is used to check whether the code handling such an event is correct 1170626677 M * ensc (e.g. returning control to program which restores old files, instead of failing with a segfault and creating corrupted files) 1170626765 M * daniel_hozac ah, so the segfault is semi-expected, and indicates a buggy kernel? 1170626847 Q * dna Quit: Verlassend 1170626866 M * ensc there should not happen a segfault when my code, the libc and the kernel are correct 1170626925 M * ensc current problem was, that kernel makes some optimizatios so that the SIGBUS does not happen when reading after the truncation happens within a 2*PAGESIZE range 1170626972 M * ensc (never heard about; just brought up by dwmw2 because ppc64 has a huge PAGESIZE) 1170627067 M * daniel_hozac i don't think the huge pagesize is the default, IIRC that's a Fedora thing. 1170627077 M * daniel_hozac (though i think it's configurable even in vanilla) 1170627296 M * daniel_hozac (CONFIG_PPC_64K_PAGES, if anyone's interested) 1170627375 M * Aiken http://paste.linux-vserver.org/1057 1170627388 M * Aiken make check fails a couple of tests 1170627399 M * daniel_hozac interesting, you have both tests failing... 1170627462 M * daniel_hozac could you paste strace -fF lib_internal/testsuite/sigbus too? 1170627528 M * daniel_hozac could you also try applying http://svn.linux-vserver.org/projects/util-vserver/changeset/2484?format=diff&new=2484 and see if the problem persists? 1170627609 M * Aiken http://paste.linux-vserver.org/1058 1170627655 Q * meandtheshel1 Quit: Leaving. 1170628037 M * Aiken still fails 1170628121 M * ensc this looks like a problem while establishing the signal handler 1170628198 M * daniel_hozac lchown looks really strange. 1170628628 M * daniel_hozac AFAICT lchown should be sigsetjmp in lib_internal/unify-copy.c:copyMMap, no? 1170628685 M * daniel_hozac Aiken: how about an strace -fF lib_internal/testsuite/sigbus-gnu? 1170629156 M * Aiken http://paste.linux-vserver.org/1060 1170630381 M * daniel_hozac hmm, well... i have to admit that i don't see why. 1170630396 M * jjpacheco anyone here play with gfs2? 1170630532 M * daniel_hozac why do you ask? 1170630537 Q * FireEgl Ping timeout: 480 seconds 1170630554 M * jjpacheco i can't compile the 2.6.19.2 kernel with it 1170630562 M * jjpacheco and i want to know if im missing something 1170630621 M * daniel_hozac why not? 1170630629 M * daniel_hozac (paste.linux-vserver.org) 1170630650 M * jjpacheco i ran the compile last night 1170630659 M * jjpacheco that terminal info is gone 1170630682 M * jjpacheco just wondering if anyone has successfully compiled a kernel with gfs support 1170630697 M * daniel_hozac my kernel includes GFS2 as a module. 1170630706 M * jjpacheco mmmm 1170630711 M * jjpacheco i made my static 1170630720 M * jjpacheco do you think it would make a difference? 1170630744 M * daniel_hozac no idea. without the errors, it's just guessing... 1170630744 M * jjpacheco well lets see :) 1170630771 M * jjpacheco r u using a shared storage device with gfs? 1170630802 M * jjpacheco k 1170630806 M * jjpacheco im making 1170630816 M * jjpacheco GFS2 file system support M 1170630824 M * jjpacheco GFS2 "nolock" M 1170630832 M * jjpacheco and GFS2 DLM M 1170630859 M * daniel_hozac no. i just rebuild the Fedora kernels. 1170630874 M * jjpacheco oh 1170630883 M * jjpacheco so you've never really used it? 1170630890 M * daniel_hozac no. 1170630968 M * jjpacheco ic 1170631212 J * FireEgl ~FireEgl@adsl-147-81-6.bhm.bellsouth.net 1170631323 M * jjpacheco http://paste.linux-vserver.org/1061 1170631329 M * jjpacheco there u go daniel 1170631337 J * dna ~naucki@89-213-dsl.kielnet.net 1170631351 M * daniel_hozac hmm. 1170631373 M * daniel_hozac i take it you have # CONFIG_CONFIGFS_FS is not set in your .config? 1170631388 Q * ema Quit: leaving 1170631396 M * jjpacheco im just reading about this bug now 1170631419 M * daniel_hozac i take it it's already fixed in 2.6.20? 1170631429 M * jjpacheco i don't know 1170631449 Q * ensc Killed (NickServ (GHOST command used by ensc_)) 1170631459 J * ensc ~irc-ensc@p54B4F959.dip.t-dialin.net 1170631604 M * jjpacheco anyone know what's going on with hans reiser? 1170632013 M * cehteh he gets his trial sometime next 1170632034 M * jjpacheco http://en.wikipedia.org/wiki/Hans_Reiser 1170632036 M * jjpacheco damn 1170632048 M * jjpacheco he had to sell his company to pay for legal costs 1170632061 M * cehteh did he managed to sell namesys? 1170632063 M * jjpacheco woohoo 1170632065 M * jjpacheco compile worked 1170632067 M * jjpacheco :) 1170632075 M * jjpacheco thx daniel 1170632096 M * daniel_hozac for? 1170632126 M * jjpacheco gfs, etc 1170632151 M * daniel_hozac but, i didn't do anything. 1170632161 M * jjpacheco the configfs thingy 1170632402 M * daniel_hozac so what actually fixed it? 1170632506 M * daniel_hozac what kernel is that, btw? 1170632507 Q * tso Quit: BitchX-1.1-final -- just do it. 1170632531 M * daniel_hozac IMHO 2.6.19.2 is fine.