1183420973 Q * ema Quit: leaving 1183421228 J * pusling pusling@88.212.70.38 1183421532 Q * bzed Quit: Leaving 1183421570 J * zLinux ~zLinux@88.213.31.27 1183421608 Q * zLinux 1183421632 M * Radiance any one recognizes this symptom: outside the vserver pinging and surfing sites goes quickly but inside the vserver pinging goes very slow (like 6 packets in 25 seconds but no packetloss reported) and surfing is also ultra slow. 1183422536 J * marcfiu ~mef@aegis.CS.Princeton.EDU 1183422577 P * marcfiu 1183423200 M * coderanger Bertl_zZ: Still sleeping? 1183423292 J * FireEgl ~FireEgl@Sebastian.Atlantica.CJB.Net 1183423526 J * DoberMann_ ~james@AToulouse-156-1-169-199.w90-38.abo.wanadoo.fr 1183423636 Q * DoberMann[ZZZzzz] Ping timeout: 480 seconds 1183426090 Q * gerrit Quit: Client exiting 1183426163 N * onox onox`zZzZ 1183427353 N * Bertl_zZ Bertl 1183427368 M * Bertl coderanger: needed a nap ... 1183427447 M * Bertl slack101: usually one uses rsync or dump/rstore or simply tar (but there are quite a number of backup solutions for linux) 1183427456 M * Bertl coderanger: what's up? 1183427904 M * slack101 lol 1183427925 M * slack101 lol @ russians trying to spend phony money in the usa 1183427950 M * slack101 Bertl, i thought vserver might have implemented something for backups 1183428088 M * Bertl no, we tend to avoid reinventig the wheel :) 1183428100 M * Bertl re-inventing even :) 1183428123 M * slack101 yea 1183428132 M * slack101 just curious 1183428137 M * slack101 i do it manually myself 1183428163 M * Bertl rdiff backup would probably be a good choice for guests 1183428204 M * slack101 rdiff ? 1183428233 M * Bertl google: rdiff (feeling lucky :) 1183428328 M * slack101 Bertl, thats written in Python 1183429044 M * lylix rsnapshot is nice as well 1183429148 M * slack101 lylix, perl ;) 1183429217 M * lylix but looks like rdiff is more efficient 1183429231 J * undefined ~undefined@adsl-68-94-189-15.dsl.rcsntx.swbell.net 1183429242 M * slack101 hmm 1183429250 M * slack101 written in perl :P 1183429253 M * slack101 what ever happened to C 1183429263 M * lylix slack101: perl? does that mean good or bad for you? heh 1183429272 M * slack101 i prefer C 1183429273 M * slack101 l( 1183429275 M * slack101 ;) 1183429586 M * lylix so use rsync directly from the shell then... 1183429635 M * cehteh mhm 1183429644 M * cehteh http://www.pipapo.org/pipawiki/BackupTools << check that out 1183429661 M * cehteh my backup thing based on rdiff-backup 1183429777 M * lylix oh no, even worse!! a shell script... lol 1183429779 M * slack101 i dont even really need backups to be honest 1183429784 M * slack101 i let users do that themselves 1183429800 M * Bertl works with the ftp access too 1183429819 M * Bertl but it could be an advantage for both sides to provide a full backup on site 1183429822 M * cehteh lylix: just a script driving rdiff-backup (which is written in python) 1183429850 M * lylix i know... just spooning out some sarcasm... sri ;) 1183429855 M * Bertl slack101: something like 'one click backup & restore' of the entire guest 1183429863 A * cehteh has that running since some years now, really simple and painless backup 1183429913 M * slack101 i dont see a reason to backup peoples stuff 1183429923 M * slack101 i give clear instructions how they can do it themselves 1183429932 M * slack101 although i do allow 1 backup to be stored on mmy server 1183429933 M * slack101 just 1 1183429940 M * cehteh well .. depends on your policies .. 1183429942 M * slack101 i might move it to 2 but i think 1 is fine 1183430018 M * cehteh people are basically just me and my wife here .. and my webserver where some more people have access .. but i think noone of them uses the backup (while i forcibly backup their ssh keys :P) 1183430093 M * lylix Bertl: what would it take to implement a stateful freeze on the guest, ie. for host migration, instead of having to stop a guest 1183430102 M * coderanger Bertl: We were wondering how the CoW stuff worked w.r.t. directories 1183430145 M * lylix i know w/ rsync such a thing can be minimized by taking a full sync on the running guest, and then momentarily stop and re-rsync the "stopped state" differences 1183430164 M * Bertl coderanger: it doesn't affect directories 1183430193 M * lylix i remember something similar in UML a few years ago (pausing i think they termed it) 1183430213 M * coderanger Bertl: Hmm, so if you have an immutable hardlink to a directory, it won't try to CoW it? 1183430230 M * Bertl coderanger: you can't put hardlinks on directories :) 1183430267 M * coderanger You can, ln just won't let you without a force 1183430316 M * Bertl well, if you really force it, it shouldn't affect the CoW, IIRC, the check is done for files only 1183430338 M * coderanger Hrmm, okay. 1183430345 M * coderanger This may result in hilarity ;-) 1183430397 M * Bertl hehe, how so? :) 1183430514 M * coderanger We had been assuming that we could CoW-link whole directories 1183430550 M * Bertl why would you want to do that? 1183430575 M * cehteh are you working on freeze/migration or is this just a braindump? 1183430592 M * coderanger Well at the very least, for the fake host context, and other things that should have CoW access to the whole filesystem 1183430631 M * lylix cehteh: braindump 1183430648 M * cehteh ok 1183430691 M * Bertl coderanger: directories are cheap, you simply do the exquivalent of 'cp -la ' for writeable things and use --bind mounts for read-only stuff 1183430712 M * coderanger Bertl: Yes, just much more work than we had been planning 1183430773 M * Bertl coderanger: happens at install time, and only for a few libraries IMHO 1183430791 M * coderanger Bertl: No, this is done continually, on the fly 1183430794 M * Bertl coderanger: you should not put unnecessary files into the activity jail anyways 1183430810 M * Bertl coderanger: how so? 1183430832 M * coderanger Bertl: Define "unnecessary" 1183430849 M * coderanger Bertl: We need to link in things like /usr/lib and /usr/bin 1183430862 M * Bertl part of it, yes 1183430874 M * coderanger No, all of it 1183430888 M * coderanger Or at least a static subset 1183430889 M * Bertl no, you won't need libraries which aren't used for example 1183430898 M * coderanger We don't know what libraries are used 1183430901 M * Bertl but let's assume you copy all of it 1183430921 M * Bertl that will be, hmm, maybe 10 inodes? 1183430960 M * Bertl maybe more if you put perl there, but let's say 100 1183430970 M * coderanger yes, its not an issue of resource usage, its just more code complexity 1183430976 M * coderanger and I don't like code complexity ;-) 1183430985 M * Bertl and when will this cp -la happen? 1183431008 M * Bertl only when you 'install' an activity or 'restore' the entire system 1183431015 M * coderanger When you run an activity 1183431032 M * Bertl why would you do that? 1183431058 M * coderanger There is another way? 1183431079 M * Bertl well, I would create the activity setup on installation 1183431087 M * coderanger How would that work 1183431099 M * coderanger Each instance needs its own container 1183431106 M * Bertl the very same way a Linux-VServer guest works now? 1183431119 M * Bertl *work 1183431122 M * coderanger Thats doing generally static allocation 1183431130 M * coderanger which we can't do 1183431141 M * Bertl please elaborate! 1183431183 M * coderanger If you create the FS containers at install time, how do you prevent multiple instances of a single activity from seeing each other's file space 1183431232 M * Bertl do multiple instances make sense for activities? 1183431325 M * Bertl besides that, regarding the libs, shouldn't be a read only mount be sufficient for that? or are activities supposed to change libraries on the fly? 1183431347 M * coderanger absolutely 1183431370 M * coderanger Yes, but they need to not share scratch space and so on 1183431381 M * Bertl which should be fine 1183431417 M * Bertl I think you can get what you want with a few bind mounts and do not even need to use/apply CoW stuff to the activities themselves 1183431424 M * coderanger For libraries in activity containers ro bind mounts are an option 1183431442 M * coderanger but that doesn't solve the problem of the whole-FS CoW 1183431517 M * Bertl well, if you _want_ an activity to be able to modify the entire FS (for the short time of its existance) then you have to do a cp -la for the entire filesystem 1183431544 M * coderanger Okay, our hope was to avoid that 1183431558 M * Bertl as I said, I don't see a point in doing so 1183431584 M * Bertl i.e. I would not even think about allowing the activities to mess with the python installation for example 1183431612 M * coderanger For Shell.activity or anything else running with full FS access, how would you do it? 1183431641 M * Bertl give me an example for that 1183431671 M * Bertl i.e. what is the user expected to get/do with that? 1183431688 M * coderanger Kid runs Shell.acitivty. They get a shell (with root available) in a container, letting them do whatever they want. 1183431710 M * Bertl and the changes they make are gone when they quit the shell? 1183431719 M * coderanger Yes 1183431733 M * Bertl is that really useful? 1183431734 M * coderanger Another use case is running non-sugarized apps 1183431737 M * coderanger Um, yes 1183431751 M * coderanger The whole point of this is to encourage kids to tinker 1183431768 M * Bertl yeah, but I would go nuts if I had to do my changes all over again :) 1183431782 M * Bertl I thought more it would be like this: 1183431796 M * Bertl - kids get access to the filesystem and can do whatever they want there 1183431806 M * coderanger No way 1183431810 M * Bertl - when they messed up pretty badly, the push a magic button 1183431830 M * Bertl - the system reverts all the changes and gives them the 'prestine original' 1183431837 M * coderanger That defeats the whole idea of a sandbox learning environment 1183431845 M * coderanger reseting the laptop is far too heavy of a hammer for this 1183431860 M * coderanger This shouldn't even involve a reboot 1183431867 M * Bertl I'm talking about resetting the FS here :) 1183431946 M * Bertl note: I don't think that a 'volatile shell' is such a bad idea, i.e. it should be fine to do that (needs the cp -la) but that is not the 'default' for 'normal' activities 1183431946 M * coderanger To implement bitfrost as written, we need FS-level CoW 1183431976 M * coderanger that cp is going to be a big resource hit as compared to doing this via unionfs 1183432007 M * Bertl well, then go for unionfs :) 1183432040 M * coderanger Heh, again an option, just not what was planned 1183432048 M * coderanger I'll let Ivan know and see what he says 1183432076 M * Bertl you should probably do a few tests first regarding performance 1183432092 M * Bertl i.e. try with unionfs vs. a full cp -la for example 1183432123 M * Bertl check how much overhead you get for creation, normal operation (i.e. fs performance) and such 1183432183 M * Bertl but it might be, that for those volatile activities, unionfs is indeed the better choice 1183432211 M * coderanger Yeah, the question is how much of the FS we expect will get accessed and/or changed 1183432252 M * coderanger (and that will change in different use cases) 1183432282 M * Bertl the shell is definitive a special case IMHO 1183432304 M * Bertl and there you probably want to actually link the entire filesystem tree 1183432329 M * Bertl for typical 'activities' OTOH, IMHO an almost complete RO fs would suffice 1183432347 M * Bertl probably only needs a /tmp or so to be happy 1183432360 M * coderanger Yes, activities wouldn't need unions 1183432384 M * coderanger They get a few megs of scratch space, but we can do almost everything they need with RO binds 1183432436 M * coderanger Some hard linking will be needed for the datastore file access, but that should be simple too 1183432454 M * Bertl the data store is persistant, no? 1183432463 M * coderanger Yes 1183432476 M * Bertl i.e. that should not be inside an activity? 1183432490 M * coderanger Not the datastore's storage 1183432493 M * coderanger The access system 1183432504 M * Bertl and that needs to be volatile? 1183432517 M * coderanger i.e., you request a file from the datastore, users selects which, it gets hardlinked into your container 1183432534 M * coderanger and you get passed a name for it 1183432540 M * Bertl ah, okay 1183432548 M * coderanger no, not volatile, just a RO link 1183434701 M * undefined has anybody seen "RTNETLINK answers: Numerical result out of range" when starting & stopping a vserver on kernel 2.6.21 1183434740 M * undefined and only one of two interfaces for the guest is active 1183434761 M * Bertl looks like you have a typo in the guest config somewhere 1183434782 M * Bertl undefined: use --debug for startup and upload the output to paste.linux-vserver.org 1183434795 M * undefined this config works fine with 2.6.15, 2.6.18, & 2.6.20 1183434800 M * undefined Bertl: ok 1183435035 M * undefined Bertl: http://paste.linux-vserver.org/3280 1183435203 M * undefined Bertl: is my alias too long? 1183435248 M * undefined Bertl: "label gigabit:apt-proxy" generates the error 1183435263 M * undefined Bertl: but "label gigabit:apt" doesn't 1183435268 M * Bertl is your interface really called gigabit? 1183435272 M * undefined yep 1183435284 M * undefined Bertl: ifrename 1183435298 M * Bertl yep, seems like that is too long 1183435322 M * Bertl but you probably do not need to use aliases anyway 1183435322 M * undefined so i take it this name length rule changed sometime in 2.6.21 1183435343 M * undefined because this exact config worked fine previously 1183435345 M * Bertl yep, likely, I would suggest removing the 'name' entry in the config for all interfaces 1183435357 M * undefined ok 1183435363 M * Bertl will make your guest look 'more natural' anyway 1183435388 M * undefined thanks for pointing out the "--debug" option to vserver 1183435398 M * Bertl you're welcome! 1183435413 M * undefined i knew there had to be something (environment variable, option, etc) but couldn't find anything documented 1183435434 M * Bertl usually 'vserver --help' is a good source of information 1183435439 M * undefined i tried to correlate the error message to a command using strace (what i normally do for userland) but that didn't work 1183435452 M * Bertl (and of course, 'vserver - build --help' :) 1183435502 M * undefined "vserver --help | grep debug" doesn't show anything 1183435526 M * Bertl what version= 1183435532 M * Bertl s/=/? 1183435545 M * undefined vserver 0.30.213 1183435575 M * Bertl daniel_hozac: maybe add that to the --help? 1183436195 N * onox`zZzZ onox 1183437354 Q * undefined Quit: Leaving. 1183437997 M * slack101 and i will attempt to compile the kernel Again :P 1183438038 M * Bertl ah, right, the kernel compile failed for you? 1183438648 M * slack101 first 2 times 1183438694 M * slack101 not really sure why either 1183438702 M * slack101 looks like it couldnt mount any file system 1183438714 M * slack101 but since i dont have my remote console there right now its a little more difficult 1183438765 M * Bertl aha, so not the compile failed, the boot failed ... 1183441175 J * hallyn_ ~xa@adsl-75-0-154-11.dsl.chcgil.sbcglobal.net 1183441175 Q * hallyn Read error: Connection reset by peer 1183442293 Q * Adrinael_ Ping timeout: 480 seconds 1183442652 J * neuralis ~krstic@solarsail.hcs.harvard.edu 1183442660 M * neuralis Bertl: 'evening. around? 1183442688 M * Bertl neuralis: yep, still around 1183442696 M * neuralis you pang earlier. what's up? 1183442729 M * Bertl we tried to figure where, when and how the filesystem for activities is created 1183442750 M * neuralis okay 1183442775 M * neuralis i still believe it should be a one-time operation at install time, generally 1183442789 M * Bertl that was my impression too 1183442809 M * Bertl coderanger: thinks it is better/intended to create it on activity start 1183442826 M * Bertl and probably destroy it again after the activity is terminated 1183442830 M * neuralis that doesn't sound right at all 1183442847 M * neuralis the files exported to activities only change when os updates are performed 1183442852 M * Bertl he has a point for starting one and the same activity several times though 1183442860 M * neuralis right 1183442864 M * coderanger neuralis: Except for the per-instance files 1183442879 M * neuralis coderanger: which i still think is somewhat of a special case 1183442895 M * neuralis but there's an easy compromise here 1183442910 M * Bertl iif the entire activity needs to create/destroy the filesystem on each and every startup, it might be better to consider something like unionfs 1183442910 M * coderanger neuralis: In any case, thats not the problem 1183442919 M * neuralis preallocate the guest fs at install time, then make dynamic copies when additional instances are started 1183442941 M * coderanger neuralis: Like my email said, the biggest problem is P_FS_RUN, Shell.activity, etc etc 1183442949 M * neuralis right 1183442953 M * neuralis so, unionfs isn't much of an option 1183442961 M * neuralis too buggy/unstable in production 1183442970 M * Bertl for the Shell activity I would see two usage scenarios 1183442982 M * neuralis otoh, i think manual VS COW is fine, because the OS files won't change underneath us 1183442984 M * Bertl (at least that is how I would prefer it) 1183443009 M * Bertl the 'admin shell' which basically gives you access to a volatile copy of the OS 1183443023 M * Bertl the 'one time play shell' which gives you a complete sandbox 1183443035 M * Bertl (which doesn't outlive the activity) 1183443063 M * neuralis right 1183443081 M * neuralis the former doesn't need its own context created 1183443082 M * Bertl in the latter case, a full 'cp -la' would be required 1183443084 M * neuralis it can just be migrated into the shadow context 1183443117 M * neuralis i don't think the 'one time play shell' is something we want or need initially 1183443142 M * neuralis when you get your machine, P_SF_RUN is enabled, meaning the host context exports the underlying FS to the shadow context as a r/o bindmount 1183443149 M * neuralis you can get a shell in the shadow context, but you can't touch the files 1183443156 M * coderanger neuralis: "one time play shell" is the basis for running legacy linux apps, which some people have been insisting very strongly we support 1183443162 M * neuralis if you disable P_SF_RUN and reboot, instead of a bindmount, we do a "cp -la" into the shadow context 1183443188 M * neuralis at which point the shell you launch in the shadow context can mutate the (COW copy of the) FS 1183443189 M * Bertl coderanger: but RO should suffice there 1183443204 M * coderanger Bertl: Not for all things. 1183443209 M * neuralis coderanger: we don't need to support all things 1183443214 M * Bertl neuralis: yeah, that is how I remembered it 1183443216 M * neuralis coderanger: in fact, full support for legacy apps has never been a requirement 1183443223 M * neuralis coderanger: and it's sure as hell not getting imposed as one now 1183443241 M * coderanger neuralis: You should have a chat with SJ about this at some point 1183443244 M * Bertl coderanger: still there are options to work around that for 'special' cases (inside the activity) 1183443270 M * neuralis coderanger: he can talk to me about it if i need to clarify something; now is way too late to be coming around about this 1183443275 M * Bertl coderanger: for example, the activity could 'copy' the read only files it wants to modify, and do a bind mount? 1183443317 M * neuralis coderanger: also, legacy apps might be handled through a "legacy activity" 1183443319 M * Bertl so I think, except for the 'user play shell' case, activities should be fine 1183443320 M * coderanger Bertl: That would assume the app have some knowledge of the OLPC environment, which makes it not legacy 1183443336 M * Bertl coderanger: the magic word here is wrapper :) 1183443340 M * neuralis coderanger: such that it's one activity that has a prepopulated full COW copy of the underlying fs 1183443348 M * neuralis coderanger: and you install and run all legacy apps within that. 1183443357 M * coderanger neuralis: This is then P_SF_FULL we talked about earlier 1183443373 M * neuralis correct 1183443378 M * neuralis which doesn't bother me 1183443396 M * coderanger neuralis: You had said earlier today that it did ;-) 1183443409 M * coderanger (doing this via a full deep copy) 1183443428 M * Bertl coderanger: actually it is a shallow copy :) 1183443480 M * coderanger Not if we are going for a full CoW image 1183443481 M * Bertl I mean, only a shallow copy is required thanks to CoW to get a fully functional filesystem (mutable) 1183443499 M * Bertl coderanger: yes, only directories will be created 1183443516 M * coderanger That doesn't make it a shallow copy, it still needs to do a full recursion 1183443544 M * Bertl full copy: cp -a shallow copy: cp -la 1183443624 M * coderanger different use of the term I suppose. I would call it a shallow copy only if there was a fixed recursion horizon 1183443627 M * harry Bertl: what's the reason that by default /tmp is ramdisk? 1183443639 M * neuralis coderanger: why do full link copies bother you? 1183443647 M * neuralis coderanger: i tried explaining earlier that this is extremely cheap 1183443663 M * Bertl harry: performance and reduced I/O 1183443694 M * harry Bertl: i just wonder if /tmp is used that much, and if it's in real life, such a big deal 1183443695 M * coderanger neuralis: Because its more time consuming and resource intensive than not doing it? I understand that its not a huge hit, but its non-0. 1183443713 M * Bertl harry: many apps create short lived temporary files, many guests means a lot of those files, which causes quite some I/O 'noise' 1183443723 M * harry true 1183443740 M * coderanger neuralis: Which is why I think its better if we try to reduce the number of cp -la's and use RO bind mounts where possible 1183443756 M * neuralis coderanger: you mean, exactly like the spec currently proposes? 1183443763 M * neuralis coderanger: you're making a mountain out of a mole hill 1183443768 M * neuralis coderanger: we're only talking about COW in special cases 1183443772 M * neuralis coderanger: where it doesn't bother me one bit 1183443778 M * Bertl :) 1183443805 M * neuralis let alone eight of them, or two bytes 1183443809 M * coderanger neuralis: Okay then, that wasn't the impression I got from you earlier today when asking about the directory-CoW stuff 1183443836 M * harry i have a question about that 1183443841 M * harry if i cp -al a guest to a new one 1183443849 M * harry and i want to upgrade one of them 1183443859 M * harry doesn't that necessarily break the link 1183443877 M * Bertl yep, given that the files are marked properly as CoW links 1183443896 M * harry how do you mark them? 1183443912 M * harry and... if i update 1, and then the second 1183443915 M * neuralis coderanger: we're not using COW anywhere by default. it's a special case of the shadow context when P_SF_RUN is disabled, and another special case for the updater. adding a third special case for the "legacy apps activity" is fine. it's still three cases, which is nothing, particularly because the operation is cheap to begin with. 1183443915 M * Bertl by setting the immutable and unlink (likage invert) flag 1183443919 M * harry they will still be entirely copied? 1183443963 M * Bertl harry: that is, why it is suggested to re-run the unification from time to time 1183443978 M * harry ah, nice :) 1183443990 M * Bertl harry: thus guests which came to some kind of sync via different pathes, will share those files again 1183443998 M * harry Bertl: and how good an idea is it to bindmount all guests to eachother 1183444013 M * harry but only make a different mountpoint for /var and /usr/local ? 1183444023 M * harry and et 1183444024 M * harry c 1183444034 M * Bertl depends, with RO bind mounts, it's probably fine 1183444062 M * harry that would be extremely nice, because every server can have it's own config / startup scripts and programs 1183444069 M * Bertl with rw bind mounts across guests, you might run into security and tagging issues 1183444069 M * harry while the base system can be upgraded from 1 system 1183444089 M * harry yeah, ro bind mounts off course :) 1183444148 M * harry that's faster than copying entire guests to a different volume, right? 1183444163 M * harry (copying makes it use different files => different caches) 1183444163 Q * onox Ping timeout: 480 seconds 1183444279 M * harry Bertl: what should i do with the ipv6 patch from bonbons? should i document that one somewhere? (it's only for grsec, as i don't want to take over bonbons great work on ipv6) 1183444296 M * harry seems the only patch available for 2.6.21.5 1183444559 J * cedric ~cedric@rny93-2-82-66-66-30.fbx.proxad.net 1183444817 J * onox ~onox@kalfjeslab.demon.nl 1183444867 J * onox_ ~onox@kalfjeslab.demon.nl 1183445116 M * onox test 1183445125 Q * onox_ 1183445167 M * onox hmm, stupid irssi 1183445205 J * HeinMueck ~Miranda@host-88-217-199-211.customer.m-online.net 1183445330 Q * onox Quit: leaving 1183445350 J * onox ~onox@kalfjeslab.demon.nl 1183446023 Q * cedric Quit: cedric 1183446584 J * rgl ~Rui@84.90.10.107 1183446867 J * meandtheshell ~markus@85.127.114.183 1183448049 J * dna ~naucki@55-246-dsl.kielnet.net 1183451457 J * cedric ~cedric@80.70.39.67 1183451759 N * DoberMann_ DoberMann 1183453086 Q * rgl Ping timeout: 480 seconds 1183454012 Q * eSa| Quit: Coyote finally caught me 1183454429 Q * Guy- Ping timeout: 480 seconds 1183455338 Q * Radiance Ping timeout: 480 seconds 1183455374 J * esa bip@ppp-62-123-66-215.dial.atlanet.it 1183455377 N * esa eSa| 1183455461 Q * eSa| 1183455462 Q * Greek0 Ping timeout: 480 seconds 1183455497 Q * tanjix 1183455747 J * esa bip@ip-87-238-2-45.adsl.cheapnet.it 1183455756 N * esa eSa| 1183456816 J * rgl ~Rui@84.90.10.107 1183457080 J * Vudu 77f539392c@perverz.hu 1183457167 M * Bertl welcome Vudu! 1183457188 Q * Vudumen Ping timeout: 480 seconds 1183457623 M * Bertl okay, finally off to bed now ... have a good one everyone! 1183457629 N * Bertl Bertl_zZ 1183458525 J * starcode ~starcode@host-82-135-57-200.customer.m-online.net 1183458644 M * starcode Hi, I have a logical problem with mounting a directory into a vserver 1183458651 M * starcode I need to mount --bind a directory from the vserver host into a guest, whos root is mounted via nfs 1183458719 M * starcode But when I mount the nfs root the server is already switched in the vnamespace of the vserver so that I cannot access the vserver hosts namespace anymore to mount --bind a directory from it (it's out of scope already) 1183458727 M * starcode Any idea? 1183458825 J * Adrinael adrinael@rid7.kyla.fi 1183459091 M * onox starcode: mount in fstab in /etc/vservers/...? 1183459176 Q * dna Quit: Verlassend 1183459257 M * starcode Yes 1183459270 M * starcode But thats not the problem 1183459295 M * starcode My root of the vserver is only visible within the vserver (vnamespace) 1183459360 M * starcode But at the moment when the root is mounted, all paths in fstab are already relative to the vserver root... so i cannot mount --bind a directory from outside of the vserver root 1183459393 M * starcode But I cannot mount --bind the directory when still in the hosts vnamespace context, as at this time the vserver root is not mounted yet 1183459600 J * Piet hiddenserv@tor.noreply.org 1183459632 M * starcode Currently i start the vserver normally and afterwards run "vnamespace -e guestname mount --bind /hostdir-to-mount /vserver/guestname/guest-path" as with this I have access to both vnamespaces 1183459694 J * dna ~naucki@55-246-dsl.kielnet.net 1183460327 M * doener starcode: the source path in /etc/vservers/.../fstab is relative to the host's root, while the destination path is relative to the vserver's root 1183460488 M * starcode doener: Ohhh, cool. Until now I only needed to mount proc, tmpfs and nfs-directories as my vserver hosts are diskless... 1183460538 M * starcode But it seems not to be well documented... at least not in http://linux-vserver.org/util-vserver:Documentation 1183460631 M * starcode I'll add this to the docs when I get it working this way for me 1183462004 J * phedny_ ~mark@ip56538143.direct-adsl.nl 1183462348 J * Guy- MPHK444AXO@chardonnay.math.bme.hu 1183462407 Q * phedny Ping timeout: 480 seconds 1183462641 Q * Aiken Quit: Leaving 1183464181 Q * nkukard Quit: Leaving 1183466878 J * zul_ ~chuck@CPE0006258ec6c1-CM000a73655d0e.cpe.net.cable.rogers.com 1183466878 Q * zul Read error: Connection reset by peer 1183467032 J * ema ~ema@rtfm.galliera.it 1183468464 Q * ema Read error: Connection reset by peer 1183468465 J * ema ~ema@rtfm.galliera.it 1183469223 Q * eSa| Read error: Connection reset by peer 1183469358 J * esa bip@ip-87-238-2-45.adsl.cheapnet.it 1183469367 N * esa eSa| 1183471629 N * DoberMann DoberMann[PullA] 1183471734 Q * haxier Quit: Konversation terminated! 1183473282 J * fatgoose ~samuel@76-10-153-123.dsl.teksavvy.com 1183473286 N * fatgoose samueltc 1183477827 Q * HeinMueck Quit: Aah! 1183478440 N * ensc Guest362 1183478450 J * ensc ~irc-ensc@p54B4EAA8.dip.t-dialin.net 1183478557 Q * Guest362 Ping timeout: 480 seconds 1183478812 J * gerrit ~gerrit@c-67-169-199-103.hsd1.or.comcast.net 1183478980 J * bonbons ~bonbons@2001:5c0:85e2:0:20b:5dff:fec7:6b33 1183481523 Q * bonbons Quit: Leaving 1183481996 Q * micah Remote host closed the connection 1183482075 J * oauto ~micah@micah.riseup.net 1183482282 Q * gerrit Ping timeout: 480 seconds 1183482760 J * HeinMueck ~Miranda@dslb-088-065-041-214.pools.arcor-ip.net 1183482924 Q * ema Quit: leaving 1183482998 J * gerrit ~gerrit@c-67-169-199-103.hsd1.or.comcast.net 1183483190 N * DoberMann[PullA] DoberMann 1183483257 Q * Wonka Ping timeout: 480 seconds 1183483496 J * Wonka produziert@chaos.in-kiel.de 1183484539 J * bzed ~bzed@dslb-084-059-121-172.pools.arcor-ip.net 1183484659 Q * cedric Quit: cedric 1183485160 Q * samueltc Quit: samueltc 1183485557 J * Greek0 ~greek0@85.255.145.201 1183486591 N * Bertl_zZ Bertl 1183486596 M * Bertl morning folks! 1183486660 M * daniel_hozac morning Bertl! 1183486896 Q * Guy- charon.oftc.net unununium.oftc.net 1183486896 Q * SadMan charon.oftc.net unununium.oftc.net 1183486896 Q * Loki|muh charon.oftc.net unununium.oftc.net 1183486896 Q * vasko charon.oftc.net unununium.oftc.net 1183486896 Q * jannic charon.oftc.net unununium.oftc.net 1183486896 Q * pyquila charon.oftc.net unununium.oftc.net 1183486896 Q * fosco charon.oftc.net unununium.oftc.net 1183486896 Q * kaner charon.oftc.net unununium.oftc.net 1183487065 J * Guy- MPHK444AXO@chardonnay.math.bme.hu 1183487065 J * pyquila gerbens@82.94.222.35 1183487065 J * kaner kaner@strace.org 1183487065 J * jannic ~jan@gasko.hitnet.RWTH-Aachen.DE 1183487065 J * fosco fosco@konoha.devnullteam.org 1183487065 J * vasko ~vasko@unreal.rainside.sk 1183487065 J * Loki|muh loki@satanix.de 1183487065 J * SadMan sadman@sadman.net 1183487959 M * slack101 Bertl, Morning Bertl 1183488131 J * zul ~chuck@CPE0006258ec6c1-CM000a73655d0e.cpe.net.cable.rogers.com 1183488131 Q * zul_ Read error: Connection reset by peer 1183488317 M * Bertl slack101: good morning! 1183488753 M * daniel_hozac Bertl: http://people.linux-vserver.org/~dhozac/p/k/delta-fdiattr-feat01.diff is there anything i've missed? 1183488787 M * daniel_hozac (other than being consistent in the coding style...) 1183488838 J * marcfiu ~mef@aegis.CS.Princeton.EDU 1183489179 J * dna_ ~naucki@55-246-dsl.kielnet.net 1183489277 M * rgl what is an iattr daniel_hozac ? 1183489323 M * daniel_hozac inode attribute. 1183489339 M * rgl but what is it? :D 1183489414 M * daniel_hozac barrier, iunlink, immutable, etc. 1183489572 Q * dna Ping timeout: 480 seconds 1183489934 M * Bertl daniel_hozac: hmm, couldn't we put the fd in id? 1183489974 M * daniel_hozac i guess so. 1183489980 M * Bertl (otherwise we need compat wrappers for the int) 1183489988 M * daniel_hozac really? how come? 1183490050 M * Bertl hmm, wasn't there an arch with different packing rules or so? 1183490064 M * Bertl anyway, id is unused, I guess? 1183490069 M * daniel_hozac yeah. 1183490085 M * daniel_hozac but shouldn't the packing rules affect the structure either way? 1183490108 M * daniel_hozac i mean, unsigned or signed shouldn't matter for that sort of thing, IMHO. 1183490196 M * Bertl IIRC, it's more int32_t vs int , but maybe I'm remembering it completely wrong 1183490260 M * Bertl rest looks fine, I take it was run tested? 1183490262 M * daniel_hozac isn't int == int32_t on all gccs? 1183490271 M * daniel_hozac yeah, returns the same things as vc_get_iattr. 1183490315 M * Bertl good, let's move the fd to id, then I'm fine with it, tx! 1183490326 M * daniel_hozac okay, great. 1183490426 M * marcfiu hello vsWorld 1183490439 M * marcfiu daniel_hozac and Bertl: good to see things moving forward. :) 1183490444 M * Bertl hey marcfiu! how was LXS? 1183490488 M * marcfiu LXS? 1183490491 M * marcfiu You mean OLS? 1183490504 M * Bertl yeah, right! 1183490533 A * Bertl is not fully awake yet ... 1183490536 M * marcfiu OLS was good. There was a good audience that attended the Linux-VServer presentation that I gave on your behalf. 1183490559 M * marcfiu There were a few questions at the end, but mostly from Kir K. from the OpenVZ group. 1183490576 M * marcfiu His buddies Pavel and another guy were there, too. 1183490601 M * marcfiu I wish there would have been more questions & interaction. 1183490618 M * marcfiu Got much more feedback and interest about Linux-VServer at the PlanetLab Birds of a Feather session. 1183490698 M * Bertl sounds good to me .. what were the 'OVZ' questions? 1183490726 M * marcfiu mostly related to resource management. 1183490766 M * marcfiu Nothing really deep... 1183490820 M * Bertl i.c. well, thanks for doing our presentation ... I take there is a video somewhere? 1183490850 M * marcfiu Benoit from Revolution Linux was at the PlanetLab BoFS. He's got a bunch of customers, including some large school district in Canada / Montreal, for which he runs a large number of xterminals off of VServer-based servers. 1183490866 M * marcfiu I have no idea if there is a video somewhere. 1183490925 M * Bertl ah, we'll see ... 1183490943 M * marcfiu We are going to sync up with Benoit to figure out if there is any way to collaborate on VServer management for large scale systems. 1183490953 M * marcfiu Benoit is buddies with Jacques. 1183490962 M * bXi Linux venice 2.6.20-vs2.3.0.11-gentoo #1 Tue Jul 3 19:31:51 CEST 2007 i686 AMD Athlon(tm) 64 Processor 3000+ AuthenticAMD GNU/Linux 1183490973 M * marcfiu Jacques == VServer father?! 1183491013 M * Bertl marcfiu: yep, the guy who started Jails for Linux :) 1183491038 M * bXi all hail jacques then 1183491280 M * bXi Bertl: do vservers have full access to the graphics card? 1183491329 M * Bertl given the necessary permissions, yes 1183491367 M * bXi so gaming in a vserver would be possible 1183491405 M * Bertl there is a project called MoreUbuntu, which realizes multi seat linux with Linux-VServer ... 1183491431 M * bXi cool 1183491483 M * Bertl not something which works out of the box though, as you usually need to lower securityy somewhat 1183491501 M * Bertl (hardware access is always problematic) 1183491969 J * fatgoose ~samuel@76-10-153-123.dsl.teksavvy.com 1183492578 M * Bertl wb fatgoose! 1183492588 M * fatgoose yo 1183492594 N * fatgoose samueltc 1183493526 M * slack101 Bertl, what do you think the best way to control a cluster of Vservers would be ? 1183493550 M * slack101 for my web panel i am making ....do you think i should use a daemon ....or have it right directly to the system 1183493551 M * slack101 hmmm 1183493567 M * daniel_hozac myPLC! :) 1183493584 M * slack101 myPLC? 1183493692 M * slack101 thing is using daemon i could cluster systems ..........but with no daemon i could right directly to the system from the script 1183493692 M * slack101 also 1183493700 M * slack101 how do i prevent port scans anyone know ? :P 1183493722 M * marcfiu slack101: http://www.planet-lab.org/doc/myplc 1183493724 M * daniel_hozac "prevent"? 1183493751 M * slack101 disallow 1183493752 M * slack101 stop 1183493772 M * daniel_hozac which would mean what, exactly? 1183493778 M * marcfiu we've developed MyPLC to manage the PlanetLab facility, which consists of 700+ machines located in 30+ countries. 1183493784 M * marcfiu see www.planet-lab.org for more info. 1183493810 M * marcfiu slack101: We also use MyPLC to manage a local cluster of servers, probably in a manner consistent with what you are after. 1183493845 M * marcfiu daniel_hozac is spending the summer with us working on MyPLC. 1183493855 M * slack101 i seeeee 1183493858 M * marcfiu So in that sense his opinion might not be entirely unbiased. :) 1183493870 M * marcfiu Our software still requires a bunch of polishing. 1183493873 M * slack101 how exactly does it cluster then ? 1183493887 M * slack101 make it all look like one system ? 1183493889 M * slack101 or what ? 1183493895 M * marcfiu it does not make it look like one system. 1183493929 M * marcfiu It is used to manage & provision vservers on a large set of machines. 1183493942 M * slack101 hmmmmmm 1183493949 M * slack101 how does it go about doing that ? 1183493979 M * marcfiu it keeps information about nodes (machines), mappings of vservers -> nodes, and mappings of users -> vservers. 1183494040 M * marcfiu Each machine runs a daemon (called the node manager), which communicates via XMLRPC to an API server (database) to find out which vservers are supposed to be on the local node and in what way they should be provisioned. 1183494052 M * samueltc So in theory, these distributions should work on virtually any Linux 2.6 based distribution, whether it supports rpm or not. 1183494066 M * samueltc but the word 'fedora' is in almost each sentence :/ 1183494101 M * slack101 marcfiu, i seee 1183494106 M * marcfiu You can manage each MyPLC via a Web GUI (based on drupal and you can skin/theme it in whatever way you want). 1183494117 M * slack101 manage to do what ? 1183494129 M * slack101 i am making a customer UI right now 1183494134 M * marcfiu But you can also ignore the Web GUI and just communicate with the API server to create the vserver->node mappings. 1183494136 M * slack101 so they can start stop restore 1183494144 M * marcfiu slack101: sure 1183494173 M * slack101 is this the safest route ? :P 1183494187 M * marcfiu I don't know what you mean by 'safest route'. 1183494189 M * slack101 then having a daemon on each machine and have the web script directly communicate with it 1183494194 M * marcfiu I have to leave now... 1183494206 M * slack101 ah 1183494218 M * marcfiu please drop me a line: mef@cs.princeton.edu or join the devel@lists.planet-lab.org mailing list to get further info. 1183494228 M * marcfiu cheers 1183494235 P * marcfiu 1183494252 M * slack101 daniel_hozac, do you recomend this route for what i am trying to do ? 1183494259 M * slack101 is this specifically for vservers or no ? 1183494272 M * daniel_hozac the whole system is based on Linux-VServer. 1183494296 M * slack101 i dont see 1 word of linux vserver on that site 1183494333 M * slack101 well i see it 3 times 1183494340 M * slack101 but im jus curious whats it real purpose is 1183494410 M * slack101 im not seeing why it uses its own chroot enviroment ? 1183494459 M * daniel_hozac hmm? 1183494524 M * slack101 im lost 1183494526 M * samueltc daniel_hozac: how does it interact with the vs api? 1183494535 M * slack101 about really what i want and need ot use to accomplish it in a safe manner 1183494546 M * daniel_hozac mostly through a python library. 1183494576 M * slack101 but what does this do thats soo great ? 1183494593 M * daniel_hozac it manages a bunch of hosts. 1183494605 M * daniel_hozac (700+, as Marc said) 1183494608 M * slack101 what can it do to the host ? 1183494644 M * daniel_hozac deploy guests. 1183494663 M * samueltc fedora guests? 1183494686 M * daniel_hozac any guests, i guess. 1183494706 M * daniel_hozac note that it's not really on par with say util-vserver with regards to configurability. 1183494710 M * slack101 i think this would complicate my little web panel i am trying to make 1183494723 M * slack101 then again i do need a way to link the servers so they can be controlled from a certain point 1183494728 M * daniel_hozac it doesn't really use network contexts for instance. 1183494787 M * slack101 wait what am i doing 1183494793 M * slack101 i will just use OpenVCP 1183494802 M * slack101 and change the design 1183494837 M * slack101 since they already have a daemon written in C 1183494838 M * slack101 etc 1183495061 M * samueltc slack101: what do you want in your pannel? 1183496766 J * Aiken ~james@ppp121-45-220-241.lns2.bne1.internode.on.net 1183497533 J * fatgoose ~samuel@76-10-153-123.dsl.teksavvy.com 1183497854 Q * samueltc Ping timeout: 480 seconds 1183499386 J * oxylin ~jpeeters@chv78-2-88-161-189-78.fbx.proxad.net 1183499401 M * Bertl wb oxylin! 1183499457 Q * fatgoose Ping timeout: 480 seconds 1183499668 M * oxylin Hi Bertl ;) 1183499691 M * oxylin Bertl, my vservers wotk fine, thx for your help some days ago 1183499935 Q * rgl Quit: Leaving 1183499974 M * Bertl oxylin: good to hear! you're welcome! 1183500010 Q * oxylin Quit: Ex-Chat 1183500047 M * oauto Bertl: my vservers work fine too, thanks for all your work over the years! 1183500069 M * Bertl the pleasure is mine! tx for the feedback! 1183500079 M * oauto oops 1183500081 N * oauto micah 1183501117 J * oxylin ~jpeeters@chv78-2-88-161-189-78.fbx.proxad.net 1183501219 Q * wenchien Ping timeout: 480 seconds 1183501251 Q * meandtheshell Quit: Leaving. 1183501783 M * slack101 i sure hope onox and oxylin has done thier donations 1183501813 M * oxylin slack101, why do you say that? 1183501843 M * daniel_hozac where did you get onox from? 1183501843 M * slack101 well your using vserver correct ;) 1183501864 M * slack101 sorry micah 1183501876 M * Bertl he is our debian maintainer :) 1183501882 M * slack101 the only way software progresses is with money 1183501897 M * daniel_hozac ...or by actually helping distribute/make it. 1183501911 M * slack101 Bertl, is just a kind of different individual where he likes what he does i guess .......but im sure money will improve the project quite a bit ;) 1183501941 M * slack101 so if your using it for commercial purposes its only right too donate some ;) 1183501964 M * daniel_hozac you realize money isn't the only resource, right? 1183501965 M * Bertl I haven't received your donation yet ... but let me check once again :) 1183501990 M * daniel_hozac IMHO we're in desperate need of someone who can take care of the documentation... 1183502013 M * slack101 Bertl, i havent even got it to compile yet ......... ;) 1183502027 M * slack101 as ive always said when i start to use it for commercial purposes i will donate alot 1183502037 M * slack101 as of course i see the project moving on 1183502043 M * slack101 not going dead ;) 1183502052 M * Bertl slack101: you reported 20+ users last time :) 1183502175 M * micah slack101: not using it commercially myself, but I do believe in helping out in return with things that benefit me 1183502193 M * slack101 Bertl, not on vserver 1183502210 M * slack101 i converted a few over as i said last time 1183502218 M * slack101 but i put them back on the other server 1183502224 M * slack101 until i get this how i want it ;) 1183502231 M * slack101 and i will take screens when i donate ;) 1183502254 M * slack101 Bertl, i was looking out for you ;) then you try to make me look like the bad guy :) 1183502339 M * Bertl well, it would look much better if you could point to your name on the Hall'o'Fame with some impressive work/donation 1183502349 M * slack101 yes o yes 1183502364 M * slack101 you will see ;) 1183502424 M * Bertl looking forward ... 1183502469 M * daniel_hozac Bertl: i don't think i pasted this earlier, http://people.linux-vserver.org/~dhozac/p/k/delta-fdiattr-feat02.diff 1183502610 M * slack101 daniel_hozac, do you feel like you are a value to this community ? 1183502661 M * slack101 :D 1183502663 M * daniel_hozac i'm not sure how to interpret that. 1183502666 M * slack101 im kidding 1183502672 A * slack101 is waiting for kernel to compile 1183502738 M * slack101 jus curious daniel_hozac but you in austria too ? 1183502741 M * daniel_hozac no. 1183502758 M * daniel_hozac i'm from Sweden, but i'm at Princeton right now. 1183502832 M * slack101 Princeton ? 1183502840 M * daniel_hozac New Jersey. 1183502841 M * slack101 the college in the ivy league ? 1183502855 M * slack101 ah 1183502859 M * slack101 why ? :P 1183502881 M * slack101 how long you going to be there ? 1183502894 M * slack101 my brother went to yale it almost killed him 1183502903 Q * dna_ Quit: Verlassend 1183502916 M * slack101 mentally / finance wise 1183502928 M * daniel_hozac 2 months. i'm working on planet-lab with marcfiu. 1183502929 M * onox slack101: the Skull and Bones gang went after him? 1183502988 J * Radiance df0829c314@halt.1984world.eu 1183502992 M * slack101 daniel_hozac, you going back to sweden ? 1183503008 M * daniel_hozac yes... 1183503017 M * slack101 i seee 1183503026 M * slack101 new jersey is to cold 1183503045 M * daniel_hozac i think it's too hot. 1183503064 M * slack101 you just never been to miami 1183503075 M * slack101 or down south :D 1183503115 M * Bertl daniel_hozac: looking at the existing code, I just saw that we pass the id to vc_[gs]et_iattr, which is currently unused ... I would prefer to change that and rename the 'id' to 'fd' in the new f[gs]et ... but I will fix that up when integrating the patch (shouldn't change the behaviour ... 1183503191 M * daniel_hozac what are you implying? making vc_[sg]et_iattr multiplex both names and fds? 1183503214 M * Bertl no, not at all, just removing the obsolete argument 1183503244 M * daniel_hozac ah, okay. yeah, makes sense. 1183503301 M * daniel_hozac speaking of cleanup, i noticed that vc_get_iattr_x32 vx_checks. isn't that performed in the switch these days? 1183503323 M * Bertl yep, should be obsoleted by now too 1183503381 M * Bertl of course, if you feel like cleaning up, go ahead, I should be back on real Linux-VServer tomorrow (I almost finished updating my complete home infrastructure :) 1183503394 M * Bertl *Linux-VServer work 1183503525 M * daniel_hozac btw, is there a reason why we pass an inode to get and a dentry to set? 1183503547 M * daniel_hozac AFAICT we only use the inode in set as well. 1183503575 M * Bertl I think that has a historical reason, i.e. we had to do some updating or so back then 1183503595 M * Bertl but inode should be more than fine nowadays 1183503599 M * daniel_hozac okay. cleanup? 1183503614 M * Bertl be my guest :) 1183503677 J * stefani ~stefani@c-24-19-46-211.hsd1.mn.comcast.net 1183503735 M * Bertl wb stefani! 1183503774 M * daniel_hozac should i remove everything about vc_[gs]et_iattr_v0 as well? doesn't seem to be in the kernel. (from the headers, that is) 1183503859 M * Bertl check out the OLPC branch in this regard, I think I removed it there already 1183503862 M * stefani Bertl: hit by car broken bones coiuld have been much worse 1183503867 Q * sladen Ping timeout: 480 seconds 1183503881 M * Bertl stefani: eeek! hospital? 1183503891 M * Bertl stefani: I mean, right now? 1183503914 M * stefani Bertl: yesterday out of hospital now :) 1183503915 J * sladen paul@starsky.19inch.net 1183503941 M * daniel_hozac stefani: damn, i'm sorry. what bones did you break? 1183503959 N * DoberMann DoberMann[ZZZzzz] 1183503999 M * stefani hand 1183504019 M * stefani hard to type with hands in casts 1183504023 M * stefani :) 1183504029 M * Bertl stefani: well, you are lucky ... remember lilo? 1183504051 M * stefani LILO... of course 1183504058 M * Bertl (the guy who did start freenode :) 1183504090 M * stefani guess not thst lilo :\ 1183504113 M * Bertl he had a car accident too (was driving a bike), a few months ago, died back then ... sad story :/ 1183504172 M * Bertl well, you can learn from that ... make damn sure that you do what you really want to do .. right now :) 1183504229 M * Bertl anyway, hope you will recover soon ... 1183504231 M * daniel_hozac Bertl: sure you want to remove the id argument? it'll require a copy of the __COMPAT macro in the switch. 1183504262 M * Bertl hmm, that's probably the reason why we left it there ... 1183504264 M * daniel_hozac indeed, stefani, get well. 1183504270 M * Bertl let me think about that for a moment 1183504274 M * slack101 Bertl, i talked to him 1-2 days before that happened 1183504277 M * daniel_hozac yeah, i'd assume so. 1183504311 M * Bertl slack101: yeah, me too, he was always around on this channel 1183504316 M * slack101 o? 1183504328 M * slack101 i dont know what these bikers think 1183504336 M * slack101 government says yea bike on the road 1183504338 M * slack101 so they do 1183504347 M * slack101 and 1-10 of every 10 cars got to swirve to miss them 1183504366 M * slack101 and in morning and at night 1183504370 M * slack101 even harder 1183504383 M * slack101 i dont care what the government said i would be on the sidewalk 1183504434 M * Bertl where you would driver over harmless pedestrians :) 1183504444 M * slack101 lol 1183504450 M * slack101 theres hardly anyone on the sidewalk 1183504463 M * slack101 maybe in downtown districts but yea man 1183504468 M * daniel_hozac that depends on the sidewalk, no? 1183504472 M * Bertl because they know you are out there to get em' 1183504476 M * daniel_hozac haha. 1183504476 M * slack101 i dotn think on a busy road with idiots going 50 mph on a 35 is good either 1183504511 M * slack101 he was older too 1183504519 M * slack101 and not sure of his physcal condition 1183504539 M * slack101 but a wack with a car when your on a bike cant be good 1183504561 M * slack101 and i jus know he was prolly on a busy street 1183504622 M * slack101 but yea i dont think he liked me to well either 1183504626 M * slack101 BUT RIP 1183504732 M * slack101 lol what were those russians thinking 1183505158 Q * oxylin Quit: Ex-Chat 1183505275 Q * stefani Read error: No route to host 1183505651 M * trippeh_ I've witnessed so many bycycle accidents involving cars lately I never cycle on the road unless absolutely necessary 1183505722 M * trippeh_ Few weeks ago my dad got hit'n run'ed by a car in the bicycle lane, too 1183506048 M * slack101 trippeh_, proves my point ;) 1183506257 M * trippeh_ I'm doing on average 10 miles a day, so I guess I'm very likely to get into somthing nasty one time or another ;) 1183506279 M * slack101 trippeh_, stay off the road 1183506282 M * slack101 where you live ? 1183506317 M * trippeh_ Norway, near the capital (Oslo) 1183506331 M * slack101 do you wear a fur coat ? 1183506405 M * slack101 :P 1183506411 M * slack101 Norway to me was always super cold 1183506423 M * trippeh_ This time a year? Not likely :) Hit 90F a few weeks ago (then again, 3 days later it snowed ...) 1183506443 M * slack101 snowed ? 1183506444 M * slack101 wtf 1183506464 M * slack101 whats snow like ? 1183506469 M * slack101 i need to go see it sometime 1183506508 M * trippeh_ Winter can get really cold here, but summer in general is not too bad, even up north. 1183506514 M * trippeh_ Uhm, its cold and white ;) 1183506548 M * slack101 i need to go see snow 1183506558 M * slack101 i got to drive pretty far to see it though 1183506718 M * trippeh_ And no, no ice bears on the mainland :P 1183506727 M * slack101 lol 1183506738 M * slack101 i wonder what they did 100 years ago to keep warm 1183506744 M * slack101 man that would have been hard 1183506762 A * trippeh_ guesses fire :] 1183506774 M * slack101 lol 1183506774 M * slack101 man 1183506779 M * slack101 that would have been hard in the winter 1183506936 M * trippeh_ The family cottage we have in the mountains is heated by fire only, no electricity, gas or other source of heat 1183506989 M * trippeh_ Laptops doesnt like beeing booted in -30C and lower, I found out once :> 1183507005 M * slack101 lol 1183507007 M * trippeh_ Weird noises... 1183507008 M * slack101 do you freeeze ? 1183507090 M * trippeh_ Can be a bit cold in the morning if nobody was up feeding the fire, but other than that its all good. 1183507108 M * slack101 do you have ot mess with the fire every few mins ? 1183507112 M * slack101 or every hour ? 1183507189 M * trippeh_ Every 1-2 hours I guess