1145145658 M * ray6 daniel: ah, OK, I'm building inside a minimalistic installed xen DomU :) 1145145689 M * daniel_hozac FYI, i'll have -14 RPMs ready in a few minutes. 1145145750 M * ray6 that's .30.210 with your 20 patches? 1145145760 M * daniel_hozac yes. 1145145778 M * ray6 just built -5 through :) 1145145789 M * derjohn daniel_hozac, are the 20+ set fc specific? 1145145812 M * daniel_hozac derjohn: 1 is Fedora/CentOS specific, 1 is Fedora Core 5 host and guest support. 1145145841 M * derjohn ah, fc5 runs now as guest without hazzles? 1145145844 M * daniel_hozac then quite a few of the patches come from the Debian and Gentoo packages. 1145145866 M * derjohn but they didnt make it into ensc's upstream release? 1145145877 M * daniel_hozac none of these are in ensc's tree. 1145145884 M * daniel_hozac he hasn't updated it since 0.30.210 was released. 1145145915 M * daniel_hozac (at least not the public copy) 1145145922 M * derjohn well, it get time someone get CVS access there (/me thinks of daniel_hozac as candidate ;)) 1145145931 M * derjohn *write access 1145145981 M * Bertl no problem with that, don't know how savannah handles this though 1145146080 M * derjohn Bertl, you wave write access? Or could you grant write access? (it's mess that all distri-maintainers have to get the secret patches ....) 1145146095 M * Bertl I don't think so 1145146145 M * daniel_hozac ensc is the only one with access to the savannah project, AFAIK. 1145146191 A * ray6 now has FC4 in Dom0, FreeBSD 7 in an DOMU and another DomU running Centos4 with a vserver kernel in which I'm bootstraping a debian guest right now :) 1145146204 M * daniel_hozac ray6: hmm, scratch that, i've got no CentOS roots setup. 1145146205 M * derjohn daniel_hozac, did anyone ever ask to et write access? or isnt that something you'd like to have .. I mean to patch it anyway 1145146238 M * daniel_hozac i wouldn't know. Enrico has been very responsive in the past, i guess he's just swamped with work. 1145146248 M * Bertl derjohn: IIRC the last time somebody (sam) branched the tools into a public repository 1145146266 M * Bertl but the commits there didn't last very long 1145146269 M * derjohn daniel_hozac, yes, no offense to enrico, he did great work! 1145146285 M * ray6 daniel: what do you mean? scratch the image? I've got a cent43minimal.tar.gz :) 1145146296 M * daniel_hozac ray6: mock configs. 1145146385 M * derjohn well at least a "progressive branch" of the utils would ne nice so see ... (besides that they should drop that silly 'alpha' status addon) 1145146402 M * ray6 daniel: hm? 1145146422 M * daniel_hozac derjohn: that's what HEAD is for. 1145146443 M * derjohn yes, I ensc would give you access to HEAD ;) 1145146443 M * daniel_hozac ray6: mock is a "rebuild source RPM in chroot" helper. 1145146466 M * derjohn I thought of a parallel branch but head would be cool, too ;) 1145146474 M * daniel_hozac i don't really mind keeping my work as patches. 1145146495 M * daniel_hozac it's easier to build RPMs from a bunch of patches than from a CVS checkout :) 1145146511 M * h01ger munin now produces graphs for number of processes and virtual memory used :) 1145146549 M * derjohn daniel_hozac, well, sure, but how will the patches go into uptream? Do you commeit them on savannah? 1145146557 M * daniel_hozac ray6: http://rpm.hozac.com/dhozac/fedora/5/vserver/SRPMS/util-vserver-0.30.210-14.fc5.src.rpm 1145146562 M * derjohn (or send them to enrico) 1145146576 M * daniel_hozac derjohn: well, the larger ones already have a patch entry at savannah's tracker. 1145146593 M * ray6 is there something fc5 specific in that SRPM? I'm using C4 as the host currently 1145146608 M * daniel_hozac ray6: no, the sources are all the same. 1145146762 M * ray6 procfs-security. Please read the FAQ for more details 1145146769 M * ray6 /proc/uptime can not be accessed. Usually, this is caused by 1145146771 M * daniel_hozac service vprocunhide start ;) 1145146771 M * ray6 hmmm 1145146795 M * ray6 I should reboot after installing the vserver utils? :) 1145146816 M * daniel_hozac no, just starting the vprocunhide service is sufficient 1145146822 M * ray6 (or "should have ..." :) 1145146934 M * ray6 ok, looks good, got the debian guest running 1145147086 M * derjohn secure-mount: chdir("/dev/pts"): No such file or directory -> was this one of the current issues with the utils? 1145147092 M * ray6 oh, running short of memory in the guest's host. 64MB and no swap there %-) 1145147137 M * daniel_hozac derjohn: no, but that should've been created by the vserver ... build. 1145147179 A * derjohn slaps himself for copying a complete physical udev-ed server in ..... 1145147520 M * shuri lol 1145147554 M * Bertl daniel_hozac: did the source mapping help/work for you? 1145147565 M * daniel_hozac Bertl: i haven't tested it yet. 1145147634 Q * shuri Remote host closed the connection 1145148015 Q * Wenix Remote host closed the connection 1145148194 P * matta 1145148668 M * Bertl daniel_hozac: cool, the osdl results finished 1145148695 M * Bertl seems there are no great differences except for the sparse run which ahs 18160 warnings :) 1145148719 M * daniel_hozac haha, wow. 1145148769 M * Bertl but you should look at the output, looks really cool 1145148820 M * daniel_hozac yeah, that's so much better than the other ones. 1145148873 M * Bertl but I think the sparse run was just not done on mainline yet 1145148985 M * Bertl or maybe an unfortunate run, as the 2.0.2 passed the sparse 1145149004 M * Bertl ah, no, sorry, wrong tab :) 1145149028 M * nebuchadnezzar derjohn: you can check the http://www.asgardr.info/twiki/bin/view/Main/MasteringVservers to look at --exclude --include stuff :-) 1145151289 M * derjohn nebuchadnezzar, yes, looks good. even with mtr-tiny :) I will try with my next build ! 1145151346 M * nebuchadnezzar ok 1145151361 M * nebuchadnezzar I have some vxW: xid=400 did hit the barrier 1145151401 M * nebuchadnezzar is there a way to make all the barrier stuff clean ? 1145151403 M * Bertl that means somebody tried to manipulate a dir with the barrier set 1145151419 M * Bertl vattrib --~barrier 1145151429 M * Bertl probably recursive and such, see --help 1145151439 M * Bertl vattr 1145151464 M * Bertl nonsense 1145151471 M * Bertl setattr 1145151481 M * Bertl setattr -R --~barrier 1145151498 M * nebuchadnezzar on the host / ? 1145151513 M * nebuchadnezzar and then setattr --barrier /var/lib/vservers 1145151515 M * Bertl well, wherever you suspect invalid barriers 1145151530 M * Bertl no, you want: setattr --barrier /var/lib/vservers/guest/.. 1145151542 M * Bertl note that the '..' is literal 1145151550 M * nebuchadnezzar arf 1145151906 M * derjohn anyone tried hylaFax with ISDN/CAPI within a guest? 1145151932 M * Bertl IIRC, folks got asterisk running, so that should be almost trivial :) 1145151977 A * derjohn @TODO: build asterisk guest 1145152102 M * derjohn ERROR: CAPI not installed, started or have no access rights on it! -> seems it uses kernel calls to the blobby AVM Fritz module, which it isnt allowed within a context 1145152132 A * Bertl is very pleased that the gcc 4.1 compiles showed no new warnings ... 1145152132 J * aRko^ aRko_@85.158.39.179 1145152139 M * Bertl welcome aRko^! 1145152259 M * derjohn n8 folks! 1145152267 M * Bertl good night derjohn! 1145152278 M * gdm hia 1145152290 M * gdm Bertl: did u get a chance to look at that page yet? 1145152308 M * Bertl well, chance yes, but I forgot :/ 1145152317 M * gdm ;-) 1145152328 M * Bertl will do so in a few minutes :) 1145152335 Q * aRko^ Quit: 1145152366 M * gdm ok, kool. the picture, btw, is mainly for some color and not because it is specifically needed 1145152392 M * Bertl we have a picture? :) 1145152432 A * Bertl goes looking ... 1145152432 M * gdm well, diagram 1145152436 M * nebuchadnezzar Bertl: where can I find explanation of the attribute flags ? 1145152464 J * matt1 ~matta@c-68-81-35-243.hsd1.pa.comcast.net 1145152473 M * Bertl nebuchadnezzar: you mean for setattr? 1145152478 M * nebuchadnezzar yes 1145152485 M * Bertl gdm: nice, colorful picture indeed :) 1145152486 M * nebuchadnezzar and showattr 1145152511 M * gdm Bertl: hehe :) 1145152531 M * gdm stolen from another website, of course (there was no copyright notice displayed) 1145152534 M * Bertl nebuchadnezzar: I think there is a howto/explanation regarding the proc hiding stuff 1145152568 M * Bertl that should cover admin, watch and hide 1145152576 M * nebuchadnezzar ok 1145152586 M * Bertl and then the unification stuff, which covers iunlink 1145152596 M * Bertl the iunlink-but-not-immutable is a special case 1145152608 M * Bertl and the barrier itself, which you already know 1145152832 M * nebuchadnezzar ok so only /var/lib/vservers has the B 1145152855 M * Bertl if that is identical to /path/to/guest/.. then yes :) 1145152861 M * Bertl (for all guests) 1145152872 M * nebuchadnezzar yes, for all guests 1145152921 M * nebuchadnezzar so guest can not escape from /var/lib/vservers, but what about escaping their root to go in another vserver root ? 1145153047 M * Bertl that's not possible without the help from outside 1145153054 M * nebuchadnezzar ok 1145153066 M * Bertl well, at least as far as we know :) 1145153305 Q * matt1 Ping timeout: 480 seconds 1145153540 M * nebuchadnezzar Bertl: vxW: xid=1 did hit the barrier <-- xid=1 is not a xid of a vserver 1145153555 M * Bertl that's the spectator context 1145153570 M * nebuchadnezzar as I read yes 1145153572 M * Bertl the one used for vtop or vps 1145153593 M * Bertl that should not happen unless you use the test scripts or so? 1145153606 M * nebuchadnezzar each time i make a vps I get a hit the barrier log 1145153626 M * Bertl hmm, maybe the tools use that to check the barrier? 1145153795 M * nebuchadnezzar vtop do not produce one 1145153817 M * nebuchadnezzar arf 1145153823 M * nebuchadnezzar vps afx produce one 1145153828 M * nebuchadnezzar not vps alone 1145153850 M * nebuchadnezzar which only give my current process 1145153914 M * nebuchadnezzar ok, thanks a lot 1145153919 M * nebuchadnezzar now it's time to sleep for me 1145153927 M * Bertl interesting, why would vps hit the barrier? 1145153943 M * nebuchadnezzar the a 1145153945 A * ray6 also should go to sleep 1145153965 M * nebuchadnezzar Bertl: it looks for all vservers 1145153994 M * nebuchadnezzar ok, so good night 1145153999 M * ray6 Bertl: cu tomorrow... 1145154004 M * Bertl yeah, but why should it be trying to modify something there ... 1145154013 M * Bertl okay, cya ray6! night nebuchadnezzar! 1145154278 M * knotty Bertl: it seem that 'vps a' hit the barrier only when I am in a guest 1145154372 M * Bertl knotty: ah, that could be 1145154380 M * gdm Bertl: here's another page, too: http://linux-vserver.org/Memory+Allocation 1145154393 M * gdm Bertl: i'd really appreciate comments on that one! 1145154414 M * gdm it is the more practical, i think, rather than the theoretical knowledge behind it 1145154520 M * Bertl hmm, the what is what is wrong :/ 1145154534 M * Bertl VM has nothing to do with RAM 1145154557 M * Bertl VM is the sum of all virtual pages inside the guest 1145154571 M * Bertl VML is the sum of all virtual pages locked into memory 1145154587 M * Bertl RSS is the number of pages currently present in RAM 1145154599 M * Bertl ANON is the number of anonymous pages 1145154612 M * Bertl SHM is the shared memory (ipc) 1145154626 M * Bertl the rest is not relevant for memory 1145154708 M * gdm that is the memory allocation page, right? 1145154720 M * Bertl yep 1145154739 M * knotty Bertl: if I am in 2 guests and I do 'vps a' on the host, that hit the barrier 2 times 1145154782 M * Bertl knotty: you sure that you don't have a barrier somewhere else? 1145154836 M * Bertl knotty: let me check the source ... 1145154962 M * knotty I don't know if I have barrier somewhere else, nebuchadnezzar did it 1145155361 M * gdm Bertl: ok, i've updated that page now 1145155395 M * Bertl knotty: hmm, it might actually happen during a normal permission check when traversing the barrier 1145155450 M * Bertl knotty: will check that and when I can confirm that, we'll add the WATCH to the exceptions for that check 1145155497 M * Bertl fs/namei.c ~234 in dx_permission() 1145155523 M * Bertl there is VX_ADMIN and probably there should be VX_ADMIN|VX_WATCH 1145155568 Q * softi42 Ping timeout: 480 seconds 1145155799 M * knotty ok thanks Bertl 1145156188 J * softi42 ~softi@p549D491F.dip.t-dialin.net 1145159055 M * micah besides what appears in vserver-stat and what is in /etc/vservers//context, is there any other way to map a vserver name to its xid? 1145159110 M * Bertl the name is purely userspace 1145159117 M * micah vserver-info XID 1145159121 M * micah apparantly does it too 1145159129 M * micah only if its static 1145159149 M * Bertl as I said, the kernel is not involved here 1145159151 M * micah it probably just parses the context file :) 1145159171 M * daniel_hozac micah: vserver-info ... CONTEXT true IIRC. 1145159232 M * micah yeah vserver-info ... CONTEXT doesn't parse the context file, wonder how it gets it 1145159308 M * daniel_hozac it does, that's what the true is for. 1145159319 M * daniel_hozac but first, it asks the kernel. 1145159328 M * micah utilvserver_fmt_long 1145159483 M * daniel_hozac micah: btw, did i tell you about http://daniel.hozac.com/vserver/util-vserver/? 1145159540 M * micah daniel_hozac: I dont think so, how is this different from the CVS? 1145159564 M * micah (besides it being a lot easier for me to figure out when things are newer :) 1145159565 M * daniel_hozac well, CVS didn't have most of the patches until a few hours ago ;) (but it does now, so i guess the point is moot) 1145159620 M * micah so whats new in here? Last I knew was the vcontextuid patch was updated 1145159663 M * daniel_hozac well, it's got all the Debian package patches, and some of the Gentoo ones. 1145159673 M * daniel_hozac plus some easy fixes for bugs in savannah. 1145159736 M * daniel_hozac (apis, condrestart, cpuset, nice, prefix, remove-init-style-gentoo, start-vservers, usage should be the new ones) 1145159942 A * micah starts pulling in patches :) 1145159968 M * micah what is fixed by the apis? 1145160015 M * daniel_hozac typo and letting you specify oldproc and olduts for --enable-apis. 1145160418 M * Bertl okay, I'm off to bed for tonight ... 1145160425 M * micah goodnight Bertl ! 1145160428 M * Bertl have a good one everyone ... cya tomorrow! 1145160433 N * Bertl Bertl_zZ 1145160434 M * daniel_hozac good night! 1145160643 J * TheGame TheGame@203.184.13.101 1145160648 M * micah daniel_hozac: what does the prefix patch fix? it looks like it modifies configure 1145160677 M * micah but configure and configure.ac syntax always boggles me 1145160755 M * daniel_hozac vserver ... build -m yum breaks when you configure with --prefix=/ rather than --prefix=. so the patch makes --prefix=/ == --prefix= 1145160823 M * micah thanks 1145160994 M * micah I had part of the start-vservers patch (the okfile fix) 1145161000 M * daniel_hozac yeah, i know. 1145161022 M * micah whats the MARK_ANY add? 1145161041 M * daniel_hozac start-vservers --all --stop will actually stop all vservers, not just unmarked ones. 1145161059 M * micah oh yeah, i just ran into that and was going to look into it :) 1145161074 M * micah i guess that solves that problem 1145161086 M * daniel_hozac it did in my tests. 1145161372 M * micah I dont think I pulled in the fc5 patch either 1145161550 M * micah which appears to make installing fedora guests work more correctly? 1145161557 M * micah i dont really understand redhat 1145161583 Q * TheGame Quit: LOL. Owned by stirk 1145161583 M * daniel_hozac the fc5 patch lets you install Fedora Core 5 guests, and use it as a host. 1145161728 J * TheGame TheGame@203.184.13.101 1145161750 M * micah lots of changes to this version 1145162196 M * micah daniel_hozac: thanks for pulling those patches together, I've just uploaded a new version of the .deb with those fixes added that were missing 1145162854 J * Zaki ~zaki@212.118.96.168 1145164592 P * TheGame 1145165899 J * reedtv ~wangchunh@219.142.143.184 1145165911 P * reedtv 1145166183 M * Zaki hi 1145167339 Q * Zaki Quit: Leaving 1145167374 J * ZLinux ~ZLinux@212.118.96.168 1145169680 Q * knotty Quit: Parti Ailleurs 1145171355 J * burak ~a@81.215.154.218 1145171388 P * burak 1145172386 J * Viper0482 ~Viper0482@p54975DA3.dip.t-dialin.net 1145174300 Q * Viper0482 Ping timeout: 480 seconds 1145174459 J * lilalinux__ ~plasma@dslb-084-058-212-040.pools.arcor-ip.net 1145174835 Q * lilalinux_ Ping timeout: 480 seconds 1145174940 J * Viper0482 ~Viper0482@p5497714B.dip.t-dialin.net 1145177168 Q * bogna Quit: 1145179319 M * nebuchadnezzar hello 1145179348 M * h01ger huhu :) 1145179562 M * nebuchadnezzar are the procs and mount related boot script usefull in a vserver ? 1145179652 M * nebuchadnezzar arf, fstab is empty so no 1145181268 M * h01ger hmmmm... vserver-stat shows the user and the systime used, but i cant find that info in /proc looking at http://linux-vserver.org/HowTo+Read+ProcFS - any pointers? 1145181275 A * h01ger is improving his muninplugins 1145181681 M * mnemoc did you run procunhide? 1145181729 M * h01ger no 1145181762 M * h01ger i can find the infos mentioned in that wiki page in /proc/virtual, but i cant find some info that vserver-stat shows.. 1145183999 Q * FireEgl Quit: Bye... 1145185183 J * Dr4g ~Dr4g@80-195-133-218.cable.ubr06.uddi.blueyonder.co.uk 1145185788 Q * mnemoc Ping timeout: 480 seconds 1145186680 M * nebuchadnezzar is here a way to make some cleanup in the .hash ? 1145186717 M * nebuchadnezzar is: find /var/lib/vservers/.hash -type f -links 1 sufficient ? 1145186804 M * nebuchadnezzar or should I rm -rf the directory and rehashify all vservers ? 1145187102 Q * Viper0482 Ping timeout: 480 seconds 1145187229 M * daniel_hozac i use the find command. 1145187294 M * h01ger does anyone use fail2ban or denyhosts with vservers? 1145187530 M * nebuchadnezzar daniel_hozac: ok, and you rm 1145187548 M * daniel_hozac right, -exec rm -f {} \; 1145187567 M * nebuchadnezzar is should be integrated to vhashify :-) 1145187587 J * Viper0482 ~Viper0482@p54976199.dip.t-dialin.net 1145187592 M * nebuchadnezzar daniel_hozac: is there a way to know who is hard linked to a file ? 1145187599 M * daniel_hozac well, it's not really part of vhashify's job. 1145187647 M * daniel_hozac well, ls -li and find -inum -xdev? 1145187657 M * daniel_hozac or maybe just find -samefile 1145187681 M * nebuchadnezzar it's not possible that 2 files in .hash are hard-links 1145187691 M * nebuchadnezzar a sort of dead-lock in the .hash 1145187711 M * daniel_hozac no, the same file will always generate the same hash. 1145187719 M * nebuchadnezzar arf, sure 1145187722 M * daniel_hozac that's sort of the point of the hashing :) 1145187727 M * nebuchadnezzar so I rm all -links 1 1145187734 M * daniel_hozac right. 1145187909 M * Hollow :) 1145187916 M * Hollow daniel_hozac: vcd has TLS suport :) 1145187926 M * Hollow gnutls is very nice 1145187934 M * nebuchadnezzar yes 1145187936 M * daniel_hozac Hollow: i thought you weren't going to implement that :) 1145187940 M * nebuchadnezzar gnutls is good 1145187947 M * Hollow well, it was so freaking easy 1145187948 M * Hollow :P 1145187954 M * Hollow with gnutls 1145187959 M * nebuchadnezzar and what is vcd ? 1145187959 M * Hollow i love it :D 1145187966 M * Hollow vserver control daemon 1145187977 M * nebuchadnezzar ok 1145187990 M * nebuchadnezzar it there some doc somewhere ? draft , 1145187990 M * nebuchadnezzar ? 1145187993 M * Hollow http://home.xnull.de/work/vserver/vcd.spec.html 1145188025 M * daniel_hozac is the login needed then? 1145188034 M * Hollow yes.. 1145188055 M * Hollow this is just _transport layer_ security 1145188073 M * Hollow i still need a username for internal vcd acls 1145188081 M * daniel_hozac yes, but you can use the email address of the client certificate for that. 1145188098 M * nebuchadnezzar the CN in fact 1145188100 M * Hollow hm.. which client certificate? 1145188107 M * nebuchadnezzar Hollow: for the TLS connection 1145188113 M * Hollow i don't have one 1145188118 M * nebuchadnezzar auth by client cert 1145188119 M * daniel_hozac but you could. 1145188119 M * Hollow and i don't need one for ssl sites :o 1145188141 M * nebuchadnezzar Hollow: vcd is developped in C/C++ ? 1145188142 M * daniel_hozac TLS allows mutual verification. 1145188156 M * Hollow then every client would need a cert.. what mess... it's already enought to create certs for the server... tbh that was the hardest part in the tls implementation :o 1145188177 M * daniel_hozac lol 1145188180 M * Hollow you should know that i hate all this crypto stuff 1145188180 M * nebuchadnezzar arf 1145188213 M * nebuchadnezzar Hollow: but it so freaking easy, remember: you love it :-) 1145188213 M * Hollow but yes... i see that there could be dumb admins not protecting their vcd and such... so at least minimum tls.. 1145188214 M * Hollow :P 1145188232 M * Hollow the spec still says no TLS for vcd 1145188233 M * Hollow lol 1145188253 M * Hollow no, i don't want to rely on client certs 1145188275 M * daniel_hozac why not? 1145188349 M * daniel_hozac it's far more secure than a username and password. 1145188373 M * Hollow see above :) 1145188438 M * Hollow (i'm just too lazy to get used to tls, ok? :) 1145188450 M * daniel_hozac :P 1145188486 M * Hollow i also just bugfixed the TLS implementation.. someone gave me a patch for it.. ;) 1145188826 J * ciphernaut ~Elive_use@dsl-58-6-115-82.qld.westnet.com.au 1145188966 M * nebuchadnezzar daniel_hozac: what do you think about ard-links with less than (num of vserver) +1 ? 1145189063 M * daniel_hozac well, assuming all of your vservers don't have the same software installed, that's perfectly legit. 1145189072 M * daniel_hozac (and the same distribution) 1145189577 J * Dr4g_ ~Dr4g@80-195-133-218.cable.ubr06.uddi.blueyonder.co.uk 1145189614 Q * Viper0482 Ping timeout: 480 seconds 1145189725 Q * phedny Ping timeout: 480 seconds 1145189913 J * Viper0482 ~Viper0482@p549750BF.dip.t-dialin.net 1145190013 Q * Dr4g Ping timeout: 480 seconds 1145190069 N * Bertl_zZ Bertl 1145190073 M * Bertl morning folks! 1145190210 J * phedny ~mark@volcano.p-bierman.nl 1145190469 M * nebuchadnezzar morning Bertl 1145190569 M * h01ger moin moin Bertl - are you reading backlog(s) or should i rephrase my questions ? :) 1145190590 M * nebuchadnezzar does /etc/vservers/.default/apps/vuniy/exclude add to the default maybe built-in list or does it completly remplace it ? 1145190661 M * derjohn Bertl, you didnt comment http://linux-vserver.org/ChangeLogExperimental that yet - does this reflect the truth? 1145190681 M * Bertl h01ger: it's simpler if you repeat/rephrase it :) 1145190688 M * daniel_hozac nebuchadnezzar: completely replace it. 1145190690 M * Bertl h01ger: but yes I do read them ... 1145190712 M * nebuchadnezzar ok, I understand now 1145190724 M * daniel_hozac h01ger: AFAICT, that's just the sum of all currently running processes, not something you should rely on at all. 1145190728 M * nebuchadnezzar if I want to add some I need to do it locally to each vservers 1145190780 M * nebuchadnezzar Is there a way to know the default list ? 1145190798 M * Bertl derjohn: well, the lines you did put there are from me (or?), so they should be fine 1145190824 M * Bertl derjohn: will add the changes between rc6 and rc15 later ... 1145190845 M * daniel_hozac nebuchadnezzar: /usr/lib/util-vserver/defaults/vunify-exclude 1145190867 M * h01ger daniel_hozac, isnt " cat /proc/virtual/$i/status |grep Tasks|cut -f2 " the number of processes on vserver $i ? 1145190869 M * nebuchadnezzar ok thanls 1145190875 M * derjohn misunderstanding: I meant the text i put in front of the changelogs and the idea to split between stable-exp and devel-exp 1145190902 M * daniel_hozac h01ger: yes. 1145190916 M * Bertl derjohn: ahh, well, I just got up :) 1145190917 M * h01ger Bertl, i want to know the milli-secs used (usertime, systime per vserver) as vserver-stat shows, but cant find it in /proc/virtual 1145190918 M * derjohn If the is correct I think that I now understood the release cycling 1145190934 M * h01ger daniel_hozac, so, that number is not reliable ? in what sense ? 1145190992 Q * ciphernaut Remote host closed the connection 1145191040 M * ray6 moring Bertl 1145191062 M * Bertl morning ray6 1145191089 M * Bertl h01ger: no such information is available, but you can get the user and sys ticks (jiffies) from /proc/virtual/42/sched 1145191099 M * Bertl cpu 0: 3 5 0 0 0 R- 25 6 50 1/4 1/8 1145191099 M * Bertl cpu 1: 0 1 0 0 0 R- 25 6 50 1/4 1/8 1145191130 M * h01ger cpu 0: 34383 18065 0 1145191134 M * h01ger :) 1145191156 M * Bertl see that's user,sys, and hold time 1145191164 M * h01ger hold ? 1145191170 M * Bertl okay, off for lunch ... 1145191177 M * h01ger (and why do you have more numbers?) 1145191181 M * h01ger Bertl, bon appetite! 1145191188 M * Bertl because I use a devel kernel? 1145191194 N * Bertl Bertl_oO 1145191202 M * h01ger Bertl_oO, sure, what numbers are those ? 1145191350 M * derjohn h01ger, bertl fiddled with scheduler in the lastest release. the scheduler file has now even two more number. and there is schedmon to watch the scheduler from outside the kernel. i _suspect_ those number comming from the new approch .. 1145191394 M * h01ger ah 1145191439 M * derjohn i think the wiki is updated already .... and the old format is still compatible if you upgrade 1145191461 M * derjohn hm, how can I find out which version of VS a running kernel has? 1145191515 M * derjohn h01ger, with your munin plugin: do you also count if a guest hits a limit ? 1145191587 M * derjohn h01ger, I mean from the host side in /proc/virtual//limit 1145191603 M * daniel_hozac derjohn: /proc/virtual/info? 1145191621 M * daniel_hozac (won't tell you exactly, but will give you a pretty good idea) 1145191641 M * derjohn daniel_hozac, hooray it a .... VCIKernel: 031101f6 .... foo? :) 1145191666 M * daniel_hozac not that one, the VCIVersion. 1145191673 M * daniel_hozac VCIKernel is what features are enabled. 1145191721 M * derjohn daniel_hozac, I would file a wish the the proc tells somethign like ct /proc/virtual/vsversion -> 2.1.1-rc28 ... my value is VCIVersion: 0002:0001 1145191762 M * daniel_hozac well, that's why there's a Makefile patch ;) 1145191772 M * daniel_hozac uname -r should contain -vs... 1145191795 M * derjohn h01ger, what VS version do sid's vserver-k7 kernels use? It's not very documented in /usr/share/doc 1145191843 M * derjohn daniel_hozac, yes, _my_ kernels do so: 2.6.15-amd64-vs2.1.0.5.1 , Debian says 2.6.16-1-vserver-k7 1145191888 M * h01ger derjohn, i'm not monitoring hits atm, just started last nite, but i'll note the idea.. 1145191905 A * h01ger starts a todo and creates a _directory_ for this little project :) 1145191937 M * daniel_hozac derjohn: ask the changelog? 1145191946 M * derjohn daniel_hozac, but it find it questionable to add it to the uanme, if you add vserver, xen, foo and bar you'll need a triple head to see what kernel is running :) 1145192003 M * derjohn daniel_hozac, Debian has that in /usr/share/doc/packagename.... but the dont mention all patches. maybe in the src-package of the kernel, i'll download it :) 1145192485 M * daniel_hozac derjohn: FYI, http://svn.debian.org/wsvn/kernel/releases/linux-2.6/2.6.16-7/debian/patches/?rev=0&sc=0 1145192523 M * derjohn daniel_hozac, hehe ... you are steadily moving towards Debain ;) 1145192557 M * daniel_hozac not at all :) 1145192766 M * daniel_hozac h01ger: kernel/vserver/sched_proc.h:vx_info_proc_sched_pc shows you what those numbers are. 1145192931 M * h01ger do you have a http-url ? i dont have the source atm... 1145192934 M * h01ger ;) 1145192965 M * daniel_hozac well, it's in the patch, so http://vserver.13thfloor.at/Experimental/ 1145192993 M * h01ger ok :) 1145193197 M * derjohn daniel_hozac, not at all? BTW: my new vserver host is called 'hozac' .. the next one will maybe be enrico ... :) 1145193212 M * daniel_hozac lol 1145193239 M * h01ger ok, for the moment i'll care more about sched_proc.h in http://www.13thfloor.at/vserver/s_rel26/v2.01/patch-2.6.14.3-vs2.01.diff but its good to understand. thanx daniel ! 1145193493 M * derjohn daniel_hozac, and: it's a debian host :) 1145193643 M * Hollow phreak``: ping 1145193814 Q * Dr4g_ Read error: Connection reset by peer 1145194487 Q * derjohn Remote host closed the connection 1145194829 N * insomnia1 insomniac 1145195946 J * ntrs_ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1145195946 Q * ntrs Read error: Connection reset by peer 1145196233 J * ntrs__ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1145196640 Q * ntrs_ Ping timeout: 480 seconds 1145198022 J * matta ~matta@c-68-32-239-173.hsd1.pa.comcast.net 1145198238 N * Bertl_oO Bertl 1145198241 M * Bertl back now .. 1145198416 M * ZLinux wb Bertl :) 1145198563 M * Bertl hey ZLinux! 1145199166 M * h01ger hey Bertl - what does hold in /p/v//sched mean ? like iowait ? 1145199189 M * Bertl no, that's the ticks the hard cpu scheduler did put the context on hold 1145199199 M * Bertl i.e. when the token bucket is empty< 1145199249 M * Bertl s/ remember, networking is IP based, not interface based. <-- sure, but a vlan should solve this, all vserver can have it's own private 127.0.0.1, AFAIK 1145220150 M * nebuchadnezzar I must test 1145220550 M * daniel_hozac nebuchadnezzar: so how would 127.0.0.1 differ from, say, 127.0.0.1? 1145220619 M * nebuchadnezzar well, with real lan, 2vlan can have the same subnet 1145220636 M * nebuchadnezzar 2 same IPs are different, depending on the vlan 1145220820 M * bonbons nebuchadnezzar: and how is it decided to which vlan to associate a socket bound to say 127.0.0.1? 1145220880 M * bonbons an app will not say "I want the IP on vlan XYZ"! 1145220887 M * nebuchadnezzar I was thinking that it was context specific 1145220912 M * nebuchadnezzar I assign the dummy0.1 to one vserver and dummy0.2 to another 1145220942 M * bonbons there is no way to associate a vlan/net device to a context until there is ngnet. 1145220952 M * nebuchadnezzar :-/ 1145220994 M * bonbons the only thing util-vserver does is configure an interface for you (if you request it to do so). But kernel never sees more of it that the pure IP address 1145221046 M * nebuchadnezzar erf, and when ngnet will arrive ? ;-) 1145221078 M * bonbons ask Bertl and ebiederm for that 1145221080 M * daniel_hozac as I see it, the next few steps will be: 1) a 2.0.2 and 2.1.1 release, 2) per context quota, 3) ipv6 in current networking mode, 4) ngnet 1145221093 M * nebuchadnezzar ok 1145221121 J * knotty ~void@fny94-1-82-67-169-3.fbx.proxad.net 1145221127 M * nebuchadnezzar :-) 1145223185 Q * Dr4g Read error: Connection reset by peer 1145223207 J * Dr4g ~Dr4g@80-195-133-218.cable.ubr06.uddi.blueyonder.co.uk 1145224133 J * FireEgl Atlantica@Atlantica.US.TO 1145224952 Q * bonbons Quit: Leaving 1145226708 Q * goldnlink Read error: Connection reset by peer 1145227149 J * shedi ~siggi@cpe-24-165-167-167.midsouth.res.rr.com 1145228481 M * orzel I'm the happy user of a multi-vhost apache+php server running inside a vserver 1145228495 M * orzel the main server has iptables rules to forward port 80 to the vserver 1145228500 M * orzel using -t nat 1145228538 M * orzel Now. where it comes to fun, is that my squid proxy, running on the main server (host), can't access my own site, because the "-t nat" rules aren't used 1145228567 M * orzel and i can't find a proper iptables line to just blindly forward port 80 to the vserver ip 1145228575 M * orzel do any of you know ? 1145229056 J * Dr4g_ ~Dr4g@80-195-133-218.cable.ubr06.uddi.blueyonder.co.uk 1145229505 Q * Dr4g Ping timeout: 480 seconds 1145231520 Q * orzel Remote host closed the connection