1153268865 N * Bertl_oO Bertl 1153268869 M * Bertl evening folks! 1153268909 M * daniel_hozac evening! 1153268912 M * doener hi Bertl 1153269437 M * Bertl somebody contacted me today (german magazine) regarding some virtualization article 1153269455 M * daniel_hozac cool! 1153269470 M * Bertl not sure they want to write something, or want me to write something 1153269492 M * doener hm, maybe (finally) iX? 1153269507 M * Bertl will get details tomorrow I guess, but if they need some article, I probably have to 'request' some help with that ... 1153269684 M * Wonka np 1153269689 M * Wonka <- muttersprachler 1153269748 M * doener dito ;) and Bertl is really good at "real german" *g* 1153269847 A * doener .oO( what was that Austrian word that one guy at LinuxTag wanted Bertl to say?... ) 1153269873 M * Wonka Oachkatzl? 1153269907 M * Wonka .oO( too many germans and stuff around... ) 1153270266 M * Bertl Wonka: des hoast oachkatzlschwoaf :) 1153270304 J * Karmek ~nichts@i53874944.versanet.de 1153270326 M * Bertl welcome Karmek! 1153271279 Q * Karmek Ping timeout: 480 seconds 1153272455 N * meandtheshell oachkatzle 1153272466 A * oachkatzle geht aufn bam aufe 1153272469 M * oachkatzle :-) 1153272567 A * oachkatzle can contribute to that mystical article as well ... 1153273333 M * Bertl excellent :) 1153273440 M * oachkatzle maybe it would be worth mentioning that Linux-VServer in FAI (FullyAutomaticInstallation) really rocks da house 1153273454 M * oachkatzle s/in/and/ 1153273495 M * oachkatzle sounds weird like I wrote it - I mean http://faiwiki.informatik.uni-koeln.de/index.php/Using_FAI_to_set_up_VServers 1153273539 M * Bertl yep, let's see _what_ they actually want from _us_ :) 1153273551 M * oachkatzle ok 1153273595 M * Bertl if I got the dates right, deadline is 31st this month 1153273623 M * Bertl so as soon as we know, we basically have to start writing something :) 1153273637 M * oachkatzle we're talking about an article not a paper - right? 1153273656 M * Bertl yes 1153273680 M * oachkatzle should be doable then 1153273761 N * oachkatzle meandtheshell 1153274032 M * Bertl okay, I'm off to bed for tonight ... have a good one everyone, and hopefully cya tomorrow! 1153274046 N * Bertl Bertl_zZ 1153274259 J * zob000 ~zob000@wsp05974758wss.cr.net.cable.rogers.com 1153274262 M * zob000 howdy 1153274284 M * zob000 i am googling the usage of vrrp or ucarp inside vserver guest .. any pointers ? 1153275551 P * Exomorph 1153275671 J * Aiken_ ~james@tooax8-013.dialup.optusnet.com.au 1153275987 Q * Aiken Ping timeout: 480 seconds 1153276007 Q * balbir Read error: Connection reset by peer 1153277139 J * balbir ~balbir@59.144.57.199 1153278391 J * Term_ ~PhAnATiC@201.152.93.213 1153280872 Q * balbir Read error: Connection reset by peer 1153280948 J * Tuhu ~vmware@200.103.132.43 1153281604 J * gerrit ~gerrit@wm402rot.66.ADSL.NetSurf.Net 1153281655 M * zob000 anyone can say what vservers/.defaults/run.rev/ is for ? can it be removed if the dir it points to was removed ? 1153281985 J * balbir ~balbir@203.101.54.29 1153283613 M * Skram i only have 11 content slides, its either i get way more technical or keep it at this 1153283782 Q * Greek0 Ping timeout: 480 seconds 1153284949 Q * Tuhu Quit: BitchX: the Cadillac of all clients 1153285172 J * Greek0 ~greek0@85.255.145.201 1153285464 Q * Term_ Quit: (-(PS)-) [v5.0.r02] http://www.kalendas.net 1153287677 Q * meandtheshell Quit: bye bye ... 1153290864 J * dna ~naucki@dialer-165-12.kielnet.net 1153292917 Q * Aiken_ Ping timeout: 480 seconds 1153293605 J * ||Cobra|| ~cob@146.50.22.204 1153294846 M * meebey morning 1153295358 M * phedny morning 1153295393 Q * m4z Quit: http://www.catb.org/~esr/faqs/smart-questions.html 1153295683 J * m4z m4z@bastard-operator.from-hell.net 1153295688 J * Godzilla ~Godzilla@202.55.153.172 1153296417 Q * Godzilla Ping timeout: 480 seconds 1153296590 J * renihs ~penguin@83-65-34-34.arsenal.xdsl-line.inode.at 1153297178 Q * mcp Read error: Connection reset by peer 1153297186 J * mcp ~hightower@wolk-project.de 1153298097 J * wam ~wam@proxy2.msh.de 1153298118 M * wam Hi, does "reboot -f" shutdown all services or does it just unplug the vserver and start it again? 1153298199 M * renihs it shuts the vserver i guess...thus taking all services with it 1153298217 M * wam so I need /dev/initctl for a proper shutdown? 1153298657 M * wam waah - no. I won't need /dev/initctl. I only want to be sure that the vserver is cycled in a clean way when calling "reboot -f". 1153299058 J * yarihm ~yarihm@84-74-17-70.dclient.hispeed.ch 1153299069 N * otaku42_away otaku42 1153299314 J * pisco ~pampel@p5087BA22.dip0.t-ipconnect.de 1153299567 Q * shedi Quit: Leaving 1153299574 Q * kir Ping timeout: 480 seconds 1153299700 J * pisc1 ~pampel@p50879930.dip0.t-ipconnect.de 1153299783 M * wam How can I "reboot" a vserver from inside the vserver? I'd like to have all init-scripts executed with the "stop"-Option. Just like I'd do "init 6". 1153300037 Q * pisco Ping timeout: 480 seconds 1153300084 M * Loki|muh wam: have you tried to place a custom skript in /etc/rc6.d which e.g. echos something to /root/shutdown? 1153300147 M * wam Loki|muh: yes, I edited K11cron inside the stop-case to write something to /root/t 1153300298 M * wam Loki|muh: and now I tested it with two other Scripts: /etc/rc6.d/S01t and /etc/rc6.d/K01t, both echoing into a file. But nothing. 1153300352 M * Loki|muh wam: for me /etc/rc6.d/S01echoshit is working 1153300358 M * Loki|muh and places a file in my homedir 1153300372 M * wam a S-script is called when shutting down? 1153300372 M * Loki|muh did you make the scripts executable? ;) 1153300376 M * wam yes ;) 1153300398 M * Loki|muh seems so, I just tried it, dunno how it really works 1153300643 M * wam could it be, that rebootmgr only works correctly when the vservers are stored below /var/lib/vservers? - At least in the debian-sid util-vserver package? 1153301191 J * schimmi ~sts@aquila.tcs.ifi.lmu.de 1153301219 M * wam ll 1153301222 M * wam oops 1153301287 J * oDn ~odn@pcam.net1.nerim.net 1153301369 M * oDn hi folks 1153301719 M * renihs welcome oDn 1153301735 M * oDn cheers renihs 1153301767 Q * gerrit Ping timeout: 480 seconds 1153301914 Q * wam Quit: Verlassend 1153302536 J * lilalinux ~plasma@80.69.35.186 1153302753 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1153302789 Q * yarihm Read error: Connection reset by peer 1153302857 J * yarihm ~yarihm@84-74-17-70.dclient.hispeed.ch 1153303578 J * shedi ~siggi@213.190.108.50 1153303629 J * gerrit ~gerrit@wm402rot.66.ADSL.NetSurf.Net 1153304114 Q * weasel Ping timeout: 480 seconds 1153304147 J * weasel weasel@asteria.debian.or.at 1153304154 J * cdrx ~legoater@wm402rot.66.ADSL.NetSurf.Net 1153304947 J * turkei ~Kacak-4-@85.102.255.2 1153304947 Q * yarihm Read error: Connection reset by peer 1153305164 J * yarihm ~yarihm@84-74-17-70.dclient.hispeed.ch 1153305408 J * gypsymauro ~Io@84.18.151.77 1153305411 M * gypsymauro hello 1153305429 Q * turkei Ping timeout: 480 seconds 1153305430 M * gypsymauro someone experienced a succesfull installation of vserver on dapper? 1153306270 Q * yarihm Read error: Connection reset by peer 1153307367 Q * schimmi Ping timeout: 480 seconds 1153307453 N * Ben_zZz Ben__ 1153307698 M * daniel_hozac have you read the vserver on ubuntu page? 1153307908 M * derjohn does anyone have experience with nforce sata chipsets on linux? 1153308123 Q * virtuoso Ping timeout: 480 seconds 1153308249 M * daniel_hozac i'm using it. 1153308721 M * doener dito 1153308791 J * dna_ ~naucki@dialer-168-232.kielnet.net 1153308827 M * derjohn daniel_hozac, i'm just thinking about changeing from via ktXXX to something nforce based ... in former times nforce was not very well supported by linux. so: could you (hey doener) recommend it? what read performance do you get (hdparm -t ..) ? did you every try using 4-port? 1153308855 M * daniel_hozac i'm using a 3-drive RAID right now ;) 1153308874 M * derjohn with what type of nforce ? 1153308881 M * daniel_hozac nforce 430. 1153308913 M * derjohn so, you get some non-sucking performance (I would say > 50 MB/sec cont read ....) ? 1153308924 M * FaUl derjohn: i have nforce-3 chipset with sata onboard, but no sata-harddisk to test 1153308926 M * daniel_hozac Timing cached reads: 3236 MB in 2.00 seconds = 1618.39 MB/sec 1153308927 M * daniel_hozac Timing buffered disk reads: 174 MB in 3.03 seconds = 57.38 MB/sec 1153308984 M * derjohn looks promising, I think i will make a test setup with that kind of hw for our next generation of hosts ;) 1153308986 M * doener Nforce4 sth. here, Tyan Tomcat K8E using Linux Software RAID1 (the stupid HostRAID isn't worth the trouble) 1153309011 M * daniel_hozac yeah, i'm using software RAID5. 1153309020 M * derjohn doener, of course linux-softraid! All other is shit, even 3ware (didnt try acreca yet ...) 1153309032 M * doener hm, hdparm values are pretty useless except for testing that dma is enabled on PATA controllers... 1153309055 J * s0undt3c1 ~s0undt3ch@bl7-249-193.dsl.telepac.pt 1153309091 M * doener and what you don't get is NCQ as NVidia does not hand-out any information about their chipsets... (at least that's the latest I know on that topic) 1153309118 Q * dna Read error: Operation timed out 1153309162 M * derjohn doener, is that so? i have a case wit some amd chipset and 4-port sata, where disk one has 25 meg/sec, disk two 55 meg/sec, disk 3 and 4 about 10 meg /sec in a non-concurrent (!) setup. I asume ACPI/IRQ probelems but was not able to find out. 1153309220 M * doener same disks? 1153309221 M * derjohn doener, about NCQ: without sata2 it's not worth trying. many disk you have to downjumer to sata1 at the time anyway. 1153309239 M * derjohn doener, 100% same disks. 300 GB maxtor. 1153309267 M * FaUl Bertl_zZ: good news, the las t 1153309276 N * Bertl_zZ Bertl 1153309277 M * derjohn IME Samsung HDDs run fine. 1153309282 M * Bertl morning folks! 1153309296 M * FaUl Bertl_zZ: vserver just migrated to our new server, so the sun can now move to vienna ;-) 1153309313 M * doener morning Bertl 1153309315 M * FaUl morning bertl 1153309318 M * FaUl thats timing ;-) 1153309324 M * Bertl derjohn: I have to disagree here, the 7200.10 baracuda rocks :) 1153309329 J * virtuoso ~s0t0na@80.253.205.251 1153309341 M * Bertl FaUl: perfect! 1153309367 M * Bertl welcome virtuoso! 1153309370 M * derjohn seagate ok ok, too, but maxtor sucks. ibm/hitachi i dont trust since click-of-death. 1153309420 M * doener derjohn: my maxtor disks are running like a charm... 1153309424 M * FaUl derjohn: well - 3ware-raid rocks if you have 9xxx-controllers ;-) 1153309427 M * Loki|muh gypsymauro: yes, use the debian util-vserver package and pay attention to http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=374188 1153309436 M * derjohn FaUl, nope. want to buy some form us? :) 1153309446 M * FaUl derjohn: which one? 1153309457 M * derjohn FaUl, besides that the cli/tools of 3ware is not opensource. 1153309487 Q * s0undt3ch Ping timeout: 480 seconds 1153309488 N * s0undt3c1 s0undt3ch 1153309513 A * FaUl uses hitachi-disks without any trouble - since click of death they have invested much into quality-management 1153309631 M * derjohn FaUl, Buzzword: quality-management! Bingo ;) 1153309661 M * FaUl derjohn: you need at least 5 buzzwoards in a row ;-P 1153309757 M * FaUl a friend of mine played buzzwoard-bingo in a recent company-meating - everybody looked confused when he suddenly shouted "BINGO!!" while some marketing-droid presentated a new product ;-) 1153309857 M * derjohn FaUl, argh, i have only two: invested and quality-management ... no bingo yet ;) 1153309900 M * FaUl yes 1153309951 M * FaUl maybe we should do another kick-of-meeting to do design-studies to get a cooperate identity in our bullshit-bingo-business ;-) 1153309964 Q * virtuoso Remote host closed the connection 1153310036 M * Bertl okay, will be back later ... have fun! 1153310045 M * FaUl Bertl: we'll have 1153310049 M * FaUl its the warmth ;-) 1153310078 M * Bertl ah, I almost forgot, anybody interested in the secret mystic german article (*bingo?*) please drop me a note :) 1153310093 N * Bertl Bertl_oO 1153310166 M * weeble As a complete off-topic question - does anyone know how to hardcode IPv6 addresses in Solaris 10? I know how to make it pick up an address automatically, but I want to specify one 1153310189 M * mnemoc solawhat? 1153310206 M * FaUl mnemoc: slowlartus 1153310219 M * weeble mnemoc, It's this really nasty old styled OS that just works insanely weirdly. (From my Leenooks perspective) 1153310221 M * weeble :) 1153310236 M * mnemoc *G* 1153310295 M * weeble I think it's something to do with /etc/inet/ndpd.conf, but that's some file for a daemon that does router discovery. It's all just wrong. Bah 1153310835 M * weeble Aaah, of course. It's so obvious. Just add addif 2001::2 up to /etc/hostname6.bge1. /me shakes head. 1153310963 M * mnemoc *ROTFL* 1153311293 J * virtuoso ~s0t0na@80.253.205.251 1153311296 M * weeble But that only adds a subinterface with the address on. (bge1:1) Aaah well. 1153311310 M * FaUl weeble: don't care about that ;-) 1153311333 M * FaUl weeble: there are enough adresses to waste some for stateless-autoconfig, anyway 1153311342 M * weeble FaUl, Yeah, I think it doesn't matter. But why an interface can't have a v4 and v6 address on the main interface..... 1153311369 M * FaUl weeble: why not? 1153311410 M * weeble Because the only way I can get it to work is to use "addif 2001::1/64 up" which creates a new subif, and assigns the v6 address to that. 1153311486 M * weeble My next task: The default v6 route. Anyone knows where that lives? /etc/defaultrouter6 ? 1153311578 M * FaUl this should go with stateless-autoconfig, anyway 1153311587 M * FaUl should work 1153312198 J * coocoon ~coocoon@p54A0674B.dip.t-dialin.net 1153312387 M * gypsymauro Loki|muh: not on debian I know it works and I use it, but on ubuntu dapper 1153312538 M * daniel_hozac gypsymauro: you're aware of the Ubuntu<->Debian relationship, right? 1153312702 M * gypsymauro daniel_hozac: of course I'm debian user by about 10 years:) 1153312724 M * gypsymauro the problem is that I can't patch ubuntu kernel with vserver 1153312751 J * zagor ~andrea@service.cab.unipd.it 1153312758 M * daniel_hozac util-vserver is the userspace part. 1153312829 M * gypsymauro isnt' a problem of util-vserver is a problem of applying the vserver patch to the ubuntu kernel 1153312829 M * daniel_hozac and i thought there was someone doing Ubuntu kernels. 1153312845 M * gypsymauro yes but is broken 1153312854 M * gypsymauro http://ubuntu.uni-klu.ac.at/ubuntu.uniklu/ 1153312869 M * daniel_hozac broken in what way? 1153312869 M * gypsymauro I can boot but when I login in gdm it freezes 1153312880 M * gypsymauro instead the original kernel works well 1153312918 M * gypsymauro and using the kernel patch of ubuntu too 1153312919 M * gypsymauro https://launchpad.net/distros/ubuntu/+source/kernel-patch-vserver/+bugs 1153312926 M * gypsymauro same bug here 1153312926 M * daniel_hozac have you tried using a vanilla kernel instead? 1153313110 M * gypsymauro I'm trying that right now 1153313126 M * gypsymauro so sad:) 1153313477 Q * cdrx Ping timeout: 480 seconds 1153313910 J * wam ~wam@proxy2.msh.de 1153314165 M * Loki|muh i thought of a problem with ubuntu, not with the kernel, which is for me an independent thing ;) 1153314177 A * Loki|muh uses vanilla kernels for the hosts 1153314193 M * eyck that's not very #vserver-ish... 1153314232 Q * oDn Ping timeout: 480 seconds 1153314291 M * wam do I need to give the CYP_SYS_BOOT capability to my vservers if they should be able to reboot cleanly? 1153314538 M * Hollow Bertl_oO: i could contribute to this mysterious article too 1153314593 M * gypsymauro Loki|muh: well I think ubuntu ships sources and binary kernels with a lot of patch applied.. 1153314829 M * doener wam: no. use either "reboot -f" (which will trigger the vshelper which will stop and start the vserver) or use plain init-style, so that you have an init process that "reboot" can talk to, the vserver will switch to runlevel 6 itself then, finally call "reboot -f" which then also triggers the vshelper 1153314853 M * doener the reboot should be clean in both cases 1153314877 M * wam doener: How does reboot -f shut the services down? Just by killing them or by the init-style scripts? 1153314937 M * doener reboot -f does not kill them at all. on a regular host it triggers a hard, unclean shutdown/reboot 1153314965 M * doener in a vserver, you're not allowed to do that, instead, the vshelper is started which does "vserver foo restart" 1153314986 M * wam doener: I'm just a little bit confused, because when I call "reboot -f" inside my vserver, it looks like no init-scripts would be executed. 1153314999 M * doener and the restart does "stop" and "start", ie. switching to runlevel 0 and back to 3 (well, kind of) 1153315028 M * doener wam: because the init-scripts are called from the outside of your session, you just don't see their output 1153315035 M * wam doener: I tried to add K-scripts to /etc/rc0.d which should have written something to a file. But the file was never created. 1153315082 M * doener hm, do you have a matching S-script in the previous runlevel? IIRC K-scripts get called only for those 1153315096 J * meandtheshell ~markus@85-124-174-200.dynamic.xdsl-line.inode.at 1153315097 M * wam ah - I'll try. Thanks. 1153315106 M * doener I'd use an S-script instead 1153315141 M * doener that will be started _unless_ there is a matching S-script in the previous runlevel which is easier/cleaner to get ;) 1153315148 J * oDn ~odn@pcam.net1.nerim.net 1153315207 M * wam doener: just for completenes: Does init work this way too or is this vserver-specific? 1153315241 M * doener that's how init works (/etc/init.d/rc actually) works 1153315274 M * wam by comparing NAMES? Hmm. That somehow doesn't sound like a good concept to me. But it seems to work ;) 1153315298 M * doener it's useless to Kill something that isn't running, and it might (for stupid programs) cause havoc to start sth. that is already running 1153315339 M * doener wam: well, you usually have your scripts in /etc/init.d and use your distros tools (chkconfig, update-rc.d,...) to manage the symlinks in the different runlevel directories 1153315468 M * wam doener: now I tested it. I wrote /etc/init.d/test, which writes "test" >> /root/test. Then I symlinked it to /etc/rc3.d/S01test and /etc/rc0.d/K01test. After a reboot -f, only one "test" is inside /root/test. So I guess, the K-Script doesn't get called. 1153315499 M * Hollow reboot is rc6 1153315505 M * wam I'll try. 1153315532 M * doener Hollow: does the vshelper actually use runlevel 6? I thought restart just calls stop and both use runlevel 0 1153315542 M * wam Hollow: same behavour. 1153315552 M * wam Hollow: I now have /etc/rc6.d/K01test 1153315577 M * wam It looks more to me like reboot -f inside the guest bahaves like it would do on a real host. 1153315581 M * Hollow doener: well, i just said it in general, but you're probably right 1153315592 M * Hollow though, if this is an initless system who knows if it ever gets called actually 1153315612 M * Hollow ah 1153315613 M * Hollow i know 1153315618 M * Hollow you use reboot -f :) 1153315627 M * wam Hollow: yep 1153315635 M * Hollow this will simply kill everything, no? 1153315646 M * wam Hollow: reboot (without -f) just says: init: /dev/initctl: No such file or directory 1153315649 M * doener Hollow: does that change anything? should just trigger the vshelper 1153315687 M * Hollow i thought reboot -f kills 1153315699 M * Hollow (before) 1153315717 M * Hollow ah.. only with VXF_REBOOT_KILL 1153315767 M * wam so do I have to use another init-style to have "good" reboots? 1153315777 M * Hollow obsoleting initless guests would be so awesome 1153315831 M * doener well, write a zero-resource-usage init ;) 1153315841 M * Hollow waldi: honestly, i don't know.. reboot never worked with vanilla util-vserver for me :p 1153315900 M * Hollow you'll need legacy support in the kernel afaicr 1153315958 M * wam Could someone please point me to a doc where I can find out how to change my init-style to something with a real init-task? 1153315965 M * wam I'd like to try this. 1153315966 M * Hollow plain? 1153315969 M * wam maybe. 1153315981 M * wam Hollow: plain == what a normal host does? 1153315988 M * Hollow yep... 1153316004 M * wam Hollow: so plain ;) Is the "style"-file in apps/init all I have to adjust? 1153316017 M * wam echo "plain" > ....apps/init/style ? 1153316022 M * Hollow should be ok.. stop the guest before 1153316025 M * wam ok 1153316205 M * wam hmm. now, "reboot" or "init 6" change to runlevel 6. But it looks like they don't call the vshelper to actually cycle the vserver... 1153316352 M * Hollow do you have legacy enabled in the kernel? 1153316364 M * wam yes 1153316378 M * Hollow hm.. 1153316400 M * Hollow but init does shutdown all services? 1153316412 M * wam Hollow: you're talking about "Enable Legacy Kernel API", yes? This is enabled. 1153316417 M * wam all services are shutdown. 1153316419 M * Hollow yup 1153316423 M * wam Only the reboot is missing. 1153316467 M * wam when I enter (vserver ... enter) and call "reboot -f", it restarts. 1153316487 M * wam Maybe I should call this as the last K-script in rc6.d ? ;) 1153316506 M * Hollow well, init should call sys_reboot 1153316515 M * Hollow iirc 1153316689 M * Hollow hm.. gentoo calls reboot -f in the reboot scripts 1153316734 M * wam would a strace help? 1153316820 M * Hollow hm, it looks like init does _not_ call sys_reboot, at least not after a quick look at the code 1153316886 M * Hollow you can try to add it to the last init script, but if init does not call it, your distros init scripts should call it anyway 1153316900 M * Hollow but maybe those got removed during build, dunno.. 1153317151 M * wam waah - my vserver now doesn't shut down any more. init is unkillable. How can I force-kill it from the host? 1153317364 M * wam forget it. thanks for your help, doener and Hollow! 1153317413 M * doener yw 1153317426 M * Hollow mkay.. :) 1153317491 M * doener Hollow: sysv init usually has a K script in rc{0,6}.d that calls {halt,reboot} -d -f -i 1153317550 M * daniel_hozac it's an S script here. 1153317562 M * doener ehr, right, K wouldn't make any sense 1153317614 M * doener got confused by a function in there that is called do_stop ;) 1153317617 M * Hollow ok, that at least proves my assumption after the glance at inits source 1153317644 Q * coocoon Remote host closed the connection 1153317671 M * daniel_hozac has anyone used kexec to speed up reboots? 1153317754 M * daniel_hozac i've got a few hosts that take several minutes to set up the hardware... being able to speed it up would be very nice. 1153317924 M * Hollow daniel_hozac: gentoo reboot.sh uses reboot -k by default.. dunno how/if/when it works, though :) 1153317951 M * gypsymauro trying the vanilla kernel..reboot:) 1153317952 Q * gypsymauro Quit: leaving 1153318024 M * cehteh daniel_hozac: afaik not all hardware works fine with it .. just give it a try ... but dont try remotely :) 1153318364 J * pisco ~pampel@p50879930.dip0.t-ipconnect.de 1153319387 M * ray6 greetings from flight lh410 to NYC :) 1153319541 M * wam ray6: wlan or mobile? 1153319570 M * ray6 wlan 1153319582 M * ray6 connexion by boeing 1153319681 M * weeble ray6, You're up there now? 1153319683 M * weeble That's cool 1153319685 M * wam ray6: you're coming over your vpn, aren't you? 1153319704 M * ray6 wam: over ssh actually 1153319717 M * ray6 weeble: yeah, just left great britain behind... 1153319722 M * weeble "wam ray6: you're coming over your vpn, aren't you?" Well, perhaps he's not *that* excited? 1153319726 M * weeble :) 1153319752 A * ray6 isn't also, already used that two times before :) 1153319752 M * wam ray6: what's the airplane's IP? ;) 1153319776 M * ray6 wam: so you can flood my plane huh? :) 1153319783 M * weeble ping lh410.aeroplanes.sky, of course, duh! 1153319784 M * Hollow heh 1153319787 M * ray6 the plane announces it's own /24 over BGP 1153319793 M * wam ray6: I'm just curious if I can check when it landed or so. 1153319806 M * wam ray6: or if ping-times change with satellites. 1153319817 M * ray6 when I switch the continent (about two hours from now) the /24 will be reannounced to the US AS 1153319840 M * weeble Yeah, ray6, on your home box, can you see what external IP it uses? 1153319846 M * ray6 wam: you can check when it changes continent - for every single plane using this all the time 1153319871 M * tdjb i've got an odd situation: after a vserver guest has been running for a while the main host interface stops responding to traffic. the guest interface usually stays up though 1153319873 M * ray6 weeble: I have BGP at home, I get the announcement pushed :) 1153319891 M * Hollow tdjb: the guest has no interface.. all network happens on the host 1153319935 M * tdjb so i've got about 10 aliases (eth0:1, eth0:2, etc) and each guest gets it's own alias, could that be the issue? 1153319947 M * eyck you've got your own private AS at home? 1153319963 M * ray6 weeble: sure I can, it uses the .2 in it's /24 :) I guess thats all they use of it 1153320009 M * Hollow tdjb: sorry, don't know 1153320019 M * ray6 eyck: "at home" is more like "in my office" but yes... 1153320031 M * wam tdjb: on my servers, this works. 1153320045 M * tdjb hollow: no problem, thanks anyways though 1153320052 M * weeble ray6, So what's the /24 they are announcing? (I just get a default route pushed via BGP here, rather than the whole tables) 1153320064 M * ray6 wam: ping time to germany will increase about 100ms when I change satelite to US 1153320087 M * wam ray6: what's your current latency? 1153320108 M * ray6 weeble: about 600, wait... 1153320127 M * ray6 rtt min/avg/max/mdev = 586.975/656.453/990.887/116.208 ms 1153320161 M * wam ray6: ugly for shells, but ok for email and surfing... 1153320198 M * sid3windr :) 1153320215 M * ray6 wam: I already used it for H323 video telephony! 1153320236 M * wam ray6: wanna have fun? teardrop the fbi now and see, if they are fast enough to catch you at the airport ;) 1153320267 M * sid3windr http://www.sid3windr.be/archives/170/tracking-planes-via-bgp ;) 1153320279 M * ray6 wam: ah, I guess they'll do anyway due to my lockpicking stuff and so 1153320297 M * ray6 sid: oh, so I'm not the only one with that idea :) do they have a list there? 1153320305 M * weeble sid3windr, I want to see pics of Diana! 1153320309 M * weeble Where are they? 1153320314 M * wam ray6: you don't have this in your hand luggage, do you? 1153320333 M * sid3windr nowhere :p 1153320344 M * ray6 wam: no, checked in of course 1153320353 M * weeble sid3windr, As an aside, don't you get a bit worried about telling last.fm about all the music you own? 1153320359 M * ray6 ah, lunchtime 1153320364 M * sid3windr ray6: no list, but the /16 is mentioned 1153320368 M * sid3windr so you can track 256 of the .2 ip's 1153320372 M * sid3windr weeble: dunno 1153320373 M * sid3windr not really 1153320382 M * sid3windr but I'm thinking of disabling it again 1153320387 M * sid3windr it slows down the loading of my page sometimes 1153320405 M * ray6 I guess the subnet is assigned per plane, will have to check if I had the same IP on this flight 9 months ago :) 1153320445 M * sid3windr yea 1153320449 M * sid3windr a /24 per plane 1153320452 M * sid3windr waste waste 1153320486 M * ray6 and one bgp announcementchange per flight 1153320487 M * sid3windr origin as 30533 1153320492 M * ray6 a few hundred per day 1153320505 M * sid3windr hehe 1153320520 M * ray6 I'm in AS 29257 currently 1153320532 J * gypsymauro ~Io@84.18.151.77 1153320536 M * gypsymauro it works:) 1153320573 M * gypsymauro well, I heard it was possible to share an image of a distro for many guests os (for example a standard sarge) is it true? I didn't found doc about that 1153320583 M * weeble Don't quite a few networks filter routes <= /24 ? 1153320602 M * weeble To cut down on the number of routes 1153320631 M * ray6 weeble: depending on the network range usually 1153320644 M * ray6 weeble: a /24 from a /19 PA aggregate is usually filtered 1153320652 M * ray6 a single /24 shouldn't 1153320659 M * weeble Aaah, OK 1153320767 M * weeble I wonder if the BGP advertisements are the best way to go? It would be easier, but increase latency to just tunnel the plane traffic out to the place it came from, say London 1153320795 M * wam weeble: read http://www.renesys.com/blog/2006/04/tracking_plane_flight_on_inter.shtml and you know why not. 1153320827 M * ray6 weeble: the announcements arent that bad, there are thousands of updates every day so why not grany boeing a few 1153320841 M * wam weeble: the network wouldn't be usabable due to the latency. 1153321103 Q * gypsymauro Quit: leaving 1153321786 Q * lilo2 Remote host closed the connection 1153321938 J * Term_ ~PhAnATiC@201.138.29.43 1153321991 J * lilo2 hiddenserv@tor.noreply.org 1153322090 J * stefani ~stefani@tsipoor.banerian.org 1153322464 Q * m4z Ping timeout: 480 seconds 1153322527 P * wam Verlassend 1153322934 Q * ||Cobra|| Remote host closed the connection 1153323031 Q * renihs Remote host closed the connection 1153323137 Q * Term_ Quit: (-(PS)-) [v5.0.r02] http://www.kalendas.net 1153323483 J * m4z m4z@bastard-operator.from-hell.net 1153323513 Q * harry Server closed connection 1153323527 J * harry ~harry@d54C2508C.access.telenet.be 1153324058 P * pisco 1153325229 Q * mire Quit: Leaving 1153325484 J * bonbons ~bonbons@83.222.36.236 1153325699 J * alexx ~alexx@proxy.ikse.net 1153325835 J * kir_home ~kir@toronto-HSE-ppp4218833.sympatico.ca 1153325963 Q * zagor Quit: I'm gonna save Rose Tyler from the middle of the Dalek fleet, and then I'm gonna save the Earth 1153326678 J * Viper0482 ~Viper0482@p549771AA.dip.t-dialin.net 1153326792 Q * oDn Read error: Connection reset by peer 1153327151 N * baggins_ baggins 1153328098 Q * tokkee Server closed connection 1153328099 J * tokkee tokkee@casella.verplant.org 1153328929 Q * kir_home Ping timeout: 480 seconds 1153329634 J * fogoh ~laurent@l2v21-1-88-161-216-139.fbx.proxad.net 1153329638 M * fogoh hello 1153329692 M * daniel_hozac hi 1153329716 M * ray6 hi daniel 1153329717 M * fogoh I've a problem when lauching an archlinux guest, the message says that init-script failed. 1153329721 Q * ray6 Server closed connection 1153329745 M * fogoh Which script is executed when a vserver is started ? 1153329781 M * daniel_hozac /etc/{init,rc}.d/rc, if you're using the sysv initstyle. 1153329798 M * fogoh ok thanks 1153329854 J * ray6 ~ray@vh5.gcsc2.ray.net 1153329864 M * fogoh I don't have any of them :) 1153329899 M * fogoh I have /etc/rc.sysinit, could it be possible to use thise script to start a vserver ? 1153329988 M * daniel_hozac probably not the same. 1153330003 M * daniel_hozac you might want to use a plain initstyle instead, to closer imitate a real system. 1153331695 Q * lilo2 Quit: brb 1153331742 J * lilo2 hiddenserv@tor.noreply.org 1153332118 J * Karmek ~nichts@i5387406E.versanet.de 1153332126 M * Karmek hello there 1153332196 M * Karmek I am using private IPs on a test Vserver ans have some difficulties with connections. What do i have to change in order to be able to grant my vserver guest internet connections 1153332246 M * cehteh do you have a firewall set up? 1153332254 M * Karmek No 1153332288 M * cehteh set the proper routing? 1153332295 M * Karmek i use NAT to map the needed ports on an internal IP like 192.168.x.x 1153332317 M * mnemoc oh, the colorful messages are back 1153332321 M * Karmek i guess i need some kind of gateway? 1153332336 M * cehteh you speak of incoming or outgoing connections? 1153332360 M * Karmek outgoing. I tried a ping inside my vserver to an external ip and it did not work 1153332399 M * cehteh dont ping .. make tcp conencts 1153332416 M * cehteh lynx www.heise.de or such 1153332446 M * Karmek lynx is not installed yet it is just a base installation. apt-setup can't reach the mirrors :( 1153332507 M * cehteh is resolv.conf correct? 1153332553 M * Karmek i added my providers nameserver there 1153332565 Q * doener Server closed connection 1153332573 J * doener ~doener@host.magicwars.de 1153332588 M * cehteh then something with your nat setup might be borked 1153332624 M * Karmek do i need to add my host system as gateway for the vserver? 1153332628 M * cehteh do you use masq or real nat? 1153332635 M * daniel_hozac there is no such thing as a gateway for a guest. 1153332643 M * daniel_hozac all networking happens on the host. 1153332647 M * cehteh niether for nat 1153332668 M * cehteh you just need to be sure the packets flow correctly .. in both directions 1153332672 M * Karmek not sure i asked here how to map port 2022 to port 22 on another ip 1153332719 M * daniel_hozac that says nothing for outgoing connections. 1153332794 M * Karmek i have nothing configured for outgoing connections 1153332899 M * cehteh iptable -t nat -A PREROUTING -p tcp -i ppp0 -m state --state NEW --destination-port 80 -j DNAT --to-dest 10.20.80.30:80 1153332899 M * cehteh iptable -t nat -A PREROUTING -p tcp --destination $PPPIP -m state --state NEW -j DNAT --to-dest 10.20.80.30 1153332899 M * cehteh iptable -A INPUT -p tcp -i ppp0 --destination-port 80 -m state --state NEW -j ACCEPT 1153332922 M * cehteh .. thats for example how my www vserver accepts incoming http 1153332943 M * Karmek thank you, maybe i can use it for me 1153332951 M * cehteh just an example ... 1153333032 M * Karmek but whats about outgoing traffic? 1153333054 M * cehteh well ... for new server i would suggetst to use firehol for setting up the firewall instead hand-written scripts 1153333116 M * cehteh iptable -t nat -A POSTROUTING -o ppp0 -s 10.0.0.0/8 -j SNAT --to-source $PPPIP ... hopefully that handles the outgoing stuff :) 1153333137 M * cehteh .. my script is very old and rusty .. i think i didnt looked at it since 3 years :) 1153333145 M * Karmek 10.0.0.0 is the internal ip then? 1153333280 M * cehteh internal net 1153333285 J * kir_home ~kir@toronto-HSE-ppp4218829.sympatico.ca 1153333320 A * cehteh likes the 10.network.machine.interface notation .... 1153333327 M * Karmek hmmmmm he does not like --to-source 1153333339 M * cehteh SNAT 1153333353 M * doener micah: around? 1153333411 M * micah doener: I am, although my network is really bad so typing this is taking forever 1153333474 M * Karmek i've tried to ping my host system, that works 1153333491 M * Karmek but my host does not forward it, as it seems 1153333503 M * doener just wanted to note that the bootclean stuff is called in rcS which is not executed for vservers with sysv init style, so the bug reporter is not that wrong, it's probably util-vserver that tries to clean up /var/run because of that (not running rcS scripts) 1153333525 M * daniel_hozac heh, i just said that in an email :) 1153333529 M * doener I have some email problem atm, thus I cannot reply on the list 1153333534 M * doener ah good := 1153333538 M * doener s/=/)/ 1153333564 M * Karmek cehteh can you give me a further hint? I can't handle it :( 1153333643 M * cehteh Karmek: look at firehol.sf.net 1153333656 M * cehteh i really recommend that .. 1153333757 M * Karmek looks interesting but i can't install it at the moment. I am a bit in time pressure because i'll go on holiday in half an hour =) 1153333899 M * cehteh well .. .little late ... but i would say apt-get install firehol and then use that is easier than debugging your iptable script bugs 1153333941 M * cehteh and maybe you delay your holiday for some hours :P 1153333949 M * Karmek i hoped there would be an easy command that somebody may give me, so that my Host forwards the requests to the internet 1153333994 M * cehteh well i've shown you ... but you have obliviously some misconfiguration and need to find out whats wrong there 1153334066 M * Karmek seems so. I get the message that --to-source is an unknown argument 1153334073 M * Karmek maybe its part of your script? 1153334079 M * daniel_hozac no. 1153334094 M * daniel_hozac try --to instead, though they're supposed to be synonyms. 1153334111 M * Karmek unknown as well 1153334162 M * Karmek hmmmmm really seems to be a missconfiguration 1153334181 Q * pisc1 Quit: Download Gaim: http://gaim.sourceforge.net/ 1153334194 M * Karmek -to says i have to use --to-source and --to-source says unknown arg 1153334222 J * pisco ~pampel@p50879930.dip0.t-ipconnect.de 1153334232 Q * kir_home Ping timeout: 480 seconds 1153334316 M * Karmek are there other possibilitys? 1153334357 M * cehteh others than fixing the bug? :) 1153334389 M * Karmek okay seems not to be the case =) 1153334413 M * cehteh .. well .. firehol, it acts like a expert system with knowledge about iptable firewall .. you make a very few statement and it does everything right 1153334470 M * romke Karmek: still colorfull? 1153334476 M * cehteh hehe 1153334486 M * cehteh has oftc a +c mode? 1153334490 M * Karmek yes, but i'll remove it after my hollidays. I Promise 1153334505 M * romke ;) 1153334597 M * Karmek is the firehol configuration long during? Installed it for a try but yet i have No clue how to use it :D 1153334626 Q * shedi Quit: Leaving 1153334709 M * Karmek well anyway i need to go offline :/ Trying it next time 1153334710 M * Karmek Cya 1153334778 M * cehteh try firehol then 1153334784 Q * Karmek Quit: Da war "Nichts" :D | [14:45] mu� die jungfrau gehaun werden? | dd bekomm ich die jungfrau biiiitttteee 1153336042 J * juggo ~lemur@h-68-166-181-4.sttnwaho.covad.net 1153336078 M * juggo hey so I was here talking about random server lockups on vserver machines but I Was unsure of whether it might be hardware related or not 1153336102 M * juggo well, it's not hardware related as the problem has stayed despite swapping almost every piece of hardware 1153336131 M * juggo so I enabled nmi_watchdog and I actually got a trace from a recent crash 1153336152 M * juggo it was in vx_update_load which I googled and saw was related to vserver stuff 1153336203 M * juggo so I'm back again to try and figure out how to resolve this issue, because we get often less than 24 hours of uptime on our vserver boxes 1153336230 M * daniel_hozac do you have the trace? 1153336251 M * juggo I have a picture of the screen 1153336281 M * juggo I suppose I can throw it on a web server real quick and send the link 1153336288 M * daniel_hozac sure. 1153336292 M * juggo hang on 1153336320 M * daniel_hozac net/serial console makes catching panics easier ;) 1153336343 M * daniel_hozac (assuming it's not one of those causing it) 1153336344 M * juggo yeah I have been reading up on how to do that, will implement this afternoon 1153337302 M * doener hm, vx_update_load sounds makes me think use-after-free... 1153337380 M * doener juggo: which kernel version are we talking about? 1153337438 M * juggo http://68.166.181.5/P1010092.JPG 1153337441 M * juggo http://68.166.181.5/P1010091.JPG 1153337507 M * daniel_hozac damn, the most important lines are right above there. 1153337510 M * juggo debian backports kernel linux-image-2.6.16-2-vserver-686 2.6.16-16bpo1 1153337543 M * juggo ok well I guess I'll get serial debugging going 1153337573 M * daniel_hozac so i guess you were cat'ing /proc/virtual//cvirt at the time of the crash? 1153337578 M * doener hm, now which vserver version might that be... 1153337609 M * juggo well I wasn't, but I do have munin monitoring the vservers 1153337615 M * daniel_hozac ah, ok. 1153337627 M * juggo do you think that could be a culprit? 1153337636 M * daniel_hozac that's quite clear from that backtrace. 1153337641 M * doener daniel_hozac: do you by chance now if there are any important changes in that path between 2.0.x and 2.1.x? 1153337663 M * doener s/now/know/ 1153337680 M * daniel_hozac doener: not that i'm aware of, although i guess the scheduler things might change it. 1153337696 M * juggo so that's 2.6.16.25 + 2.0.2-rc21 1153337709 M * doener juggo: thanks :) 1153337715 M * juggo according to the changelog 1153337743 M * daniel_hozac not rc22? that's odd, i guess. 1153337874 M * juggo well it could always be missing from the changelog, any way to verify? 1153337874 M * daniel_hozac doener: i don't see any changes in that path. 1153337926 J * shedi ~siggi@inferno.lhi.is 1153337935 M * juggo so there are 4 plugins that monitor vserver activity that read from cvirt, any chance that somehow there's some thread issue if they are run in parallel? 1153337954 M * daniel_hozac there really shouldn't be. 1153337963 M * doener daniel_hozac: hm, couldn't that be the activate_task path instead of accessing /proc/virtual/...? 1153337969 M * daniel_hozac and if there is, it would be awesome to get it fixed ;) 1153337980 M * doener oh wait... I didn't see the end of the trace... 1153338020 M * daniel_hozac doener: i never quite learned to interpret the call trace. 1153338047 M * doener me neither, but at least I think that we came from the bottom and went to the top 1153338075 M * daniel_hozac doener: IMHO it looks like x was reading from cvirt, x got interrupted and another process got switched in. 1153338082 M * doener not sure if the trace starts with the current function or with the previous one (which would mean that we miss the actual function name which would be one or two lines above that stuff= 1153338095 M * daniel_hozac it starts with the caller. 1153338143 M * doener so we miss the function where the panic/bug/oops happened, right? (has some problems understanding english today) 1153338152 M * doener s/has/I have/ 1153338154 M * daniel_hozac yeah, i think so. 1153338162 A * doener .oO( and writing as well... obviously ) 1153338172 M * daniel_hozac lol. 1153338208 M * daniel_hozac one thing i always had a problem with the traces, as it seems it's not a x called y called z kind of list. 1153338235 M * daniel_hozac more like x called y and z, and then called a which called b... 1153338266 M * doener that's because of a) interrupts, b) process switches, c) function inlining... at least that's what my experience says 1153338294 M * doener function inlining sometimes makes it one hell of a time to read call traces 1153338449 M * daniel_hozac is it possible it's some spinlock deadlock? 1153338499 Q * slava Server closed connection 1153338516 J * slava ~slava@195.22.238.42 1153338709 M * doener hm, vxi->cvirt.load_lock is always locked last... 1153338768 M * daniel_hozac ah, i think EIP is at die_nmi. 1153338790 M * daniel_hozac (from the console shuts up message) 1153338917 Q * weeble Server closed connection 1153338938 J * weeble ~weeble@81.52.144.1 1153338938 M * doener might mean that the cpu is stuck, and the last vx_update_load call is from irq context, right? 1153338952 M * doener juggo: UP or SMP box? 1153338961 M * juggo this is UP 1153338965 M * daniel_hozac i think so. 1153338983 M * juggo but it's interesting because we also have a SMP system which does not seem to be having these problems 1153339024 M * daniel_hozac so it's easy to reproduce? 1153339038 M * juggo no, hence the word "seem" 1153339043 Q * Viper0482 Remote host closed the connection 1153339056 M * juggo but generally the 2 UP boxes are locking up within 24 hours or so 1153339092 M * juggo the smp box has been up for 2 days, rebooted after the weekends kernel upgrades, and was up for several days before that 1153339116 M * doener well, on smp the other cpu can get the process that holds the lock and free it, so that the irq can continue to run 1153339128 M * doener on up you're deadlocked 1153339131 M * juggo yes, maybe I should check for those messages 1153339146 M * doener except for the one fact that spinlocks do nothing on UP 1153339160 M * daniel_hozac CONFIG_SMP without alternatives? 1153339218 M * doener can you disable them? 1153339239 M * juggo now that kernel is smp enabled, I think to get away from having too many binary kernel variants 1153339276 M * juggo but my understanding is that wouldn't be a problem, just some extra overhead 1153339282 M * juggo on a UP system 1153339378 M * daniel_hozac you're right, you can't disable it. 1153339602 M * doener daniel_hozac: but then again, .16 doesn't know about smp alternatives ;) so it is kinda disabled :) 1153339641 M * daniel_hozac doener: hah, true! i keep using my own trees as references... 1153340054 J * kir_home ~kir@toronto-HSE-ppp4218829.sympatico.ca 1153340194 M * doener daniel_hozac: do you see any solution besides disabling interrupts? 1153340319 M * daniel_hozac well, is it even a problem on 2.6.17? 1153340365 M * doener should be "solved" there, but is that a "clean" solution? 1153340437 M * daniel_hozac i guess not. 1153340550 Q * kir_home Quit: Ухожу я от вас 1153340876 M * doener http://www.13thfloor.at/~doener/vserver/patches/juggo.diff 1153340903 M * doener that should stop the box from deadlocking... 1153340985 M * juggo ok, so a question, would this occur on say 2.6.8, I ask because I may be in the minority of people running vservers + munin vserver plugins 1153341007 M * juggo but I know micah does as well, so either everyone else is running SMP systems, or something else is at play 1153341062 M * juggo and if this is the issue it should be reproducible by creating a program that spawns many threads reading from cvirt 1153341064 M * daniel_hozac or they're running UP boxes without CONFIG_SMP, or more recent kernels ;) 1153341112 Q * bonbons Quit: Leaving 1153341126 M * juggo ah, so it's a 2.6.16 + CONFG_SMP=y + UP box where spinlocks don't work the way they need to? 1153341171 M * daniel_hozac they work the way they need to, but cause a deadlock on UP ;) 1153341309 M * juggo ok I think I understand, I've actually been putting locking into the munin plugins 1153341328 M * micah hey juggo, catching up on backlog 1153341339 M * juggo yeah I'm sure you'll find it interesting 1153341378 M * juggo I've definitely been loving the vserver thing, except for our vserver hosts crashing all the time 1153341392 M * micah daniel_hozac: the reason its 2.0.2-rc21 instead of 22 is because waldi just put in the minor fix that was in 22 in the 21 patch, I think if I recall correctly. What I do recall is that I asked waldi about that and he explained and it was alright :) 1153341406 M * juggo and if it's a munin plugin thing, well, that would explain why no one else has the problem, and why it occurs so erradically, 4 crashes one day, then none for 4 days 1153341423 M * daniel_hozac micah: so it's just misrepresenting itself then? :) 1153341433 M * micah daniel_hozac: I think so :) 1153341460 M * waldi micah: hmm? 1153341472 M * micah so to answer your question, I'm running 2.6.16-2-vserver-686 on my home desktop with the munin plugins 1153341476 M * daniel_hozac juggo: you need to have a process reading cvirt at the same time that a process in that context is switched to. 1153341501 M * micah oh humm, its a xeoon, so its not technically a UP 1153341512 M * daniel_hozac that explains it. 1153341531 M * micah waldi: i think people were puzzled why its 2.6.16.25 + 2.0.2-rc21, and not rc22 1153341544 A * micah tries to think of other machines with that kernel 1153341546 M * doener < 2.6.17 + CONFIG_SMP + UP system = deadlock race 1153341556 M * juggo yeah, when I say SMP, I mean dual core as well, Athlon X2 1153341560 M * waldi micah: ah, okay 1153341579 M * waldi this patch is not completely identical anyway 1153341601 M * juggo which as I said you could probably write a short script to test 1153341612 M * micah yeah, you probably have to adjust version.c 1153341648 M * micah so the solution is to apply doener's patch, or run 2.6.17? 1153341679 M * daniel_hozac or not run a CONFIG_SMP kernel on UP. 1153341683 M * waldi version.c is completely different 1153341698 M * doener or not using CONFIG_SMP on non-smp systems... that 2.6.17 works is just a side-effect of smp-alternatives... on non x86 it'll still deadlock 1153341708 M * waldi hmm? 1153341712 M * waldi what is a deadlock? 1153341713 M * daniel_hozac doener: will it? 1153341728 M * micah waldi: < 2.6.17 + CONFIG_SMP + UP system = deadlock race 1153341734 Q * bubulak Ping timeout: 480 seconds 1153341741 M * waldi oh 1153341753 M * doener daniel_hozac: well, smp alternatives are x86 (and x86_64?) only, aren't theyß 1153341756 M * doener s/ß/?/ 1153341758 M * waldi x86 1153341763 M * waldi x86_64 will be in .18 1153341766 J * bubulak ~bubulak@whisky.pendo.sk 1153341770 M * daniel_hozac doener: yeah, but do other architectures have similar irq handling? 1153341782 M * micah waldi: juggo is running a UP linux-image-2.6.16-2-vserver-686 2.6.16-16bpo1, and he can lock the machine up every day due to accesses to cvirt 1153341793 A * daniel_hozac really isn't that well-versed in the low-level kernel stuff. 1153341815 M * juggo isn't there a way to disable smp stuff as a kernel boot parameter? 1153341819 M * doener hm, IIRC almost every arch has a timer interrupt which will call generic code 1153341850 M * waldi juggo: nosmp? 1153341885 M * waldi micah: why is cvirt thing accessed in interrupt context? 1153341891 M * doener waldi: that just disables the cpus, but does not change the spinlock code, does it? 1153341898 M * waldi doener: no, it does not 1153341904 M * juggo yeah nosmp looks like it could be it 1153341906 M * waldi micah: this looks weird 1153341908 M * daniel_hozac waldi: it's not accessed in interrupt context, vx_update_load is called from activate_task. 1153341916 Q * fogoh Quit: Parti 1153341932 M * daniel_hozac waldi: cvirt is read by the process that is running when the timer interrupt happens. 1153341933 M * doener and reading cvirt also call vx_update_load 1153341944 M * waldi daniel_hozac: so you just disable the timer interrupt, yeah 1153342007 M * doener that's what my patch does... 1153342040 M * waldi is there any definition that activate_task can be blocked this way? 1153342064 M * waldi or will this just work because all(?) arches does that in the timer interrupt? 1153342083 M * doener spin_lock_irqsave disables all local interrupts 1153342124 M * doener and preemption is disabled while a spinlock is held IIRC, so everything but interrupts doesn't matter 1153342410 M * juggo with dual-core chips becoming more prevalent it makes sense that far more kernels are going to have CONFIG_SMP enabled by default, and that leaves the possibility that they will run on UP systems 1153342567 M * daniel_hozac SMP alternatives is meant to fix that. 1153342597 M * juggo yeah, so only a problem for particular kernels, 1153343206 N * Ben__ Ben_zZz 1153343494 M * ray6 ok, landing in nyc... cu later :) 1153343656 J * Aiken ~james@tooax6-043.dialup.optusnet.com.au 1153343890 N * Bertl_oO Bertl 1153343894 M * Bertl evening folks! 1153343923 M * Hollow hey Bertl 1153343923 M * doener hi Bertl 1153343933 Q * SNy Server closed connection 1153343934 J * SNy 37871095b1@bmx-chemnitz.de 1153343953 J * cdrx ~legoater@OTWAON23-1177993857.sdsl.bell.ca 1153344269 M * meandtheshell hi Bertl! 1153344445 M * romke evening Bertl 1153344544 Q * lilalinux Remote host closed the connection 1153345281 Q * dna_ Quit: Verlassend 1153345492 J * Pazzo ~thomas@host130-250-static.72-81-b.business.telecomitalia.it 1153345672 Q * cdrx Read error: Operation timed out 1153346303 P * stefani I'm Parting (the water) 1153346880 J * klper ~heist@83.110.124.122 1153346913 P * klper 1153347134 J * coocoon ~coocoon@p54A07E82.dip.t-dialin.net 1153347165 M * coocoon evening 1153348502 J * chand ~chand@gw.net81-65-27.noos.fr 1153349664 Q * pisco Quit: Download Gaim: http://gaim.sourceforge.net/ 1153350619 N * morrigan morigan_oO 1153350779 Q * coocoon Remote host closed the connection 1153351909 Q * sladen Ping timeout: 480 seconds 1153351954 J * sladen paul@starsky.19inch.net 1153352449 Q * Pazzo Quit: ....... 1153352454 Q * alexx Quit: Ex-Chat