1190940647 Q * cehteh Ping timeout: 480 seconds
1190941328 J * cehteh ~ct@217.140.77.75
1190941721 J * _Hunger Hunger.hu@213.163.11.138
1190941799 Q * Hunger Ping timeout: 480 seconds
1190942206 Q * derjohn Remote host closed the connection
1190942440 J * nospoonuser ~nospoonus@n18-241.dsl.vianetworks.de
1190942763 Q * nospoonuser Quit: Changing server
1190945057 Q * jmcaricand Ping timeout: 480 seconds
1190945298 J * jmcaricand ~user@d77-218-94-213.cust.tele2.fr
1190946263 Q * jmcaricand Remote host closed the connection
1190947444 Q * Piet_ Quit: Piet_
1190950024 Q * zLinux Remote host closed the connection
1190953760 Q * nou Read error: Connection reset by peer
1190956806 J * JonB ~NoSuchUse@kg0-231.kollegiegaarden.dk
1190957480 Q * Aiken Quit: Leaving
1190958587 J * Aiken ~james@ppp121-45-249-108.lns2.bne4.internode.on.net
1190959984 J * jmcaricand ~user@d83-179-210-152.cust.tele2.fr
1190961072 Q * JonB Ping timeout: 480 seconds
1190962201 J * JonB ~NoSuchUse@130.227.63.19
1190963417 Q * arekm Quit: reboot
1190963592 Q * hparker Quit: *burp*.. It's broke
1190963724 N * Bertl_zZ Bertl_oO
1190963851 J * arekm arekm@carme.pld-linux.org
1190963853 A * arekm back
1190964173 J * dna ~dna@251-242-dsl.kielnet.net
1190968294 J * ema ~ema@rtfm.galliera.it
1190968453 J * hparker ~hparker@linux.homershut.net
1190968867 J * nou Chaton@2001:6f8:328:bbc:6666:6667::
1190969235 Q * mountie Read error: Operation timed out
1190969394 J * mountie ~mountie@trb229.travel-net.com
1190970511 N * phedny Guest205
1190970517 J * phedny ~mark@ip56538143.direct-adsl.nl
1190970922 Q * Guest205 Ping timeout: 480 seconds
1190971613 J * Pazzo ~ugelt@213.215.138.237
1190971657 M * Pazzo Hi @ll! Is http://lists.grok.org.uk/pipermail/full-disclosure/2007-September/066025.html affecting linux-vserver?
1190971755 M * Pazzo (sounds serious)
1190972051 M * arekm sure, it affect every kernel matching that advisory
1190972068 M * arekm vserver doesn't prevent anything in this case
1190972083 M * Pazzo arekm: thought so, thanks!
1190972114 M * Pazzo what about 32bit based vserver on a 64bit linux-vserver host?
1190972144 M * arekm patch fixing this issue applies cleanly to much older kernels, too btw (using on 2.6.16)
1190972178 M * arekm Pazzo: probably will also work but not sure
1190972220 M * arachnist Pazzo: `--# cat /etc/vservers/smgl-x86/personality
1190972220 M * arachnist linux32
1190972233 M * Pazzo I decided to stay with 2.6.20 on most servers - they already applied the patch to 2.6.20.20 :-) But still a lot of upgrade work to do right now :-(
1190972234 M * arachnist ah
1190972235 M * arachnist nvm
1190972240 M * Pazzo ?
1190972458 A * arekm waits for vserver gurus to help with the latest oops on vanilla+vserver
1190972475 M * JonB can a vserver guest not join a samba domain?
1190972756 M * Pazzo JonB: tried samba in a vserver guest long time ago, so I cannot give you a qualified answer... maybe you need broadcast CAPs? (guessing)
1190973006 M * JonB Pazzo: ok
1190973179 J * transacid ~transacid@transacid.de
1190974394 N * phedny Guest207
1190974399 J * phedny ~mark@ip56538143.direct-adsl.nl
1190974803 Q * Guest207 Ping timeout: 480 seconds
1190975019 Q * Pazzo Quit: ...
1190975040 M * JonB suppose i want to share users and passwordsw between 2 vserver guests
1190975055 M * JonB can i then just hardlink the passwd, shadow and group file togehter?
1190975678 Q * bzed Ping timeout: 480 seconds
1190975739 J * bzed ~bzed@devel.recluse.de
1190975957 J * Pazzo ~ugelt@213.215.138.237
1190975957 N * phedny Guest208
1190975963 J * phedny ~mark@ip56538143.direct-adsl.nl
1190976085 M * JonB where do etc/localtime come from ?
1190976130 M * Pazzo JonB: "man tzconfig" explains the whole procedure pretty well :-)
1190976157 M * JonB Pazzo: never mind, i was in the wrong directrory
1190976179 M * JonB i just saw that net join was accessing that file as one of the last before the error message
1190976259 N * phedny Guest209
1190976265 J * phedny ~mark@ip56538143.direct-adsl.nl
1190976369 Q * Guest208 Ping timeout: 480 seconds
1190976426 Q * ema Quit: leaving
1190976468 Q * Guest209 Ping timeout: 480 seconds
1190976560 N * phedny Guest210
1190976566 J * phedny ~mark@ip56538143.direct-adsl.nl
1190976678 N * phedny Guest211
1190976683 J * phedny ~mark@ip56538143.direct-adsl.nl
1190976845 M * Pazzo Bertl_oO: I'm out of office (at Rome for VON Italy ;-) and wrote some mails to our office to speed up Kernel-Upgrades as of CVE-2007-4573
1190976952 M * Pazzo They had some trouble applying the patch, as there are some chunks succeeding at different offsets (we are still using 2.6.20 as it currently is still actively maintained):   http://pastebin.ca/718347
1190976953 Q * Guest210 Ping timeout: 480 seconds
1190977068 Q * Guest211 Ping timeout: 480 seconds
1190977088 M * Pazzo As they are a little bit sceptic (and I have to leave right now): It would be really great if you could have a short look at the patch (and maybe provide a 2.2.x for 2.6.20.20)
1190977114 M * Pazzo Thanks a lot, and have a nice weekend!
1190977313 Q * Pazzo Quit: ...
1190977611 M * jmcaricand Pazzo: use linux-2.6.20.15.tar.bz2 not linux-2.6.20.tar.bz2
1190977648 Q * Aiken Quit: Leaving
1190978634 M * eviljonny Does anyone know which vserver branch is on Debian Sarge with the 2.6.18-4-vserver kernel?
1190978726 J * derjohn ~derjohn@80.69.41.3
1190978736 M * eviljonny I did have a look in the package information but it doesn't say. I also had a google which didn't turn up anything
1190979434 J * Baby ~miry@195.37.62.208
1190979448 P * Baby 
1190979461 N * phedny Guest213
1190979467 J * phedny ~mark@ip56538143.direct-adsl.nl
1190979812 M * jmcaricand eviljonny: apt-cache show kernel-patch-vserver
1190979871 Q * Guest213 Ping timeout: 480 seconds
1190979885 M * eviljonny Ahha, thank you
1190980277 M * jmcaricand As you see, Sarge use 2.6.8|2.6.11 kernel not 2.6.18 :-(
1190980333 M * jmcaricand but you can build __your__ 2.6.18 deb package :-)
1190980558 M * eviljonny Indeed, I was hoping I wouldn't have to. Before I go about doing this I want to find out if anyone knows if I can fix the problem I am having without going for a new kernel or newer branch
1190980760 M * eviljonny Should I be able to talk to the host server from the guests? If I try and talk to the loopback IP on the guests (only the host actually has a loopback IP, the guests only have a public IP) I get operation not permitted. Even if I try and ping the public IP of the host I still get operation not permitted. We have a custom kernal already which is 2.6.17.13-vs2.0.2.1, I am reluctant to change it because it is a live production server, if I have to t
1190980816 M * JonB how do i make a vserver guest listen to the broadcast address ?
1190980890 M * eviljonny Sorry if the setup seems a little wacky but I have come into this project after the build was complete and the server was made live
1190980948 M * transacid eviljonny: maybe that is disallowed by some iptables rules
1190981000 M * transacid hmm but then you should get "operation not permitted"
1190981031 M * eviljonny I had a look at the iptables rules and it is allowed, the rules are pretty simple. Also even if I am tcpdumping I dont see anything, I would expect to see the packets with tcpdump even if they are filtered out by iptables
1190981036 M * eviljonny Yeah
1190981043 M * arachnist etch has .18 afair
1190981054 M * arachnist right?
1190981072 M * jmcaricand arachnist: yes
1190981168 M * eviljonny Does etch have the 2.2 branch of vserver?
1190981232 M * eviljonny We were considering building a new server from scratch on etch instead of using backports and a custom kernal
1190981326 M * jmcaricand eviljonny: I believe it's 2.0
1190981333 M * jmcaricand for etch
1190981335 M * eviljonny Doh
1190981403 M * eviljonny Thanks
1190982359 Q * arachnist Quit: bbl
1190982638 M * jmcaricand eviljonny: "Even if I try and ping the public IP of the host" from guest ?
1190982891 J * arachnist arachnist@088156185052.who.vectranet.pl
1190983140 M * DavidS probably old news here, but i just stumbled upon the lwn writeup of linux' kernel containers talk from the 2007 kernel summit: http://lwn.net/Articles/249080/
1190985181 J * nospoonuser ~nospoonus@n18-241.dsl.vianetworks.de
1190985716 M * eviljonny jmcaricand: Even if I try and ping the public IP of the host....I still get operation not permitted
1190985769 M * eviljonny After some more digging I think it's possible that applications don't have permission to use Raw sockets, I'm not 100% sure yet but it looks that way (not saying this is a bug, my config error)
1190986569 J * andertx ~paul@auh-b17511.alshamil.net.ae
1190986581 P * andertx 
1190990459 M * UukGoblin hm
1190990493 M * UukGoblin 'vcontext' process takes up 100% cpu, mostly system, in one of my guests
1190990500 M * UukGoblin I killed running X in it...
1190990507 M * JonB which util-vserver?
1190990559 M * UukGoblin JonB, vserver 0.30.210 -- if that's the right version 
1190990566 M * JonB upgrade
1190990607 M * UukGoblin what about the process now? is my vserver dead? do I have to reboot?
1190990638 M * JonB i do not know. I only overheard someone speak about that problem and the answer was upgrade
1190990660 J * dowdle ~dowdle@scott.coe.montana.edu
1190991040 J * julien-vserver ~jujux@82.252.4.58
1190991043 M * julien-vserver hello
1190991111 M * julien-vserver daniel_hozac: I tried to see dmesg output after echoing 128 in /proc/sys/vserver/debug_misc, after doing the tty switch but nothing happens
1190991132 M * jmcaricand UukGoblin: you use 0.30.213 to solve this
1190991146 M * julien-vserver dmesg et echo done in host, I tried echoing in guest but I didn't have rights (as expected surely)
1190991147 M * UukGoblin jmcaricand, JonB, thanks
1190991264 M * julien-vserver daniel_hozac: I have debug kernel, bugverbose, info and vserver debug enabled in kernel
1190991433 M * julien-vserver daniel_hozac: I would like to see an version of xorg with debug symbols, but It doesn't seem there's one in debian stable, and I begin to take courses monday, I will not have time to compile my own xorg (and continue testing it)
1190991438 M * julien-vserver I'm sorry
1190991507 M * julien-vserver daniel_hozac: I don't know if I said it, but I will try debian unstable which include kernel 2.6.22 and one of the latest vserver patch as soon as possible and see if I have same problem
1190992340 M * julien-vserver and there's dbg version of xorg in unstable too
1190992358 J * igraltist ~user4@kasten-edv.de
1190993290 M * julien-vserver read you later
1190993299 Q * julien-vserver Quit: Leaving.
1190993750 M * dowdle Ok, so I have a VPS just the way I want it.  How do I create another VPS from that?  Can I just rsync the directory structure for it, and copy the config dir... do some file renaming, and start it up? (making sure to change the context ID).  If there is some docs on this somewhere, by all means, refer me to them.
1190993770 M * sid3windr you can do exactly the same thing you did to make the other one :D
1190993779 A * sid3windr grins, ducks and runs
1190993837 M * dowdle sid3windr: Yeah... funny... that's what I'm trying to avoid.
1190993869 M * dowdle I guess I could tar.gz it up and see if I can use that script that is used to install distro images... 
1190993942 M * dowdle I guess rather than asking... I could just give it a try.
1190993948 Q * UukGoblin Quit: [-;-]
1190994008 M * sid3windr I think rsync works fine
1190994021 M * sid3windr actually, I usually tarred them up when moving them to another hose
1190994023 M * sid3windr *host
1190994034 P * jmcaricand ERC Version 5.0.2 $Revision: 1.726.2.11 $ (IRC client for Emacs)
1190994036 M * sid3windr although on unpack I had some issues with the postfix sockets or something
1190994078 M * dowdle Any way to set a root password in my VPS... so I can use su as a user?
1190994092 M * sid3windr vserver enter
1190994096 M * dowdle The passwd command doesn't seem to work... it just exits with a token error.
1190994096 M * sid3windr passwd
1190994099 M * sid3windr oh
1190994104 M * sid3windr then I guess something's broken
1190994111 M * sid3windr it should just work like any other non-vserver
1190994117 M * dowdle bash-3.1# passwd
1190994117 M * dowdle Changing password for user root.
1190994117 M * dowdle passwd: Authentication token manipulation error
1190994125 M * dowdle Doesn't even give me the chance to set a password.
1190994128 M * sid3windr I guess your pam setup inside your distro thingy is broken
1190994146 M * dowdle Oh, yeah... perhaps I have some work to do on that.
1190994168 M * dowdle I did that last time (before I accidentally nuked the VPS I was working on) to fix ssh as user.
1190994178 M * dowdle Perhaps that same fix also fixes passwd for root.
1190994323 A * dowdle comments out all references to loginuid in /etc/pam.d/*
1190994343 N * ensc Guest222
1190994353 J * ensc ~irc-ensc@p54B4EB20.dip.t-dialin.net
1190994373 Q * JonB Ping timeout: 480 seconds
1190994463 Q * Guest222 Ping timeout: 480 seconds
1190994582 M * dowdle Egads, my VPS isn't using shadow passwords.  Argh.  What do I need to do to get that working?  CentOS 5 VPS.  yum search shadow (first thing)
1190994632 M * sid3windr shadowconfig on
1190994641 M * dowdle root has a * as the password so that was probably my problem, eh?
1190994681 M * dowdle no such command... shadowconfig.  I'll get it figure out here in a second.
1190994755 M * sid3windr magic:~# dpkg -S shadowconfig
1190994758 M * sid3windr passwd: /sbin/shadowconfig
1190994758 M * dowdle It has just been so long since I used a system that didn't by default come with shadow... but starting with a bare bones install created that I guess.
1190994760 M * sid3windr :)
1190994767 M * dowdle sid3windr: I *DID* say my VPS was CentOS.
1190994769 M * sid3windr I know
1190994776 M * sid3windr but shadowconfig has been around for 15 years
1190994790 M * sid3windr I do realize CentOS is gay, trust me :p
1190994805 M * sid3windr seems indeed to be a debian only thing.
1190994805 M * dowdle sid3windr: I haven't installed dpkg for one thing... and I don't think shadowconfig is a centos package... but what the hey.
1190994859 M * dowdle Red Hat based distros, pre-shadow as default... had a password converter program (which I'm assuming is similiar in function to shadowconfig).  There is also system-config-auth.
1190994874 M * dowdle password converter = take passwords from /etc/passwd to shadow
1190994951 M * dowdle Ok, just getting rid of the * as the password in /etc/passwd let me set a root password.  On down...
1190994954 M * dowdle errr. One down...
1190995207 M * dowdle I kinda did it the wuss way... but I installed setuptool, picked authentication, and put an asterisk by md5 shadow... and that created shadow and moved the passwords.
1190995631 M * dowdle Ok, installing JDK... and it says that no JRE is installed... like DUH... that's what you provide JDK.  I've not run into that before... but I do know that their install requires some packages you wouldn't think it did... like bc.
1190995812 M * dowdle Hmm, it doesn't seem to be extracting anything... and then complaining that it can't find what it didn't extract.  Something is missing for it to do that.
1190996088 J * JonB ~NoSuchUse@kg0-231.kollegiegaarden.dk
1190996814 M * brc ls returns this on some files :
1190996815 M * brc ?---------  ? ?    ?          ?            ? /fs3/vservers/user/usr/bin/which
1190996817 M * brc any hint?
1190996823 M * brc on the host and on the vserver
1190996843 M * brc if i try to rm it says input/output error. this is a raid5 
1190996845 M * brc configuration
1190996849 M * brc so i guess it is not hardware rlated
1190997025 M * JonB why do you guess that?
1190997029 M * JonB what does dmesg say?
1190997094 J * _jmcaricand_zzz ~jmcarican@d77-218-34-47.cust.tele2.fr
1190997106 J * zLinux ~zLinux@88.213.33.208
1190997145 M * JonB did you check the filesystem for errors?
1190997206 A * dowdle wusses out again... and just rsyncs a JDK install from another machine.
1190997235 J * olpau ~olpau@p57A6E488.dip.t-dialin.net
1190997237 M * brc dmesg does not say anything special. I have still not checked cause it is mounted
1190997246 M * brc Still not checked the filesystem
1190997246 M * JonB unmount
1190997261 M * brc Going to do that during the night. I found it weird because it is just inside one vserver
1190997273 M * brc No reboots, no crashes, nothing special happened
1190997366 Q * olpau 
1190997374 M * JonB sometimes stuff just break
1190997388 M * brc weird
1190997399 M * brc never saw that. do you think that this may be related to raid5 ?
1190997422 J * pbryan ~pbryan@192.18.43.225
1190997449 M * JonB no, i think it is filesystem related
1190997485 M * brc gonna do what you said, umount and fsck. I used to trust more ext3 :)
1190997510 M * JonB if the vserver is still running thats where i would look
1190997527 M * JonB i mean at the filesystem
1190997646 M * brc vserver is still running
1190997683 M * JonB it's not dead?
1190997713 M * brc not dead. the problem just happened with around 10 files
1190997743 M * JonB filesystem check
1190998162 M * brc i am doing rsync before
1190998174 M * brc afraid about having everything corrupted after umount/mount
1190998178 M * JonB what about your backup ?
1190998192 M * JonB unmount will not corrupt it
1190998226 M * brc Backup is remote would take a lot to restore
1190998230 M * brc so i am rsyncing to another local partition
1190998245 M * JonB ok
1190998248 M * arekm vserver devs alive?
1190998420 J * fatgoose ~samuel@76-10-153-239.dsl.teksavvy.com
1190998427 M * fatgoose hi
1190998427 M * fatgoose \
1190998609 J * Piet ~piet@tor.noreply.org
1190999192 M * pbryan Hey, anyone know what the Ubuntu 7.x options are for Vserver (other than rebuilding kernel from source + patches)?
1190999272 M * JonB installing a debian binary compiled kernel?
1190999304 M * pbryan Hmm. Grab the Deb kernel, install on Ubuntu. And bring across the linux-headers too, I guess...
1190999329 M * pbryan No incompatibilities between distros, to your knowledge?
1190999357 M * pbryan (There's been increasing incompatibilities on packages and dependencies between Ubuntu and Debian).
1190999376 M * pbryan Or maybe, the solution is to switch to Debian?
1190999418 Q * sladen Ping timeout: 480 seconds
1190999453 M * JonB that might also be a solution
1190999459 M * JonB maybe a better solution
1190999467 M * pbryan Yeah, I'm thinking it might be.
1190999485 M * pbryan Strangely, I'm not sensing a lot of demand for Vserver on Ubuntu.
1190999542 M * JonB i dont think serious users use ubuntu
1190999545 M * dowdle So, vtop... shows all VPSes?  How to you see only one VPS?
1190999567 M * pbryan JonB: At least on the server...
1190999582 M * JonB and vserver users are serious
1190999607 M * JonB dowdle: maybe the man page shows howto ?
1190999640 M * dowdle Wow... a vserver command that has a man page.  Who'da'thunk?
1190999655 M * _jmcaricand_zzz pbryan:  incompatibilities -> No 
1190999683 M * dowdle Ok, so it shows all processes and not just a VPS... so I guess if I want that, I run top inside of a VPS, eh?
1190999691 M * JonB pbryan: i tried using ubuntu on a server because the customer requested it
1190999701 M * JonB pbryan: i end up getting burned :-(
1190999738 M * _jmcaricand_zzz JonB: but Debian is the best :-)
1190999744 M * JonB yeah
1190999818 M * _jmcaricand_zzz dowdle: vserver guestid exec top ?
1190999880 M * dowdle _jmcaricand_zzz: I know it is hard to tell, but I'm starting to pull my head out.
1191000028 M * pbryan JonB & _jmcaricand_zzz: Thanks for the info.
1191000577 J * sladen paul@starsky.19inch.net
1191000831 M * dowdle So, bitfrost on the OLPC uses Linux-VServer.  Is that correct?
1191000913 Q * Piet Remote host closed the connection
1191001011 M * Supaplex dowdle: that's what I hear
1191001055 M * dowdle Supaplex: I've kinda heard that myself... but I don't recall where... but none of the docs I can find on OLPC mention Linux-VServer... that I can find.
1191001104 J * michal ~michal@www.rsbac.org
1191001122 M * Supaplex yea, I haven't researched or recall the specifics. I think Bertl mentioned it a few weeks ago
1191001126 N * michal Guest224
1191001137 N * Guest224 michal_
1191001406 J * Piet ~piet@tor.noreply.org
1191001747 M * dowdle I know unify saves dispace... but does it save RAM too?  Do VPSes have copy-on-write libraries that they share?
1191001757 M * dowdle errr dispace = disk space
1191001821 M * dowdle Yes, I'm reading the docs... but I haven't gotten there yet.
1191001966 N * Bertl_oO Bertl
1191002005 M * Bertl dowdle: Linux-VServer is part of the implementation of BitFrost
1191002043 M * Bertl dowdle: unified files will only be cached once (imap cache) and thus, any mapping on that inode (like for libraries or executables) will only happen once
1191002067 M * dowdle Bertl: Ok, I have a lot more questions regarding that but I'll save them for later.
1191002068 M * Bertl dowdle: so yes, if libraries and executables or even shared (read only) data is unified, it will save a lot of ram
1191002128 M * Bertl dowdle: regarding man pages: they are somewhat outdated, but all commands support up-to-date --help output
1191002153 M * dowdle Bertl: Ok, so unify saves memory too.  Does it have a copy-on-write function... so if a single VPS gets its libraries updated (or some other reasonable scenerio), it'll notice and adjust appropriately?
1191002159 M * Bertl dowdle: feel free to update the man pages and submit patches to daniel_hozac, he will gladly include them I guess
1191002173 M * dowdle Bertl: Like I didn't see that coming. :)
1191002190 M * Bertl dowdle: yes, the CoW Link Breaking takes care of updated
1191002193 M * Bertl *updates
1191002223 M * Bertl and a good admin re-unifies the guests every now and then, saving space and memory
1191002299 A * dowdle continues reading and absorbing.
1191002313 M * Bertl feel free to ask if you ahve more questions ...
1191002361 M * JonB Bertl: if i want to share users between guests, can i then just hardlink the passwd, shadow and group file?
1191002371 M * Bertl dowdle: regarding the 'copy existing guest', check out 'vserver - build --help'
1191002387 M * Bertl dowdle: you should see the 'clone' build method which is what you probably want
1191002412 M * Bertl JonB: you could do that, but I'm not sure it is a good idea for several reasons
1191002426 M * JonB Bertl: please enlighten me
1191002438 M * Bertl JonB: first, if you do not make that file immutable, folks from any guest could break that link 
1191002455 M * Bertl second, changing the password (or removing it) from one guest, will affect the others
1191002494 M * JonB but they need to be root to break it
1191002506 M * Bertl guest root, yes
1191002512 M * JonB thats me
1191002528 M * Bertl okay, then it is probably fine ...
1191002542 M * Bertl you could make it a little more secure by adding the immutable flag
1191002545 M * Radiance hiya guys :)
1191002546 M * JonB second... the entire point is that they are identical
1191002557 M * Radiance what's a safe way to add an ip to a running vserver ?
1191002575 M * JonB Bertl: but then people can not change password
1191002604 M * Bertl Radiance: naddress --help
1191002610 M * Bertl JonB: correct
1191002626 M * JonB Bertl: i suppose i could look at some ldap service
1191002627 M * epicbjorn that is unless you make it writable from all guests
1191002662 M * JonB Bertl: running samba inside a vserver guest, the faq says it needs the broadcast address assigned to that vserver
1191002663 M * Bertl JonB: but I'm not sure how typical password updating works because if it makes a copy of the file (to modify it) and then moves that copy back in place, the link won't work
1191002693 M * Bertl JonB: yeah, samba, per se, is very broadcasty ...
1191002707 M * JonB Bertl: but winbind works
1191002738 M * JonB Bertl: might the broadcasty be the reason that i could not get net ads to work inside but had to use net rpc
1191002855 M * JonB Bertl: how do i assign the broadcast address to a vserver guest?
1191002869 M * Bertl like a normal ip
1191002888 M * Bertl (just without the device info, i.e. use nodev)
1191002897 M * JonB ok
1191002926 M * Bertl Radiance: you might also want to add the ip after that to the configuration
1191002944 M * Radiance thanks :)
1191002984 M * Radiance should the ip be also already up in the host environment ?
1191003023 M * Bertl not necessarily, you can up/down it at any time
1191003036 M * Bertl but of course, you need to add it on the host if you want to use it :)
1191003068 M * Radiance ok :)
1191003099 Q * FireEgl Read error: Connection reset by peer
1191003439 M * dowdle JonB: What about home dirs?  A shared /etc/passwd[shadow] wouldn't magically make shared homedirs.
1191003474 M * Bertl yeah, but he can mount a shared /home into those guests
1191003491 M * Radiance BertI, btw, how do i see what the nid is ? 
1191003528 M * Bertl host or guest?
1191003548 M * Radiance what i need to have naddress add the ip to the vserver 
1191003564 M * JonB dowdle: one guest for svn server, one guest for cvs server
1191003572 M * JonB dowdle: they dont really use their homedir
1191003586 M * Bertl Radiance: usually the nid is identical to the xid if not specified otherwise
1191003598 M * Bertl Radiance: check /proc/virtnet/ for possible nids
1191003645 M * Radiance ah ok
1191003648 M * Radiance that would be 101
1191003673 M * Radiance had to assign it a while back after no more dynamix contexts if i remember correctly 
1191003699 M * Bertl yeah, the config has a context entry (usually)
1191003711 M * Bertl and you can also see it with vserver-stat
1191003750 M * Radiance thanks mate
1191003758 M * Bertl you're welcome!
1191003762 Q * hardwire Ping timeout: 480 seconds
1191003986 M * dowdle With the "clone" method of vserver-build, the source would be what?  The name of the source VPS?  A directory?  The context ID?
1191004023 M * dowdle Is that on the flower page?
1191004069 M * Bertl name and directory should work, not sure about the id
1191004137 P * pbryan 
1191004411 J * hardwire ~bip@rdbck-5855.wasilla.mtaonline.net
1191004440 M * dowdle How do you destroy a VPS?  1) shut it down, 2) rm -rf /vservers/<name> and 3) rm -rf /etc/vservers/<name>?
1191004460 M * Bertl vserver <name> stop
1191004467 M * Bertl vserver <name> delete/remove
1191004476 M * Bertl (can't remember which one :)
1191004512 M * Bertl just checked, it's delete :)
1191004516 M * dowdle delete it is.
1191004533 A * dowdle tries the clone method.
1191004564 M * dowdle And the vdu is a vserver aware du that'll let you figure out real space consumed by a cloned/unified system?
1191004599 M * Bertl yep, it knows about context tagging 
1191004612 M * Bertl (for shared partitions)
1191004650 J * virtuoso ~s0t0na@ppp91-122-59-139.pppoe.avangard-dsl.ru
1191004666 M * Bertl wb virtuoso!
1191004683 M * virtuoso Evening Bertl!
1191004817 Q * _jmcaricand_zzz Quit: KVIrc 3.2.4 Anomalies http://www.kvirc.net/
1191005298 M * dowdle Ok, so the clone method of vserver-build doesn't copy the config of the --source original over... what's the easiest way to clone a config?
1191005336 M * Bertl probably cp -va if you really want to do that (which is unlikely)
1191005366 M * Bertl at least certain config parts (like the symlinks) need to be per guest
1191005377 M * dowdle Well, in the vserver-build statement I gave it unique info... but it didn't create a config/dir... like other methods would.
1191005399 M * dowdle I know I can copy an existing config dir and edit the files (name, IP address, context id) manually... but is that the way everyone does it?
1191005403 J * ema ~ema@rtfm.galliera.it
1191005428 M * Bertl that sounds strange, the clone method will create a config for the new guest, what tools do you use and what was your command line?
1191005464 M * dowdle vserver mail1 build -m clone --context 4041 --hostname=mail1.cs.montana.edu --interface eth0=eth0:153.90.199.124/255.255.0.0 -- -d centos5 --source dbase
1191005484 M * dowdle dbase being an existing VPS (that was running at the time)
1191005499 M * Bertl so you should get an /etc/vservers/mail1 config tree
1191005530 M * Bertl eth0=eth0:... is probably not what you really want, btw
1191005548 M * Bertl it means that the ip will be added as eth0:eth0 alias
1191005595 M * dowdle Bertl: I just went by the example on the CentOS install page.  What would be a reasonable eth0=?  (given NIC, eth0)
1191005597 M * Bertl the rest looks good, and I think should be fine for 0.30.214
1191005612 M * Bertl dowdle: the network config works like this:
1191005650 M * Bertl  [<alias>=][<device>:]<ip>[/prefix|/netmask]
1191005670 M * Bertl so, if you want the tools to _assign_ the ip to an interface
1191005705 M * Bertl you have to specify <device>, and if you want to give it an alias name, like eth0:mail
1191005724 M * Bertl then you would use mail=eth0:<ip>/<prefix/mask>
1191005755 M * Bertl if you do not want the tools to _assign_ the ip, you use nodev instead
1191005787 M * Bertl for the kernel, the only relevant information is the ip and the prefix/netmask
1191005807 M * Bertl everything else is userspace to simplify the guest setup
1191005847 A * dowdle updates notes
1191005849 M * Bertl dowdle: what tool version (vserver --version) do you use?
1191005876 M * dowdle vserver 0.30.214
1191005903 M * Bertl please contact daniel_hozac if the clone method actually failed to create a config and didn't give you an error
1191005909 M * dowdle Hmm, maybe this caused the failure?
1191005910 M * dowdle vserver mail1 build -m clone --context 4041 --hostname=mail1.cs.montana.edu --interface mail1=eth0:153.90.199.124/255.255.0.0 -- -d centos5 --source dbase
1191005912 M * dowdle Opps
1191005913 M * dowdle bad paste
1191005927 M * dowdle vclone: Unify_copy("/etc/vservers/.defaults/vdirbase/mail1/sbin/new-kernel-pkg"): File exists
1191005934 Q * JonB Quit: This computer has gone to sleep
1191005979 M * Bertl looks like mail1 already exists (at least partially)?
1191006009 M * Bertl try to delete it first and rerun the clone with
1191006018 M * dowdle Will do.
1191006028 M * Bertl --interface eth0:153.90.199.124/255.255.0.0
1191006033 M * dowdle Making mistakes... and not getting it right the first time... is how one learns.
1191006041 M * Bertl precisely
1191006049 M * dowdle Not: --interface mail1=eth0:153.90.199.124/255.255.0.0
1191006057 M * dowdle ?
1191006060 M * Bertl you can do both
1191006070 M * Bertl but usually folks want to see eth0 not eth0:mail1
1191006083 M * Bertl (inside the guest, that is)
1191006127 A * dowdle updates notes
1191006296 J * Aiken ~james@ppp121-45-249-108.lns2.bne4.internode.on.net
1191006320 M * arekm Bertl: any idea? http://pastebin.com/m9df408a pure 2.6.22.9+vserver 2.3
1191006342 Q * hardwire Ping timeout: 480 seconds
1191006365 M * dowdle Hmmm, my clone didn't do anything but take a long time.  The things it says are there (like that error message) aren't.
1191006416 M * dowdle Should I shut down the source when I do a clone?  That sounds sensible.
1191006422 M * Bertl arekm: did the patch apply cleanly?
1191006450 J * hardwire ~bip@rdbck-2558.palmer.mtaonline.net
1191006490 M * arekm Bertl: no, two problems in fs/ - locks.c and splice.c
1191006521 M * arekm I fixed these manually
1191006542 M * arekm locks.c only actually. splice.c already has the changes
1191006598 M * arekm Bertl: I have the same problems with other versions too: http://lists.pld-linux.org/mailman/pipermail/pld-kernel/2007-July/001409.html
1191006604 M * arekm and http://lists.pld-linux.org/mailman/pipermail/pld-kernel/2007-July/001403.html
1191006615 M * arekm but 2.6.17+vserver 2.1 is stable
1191006666 M * arekm this is 2 x quad core 2.13GHz intel 2U machine, 4GB ram (if smp thing matters here)
1191006671 M * Bertl arekm: okay, do you have a way to trigger it?
1191006688 M * arekm Bertl: boot that kernel, vservers start, wait few minutes
1191006711 M * arekm 7 guests, few == like 5 minutes
1191006729 M * arekm and oops happens. guests doing typical stuff, web, ftp, ssh, bind
1191006818 M * Bertl so one of your guests suddenly exits?
1191006841 Q * bzed Quit: new kernel....
1191006867 M * arekm don't really know. machine freezes and reboots at that moment
1191006891 M * arekm (it's 300km away from me so that's only what I see over ipmi console)
1191006895 M * Bertl arekm: the BUG() you are hitting is a check that the context is actually shutting down when the contexxt is destroyed
1191006928 M * brc Bertl!
1191006939 M * Bertl arekm: it seems to me that free_vx_info() is called on a context which isn't shutting down (according to the state info)
1191006957 M * brc Can i ask you some kernel error doubts? :)
1191006967 M * Bertl brc: sure
1191007003 M * arekm Bertl: why it could be called that way? it happens on various kernels, various vserver snaps so there has to be some bug in the code IMO
1191007029 M * brc I had files like this (seeing from the host and inside a vserver), just oon /usr/bin of one vserver (around 10 files):
1191007032 M * brc ?---------  ? ?    ?          ?            ? ./vserver_name/usr/bin/crontab
1191007042 M * brc if i try to rm the file, i would receive i/o errror
1191007061 M * brc then i stoped this vserver, rsynched it to another partition and started it on this new partition (copying compromised files from a snapshot)
1191007062 M * Bertl arekm: ah, just saw it, that's on xfs, yes?
1191007068 M * arekm Bertl: yes
1191007073 M * brc Then i tied an rm -rf vserver_name/usr/sbin
1191007089 M * brc Andd i got KERNEL OOPS
1191007109 M * Bertl arekm: xfs is broken in recent kernels, there is a locking issue somewhere, but we didn't dig into it, as the xfs folks don't know either
1191007118 M * brc and some errors. Do you think that this server may crash because of this ?
1191007179 M * Bertl brc: that means your filesystem is in an inconsistant state (the kernel oops) what did it say?
1191007200 M * arekm Bertl: I wouldn't say 2.6.20 is fresh but ok, going to find xfs people 
1191007237 Q * ema Quit: leaving
1191007247 M * Bertl arekm: but I still don't see why the context should exit without getting into shutdown
1191007259 M * brc Bertl: http://pastebin.com/m7a147aa6
1191007334 M * Bertl arekm: sounds like a bug in the mm copy/drop code, will look into it shortly
1191007369 M * arekm older oops on (not vanilla) 2.6.22.9 http://pastebin.com/m9df408a
1191007393 M * arekm ups, wrong url
1191007417 M * arekm http://pastebin.com/m5f7f6115
1191007421 M * arekm this is correct one
1191007495 M * Bertl brc: yep, looks quite broken, nevertheless the ext3 folks might be interested as a bad filesystem should not be able to crash the kernel
1191008008 M * brc machine just crashed
1191008279 M * Bertl arekm: any unusual drivers or hardware you are using?
1191008401 J * JonB ~NoSuchUse@kg0-231.kollegiegaarden.dk
1191008514 M * arekm Bertl: no, standard stuff http://pastebin.com/m17b74bcf
1191008553 M * Bertl arekm: okay, could you do a test run with 2.3.0.24 and enable some more of the Linux-VServer debugging stuff?
1191008572 M * arekm Bertl: yes, more means what exactly?
1191008599 M * Bertl I'm interested in the history tracing, do you have a serial console?
1191008621 M * arekm ipmi console
1191008635 M * Bertl does that work well for longer traces on an oops?
1191008641 Q * zbyniu Quit: leaving
1191008658 M * arekm sometimes yes, sometimes no, will see
1191008664 M * arekm VSERVER_HISTORY_SIZE=64 is ok?
1191008671 M * Bertl okay, make that 256
1191008676 M * arekm I'll add syslog to remote server logging, too
1191008705 M * Bertl also try to narrow the number of guests down
1191008716 M * Bertl I suspect that a single of your guests triggers this
1191008723 M * arekm Bertl: http://pastebin.com/m2a5a4c35 fine?
1191008731 M * Bertl would be nice to figure which one and use only that one
1191008756 M * Bertl disable the MONITOR
1191008765 M * Bertl we don't need scheduling info for this one
1191008797 J * zbyniu ~zbyniu@host13-188.crowley.pl
1191008805 M * Bertl welcome zbyniu!
1191008822 M * zbyniu hello :)
1191008929 J * arekm_ ~arekm@chello089076024040.chello.pl
1191008939 Q * arekm Quit: leaving
1191008951 Q * JonB Quit: This computer has gone to sleep
1191009074 J * JonB ~NoSuchUse@kg1-20.kollegiegaarden.dk
1191009107 M * arekm_ reboot, with all vservers for now
1191009138 J * bzed ~bzed@devel.recluse.de
1191009157 M * arekm_ Sep 28 21:52:15 carme kernel: ��� tasks left: 4
1191009158 M * arekm_ Sep 28 21:52:19 carme kernel: ��� tasks left: 1
1191009166 M * arekm_ uh, some debugging print broken
1191009193 M * Bertl tasks left? interesting ...
1191009205 M * arekm_ Sep 28 21:51:37 carme shutdown[15416]: shutting down for system reboot
1191009206 M * arekm_ Sep 28 21:51:37 carme init: Switching to runlevel: 6
1191009206 M * arekm_ Sep 28 21:52:15 carme kernel: ��� tasks left: 4
1191009206 M * arekm_ Sep 28 21:52:19 carme kernel: ��� tasks left: 1
1191009206 M * arekm_ Sep 28 21:52:38 carme ntpd[1961]: ntpd exiting on signal 15
1191009208 M * arekm_ Sep 28 21:52:38 carme ntpd: ntpd shutdown  succeeded
1191009229 M * arekm_ anyway ignore for now since this is reboot on old 2.6.17
1191009257 M * Bertl hmm?
1191009278 M * Bertl ah, that was the shutdown ...
1191009282 M * arekm_ yes
1191009398 M * arekm_ now waiting 8)
1191009413 Q * transacid Remote host closed the connection
1191009434 J * transacid ~transacid@transacid.de
1191009717 M * dowdle It loads a lot of vz related modules into the kernel too.
1191009721 Q * JonB Ping timeout: 480 seconds
1191009739 M * Bertl dowdle: hmm?
1191009742 J * JonB ~NoSuchUse@kg0-231.kollegiegaarden.dk
1191009846 M * dowdle Opps... like... could you guess "wrong channel"?
1191009883 M * Bertl hehe, yeah, for a moment I was worried :)
1191010065 M * arekm_ Bertl: http://pastebin.com/m4c1e52c4
1191010132 M * Bertl hmm, sure you enabled the history on that one?
1191010145 M * Bertl (because I do not see a history trace :)
1191010202 M * arekm_ let me check again
1191010282 M * Bertl what's the mpt stuff?
1191010295 M * arekm_ SAS driver
1191010341 M * arekm_ CONFIG_VSERVER_HISTORY=y
1191010341 M * arekm_ CONFIG_VSERVER_HISTORY_SIZE=256
1191010367 M * arekm_ is set. unless some files weren't rebuilt after changing config (did only make oldconfig; make modules bzImage - no clean)
1191010384 M * Bertl clean isn't required
1191010400 M * Bertl 'make' should be all you need
1191010546 M * Bertl okay, also try to reduce the guests (e.g. disable one half of them)
1191010609 N * arekm_ arekm
1191010920 M * arekm ok, history on, verified by /proc/config.gz, waiting for crash now
1191010982 M * Bertl arekm: what tools are you using?
1191011019 M * arekm util-vserver 0.30.214
1191011034 M * Bertl built with dietlibc?
1191011053 M * arekm but with 13 patches, yes dietlibc
1191011074 M * Bertl 13 patches, nice ...
1191011094 M * arekm but nothing important in these patches I would say
1191011121 M * Bertl daniel_hozac might be interested in those if they fix something
1191011156 M * Bertl the kernel is unmodified, except for your 2.6.22.9 adaptations?
1191011179 M * arekm yes
1191011255 M * arekm vanilla built specially to reproduce this case 
1191011276 M * JonB vanilla icecream? *droool*
1191011304 M * Bertl arekm: okay, the sas stuff is required on that machine?
1191011320 M * arekm Bertl: sure. otherwise hard disks wouldn't be seen :)
1191011328 M * arekm sas is scsi successor
1191011362 J * dreamind ~dreamind@C2107.campino.wh.tu-darmstadt.de
1191011362 Q * dreamind 
1191011421 M * Bertl arekm: just checking our options to narrow it down :)
1191011464 M * Bertl arekm: to me, the issue looks like something decrements the ref count for the context (in a way which should not happen)
1191011509 M * Bertl the history tracing should give us a clue _where_ that happened, if we can reduce the number of history entries 
1191011517 M * arekm running guests one by one for 5mins each
1191011542 M * Bertl okay
1191011553 J * Punkie ~Punkie@home.pekelny.net
1191011558 M * Bertl wb Punkie!
1191011570 M * Punkie hi Bertl ;)
1191012557 M * arekm Bertl: http://pastebin.com/m6465f88
1191012562 M * arekm Bertl: 4 guests only now
1191012589 M * arekm Bertl: that history still missing right?
1191012593 M * Bertl hmm, still not a single line of history
1191012608 M * Bertl let me double check the code .. brb
1191012835 M * Bertl okay, that is the first bug(tm) we discovered today
1191012929 M * arekm heh, and the fix? :)
1191012935 M * Bertl sec
1191013327 Q * JonB Quit: This computer has gone to sleep
1191013545 J * bonbons ~bonbons@2001:960:7ab:0:20b:5dff:fec7:6b33
1191013550 M * Bertl arekm: http://vserver.13thfloor.at/Experimental/delta-history-feat01.diff
1191013561 M * Bertl that should enable the history tracing
1191013589 M * Bertl try to run only those four guest you used, and keep all the stacktraces around
1191013691 M * arekm include/linux/vserver/history.h:63: error: dereferencing pointer to incomplete type
1191013740 M * arekm 63: entry->vxi.usecnt = atomic_read(&vxi->vx_usecnt);
1191013747 M * Bertl #include <linux/vs_context.h>
1191013756 M * Bertl right before the history include
1191013769 M * Bertl (will update the patch in place in a second
1191013782 M * arekm builds now
1191013875 M * arekm reboot
1191013887 Q * Piet Ping timeout: 480 seconds
1191014216 Q * besonen_mobile_ Quit: Leaving
1191014385 Q * nebuchadnezzar Write error: connection closed
1191014385 J * nebuchadnezzar ~nebu@82.233.222.74
1191014389 Q * bonbons Quit: Leaving
1191014755 J * Julius ~julius@p57B253DF.dip.t-dialin.net
1191014760 M * arekm http://pastebin.com/m43044a22
1191014763 M * arekm only this catched
1191014778 Q * Julius Read error: Connection reset by peer
1191014789 J * julius_ ~julius@p57B253DF.dip.t-dialin.net
1191014904 M * Bertl arekm: hmm, strange, still no trace ...
1191014927 M * Bertl but on the good side, it seems easily reproduceable
1191014937 M * arekm hard lock now, need poweroff/on
1191015156 M * arekm I'll try to get another oops, maybe will be longer
1191015183 M * Bertl let's add another patch to the mix and reduce the guests further
1191015207 M * arekm 3 guests now
1191015386 J * fatgoose_ ~samuel@76-10-153-239.dsl.teksavvy.com
1191015656 J * besonen_mobile ~besonen_m@71-220-225-14.eugn.qwest.net
1191015737 Q * fatgoose Ping timeout: 480 seconds
1191015967 M * Bertl http://vserver.13thfloor.at/Experimental/delta-history-feat02.diff
1191016301 Q * julius_ Ping timeout: 480 seconds
1191016350 M * arekm http://pastebin.com/m725ffa6 2vservers this time
1191016356 M * arekm applying feat02 and retesting
1191016907 M * Bertl tv.com death notex2x~
1191016937 M * Bertl (wrong terminal)
1191016940 J * julius_ ~julius@p57B253DF.dip.t-dialin.net
1191017125 Q * dna Quit: Verlassend
1191017727 M * arekm Bertl: single guest, http://pastebin.com/m282fc1eb
1191017764 M * Bertl excellent
1191017841 M * arekm does it explain anything?
1191017864 M * Bertl will take a little to work through the history
1191018258 Q * julius_ Remote host closed the connection
1191018278 J * yarihm ~yarihm@84-75-130-73.dclient.hispeed.ch
1191019167 M * dowdle Ok, I made a VPS just like I like it.  Then I stopped it, tar.gz'ed up the filesystem for it, then used that as a template for vserver-build -m template... and everything went great.  Now I want to attempt to unify them.  Any docs for that?  Any docs that mention what the command even is so I can --help on it?
1191019198 M * daniel_hozac vserver ... hashify
1191019264 M * arekm daniel_hozac: want some patches for utils?
1191019276 M * daniel_hozac what for?
1191019301 M * arekm http://cvs.pld-linux.org/cgi-bin/cvsweb/SOURCES/util-vserver-build-umask.patch
1191019323 M * arekm http://cvs.pld-linux.org/cgi-bin/cvsweb/SOURCES/util-vserver-dev.patch
1191019335 M * dowdle vserver - hashify --help no workie?
1191019344 M * arekm http://cvs.pld-linux.org/cgi-bin/cvsweb/SOURCES/util-vserver-include.patch
1191019377 M * arekm http://cvs.pld-linux.org/cgi-bin/cvsweb/SOURCES/util-vserver-m4-diet.patch
1191019395 M * daniel_hozac i don't see the point of the umask patch?
1191019414 M * daniel_hozac dev is an absolute nono.
1191019454 M * daniel_hozac include makes no sense to me, there's no such file.
1191019472 M * baggins daniel_hozac: the point of umask patch is that with umask of 077 the vserver build process goes berserk and builds unusable guests
1191019514 M * dowdle [root@comp2 ~]# vserver mail1 hashify
1191019514 M * dowdle Failed to initialize unification for vserver
1191019534 M * daniel_hozac and, did you set it up?
1191019584 M * daniel_hozac as for the m4-diet patch, i don't see what problem that solves.
1191019599 M * daniel_hozac baggins: what build method relies on the umask? that just sounds broken.
1191019609 M * arekm daniel_hozac: vserver-compat.h is compat.h, was renamed here locally due to some reason, baggins should know why
1191019624 M * daniel_hozac compat.h is an internal header, it should not be installed.
1191019645 M * baggins daniel_hozac: the one that use mkdir? I don't remember :(
1191019730 M * arekm daniel_hozac: http://cvs.pld-linux.org/cgi-bin/cvsweb/SOURCES/util-vserver-rpm-fake-resolver-badperm-errorlogging.patch?rev=1.4
1191019778 M * daniel_hozac if you have a non-accessible root directory, you're going to have far worse problems than that.
1191019783 M * Bertl arekm: how many CPUs are on that machine?
1191019801 M * arekm Bertl: 2 x quad core
1191019813 M * Bertl so 8 CPUs total, yes?
1191019818 M * arekm yes
1191019821 M * arekm daniel_hozac: http://cvs.pld-linux.org/cgi-bin/cvsweb/SOURCES/util-vserver-vrpm.patch
1191019843 M * arekm daniel_hozac: you get the warning about root perm problem at least
1191019853 M * baggins daniel_hozac: but the "external" headers depend on it so it must be installed but file name conflicted with something
1191019867 M * daniel_hozac the external headers don't depend on it.
1191019897 M * baggins vserver.h?
1191019902 M * daniel_hozac right.
1191019918 M * daniel_hozac xid_t/nid_t/tag_t should be defined by your kernel headers.
1191019955 M * daniel_hozac what problem does the vrpm patch try to solve?
1191019960 M * baggins an "#include <compat.h>" is not a dependence? really?
1191019968 M * daniel_hozac there is no such thing...
1191019999 M * baggins ok, sorry, must be too late for my clear thinking
1191020023 M * arekm baggins: our local patch adds that include
1191020053 M * baggins yeah, so let's try to kill it and we'll see what breaks and fix it The Right Way
1191020116 A * dowdle finds -t /vservers/templates/centos-5.0-default.tar.gz
1191020118 M * dowdle Opps
1191020119 M * baggins vrpm patch makes %pre/%post script execute in corrent network context, it makes possible for those rpm scripts to work (as in restart service)
1191020123 A * dowdle finds http://linux-vserver.org/util-vserver:Vhashify
1191020128 M * dowdle I swear I used to know how to paste
1191020142 M * daniel_hozac baggins: when you do what?
1191020147 M * daniel_hozac because that's already handled...
1191020195 M * daniel_hozac (when you run vrpm, that is)
1191020254 M * baggins I don't quite understand the question, is it "how I run vrpm?"
1191020290 M * daniel_hozac no, how do you get to the code path where it's not already inside the network context?
1191020297 M * dowdle Hmmm, that page doesn't say much and vserver <guest> hashify just says it can't do that Hal.
1191020344 M * dowdle err... Hal = Dave
1191020348 M * daniel_hozac http://linux-vserver.org/Frequently_Asked_Questions#How_do_I_manage_a_multi-guest_setup_with_vhashify.3F
1191020349 M * baggins vrpm guest -- --upgrade -vh <some-packege>
1191020363 M * baggins nothing out of ordinary
1191020397 M * daniel_hozac that runs in the network context here...
1191020456 M * arekm that patch is from jan 2006, maybe it was fixed in upstream later?
1191020487 M * dowdle Now HOW did I miss that.
1191020488 M * dowdle ?
1191020586 M * dowdle Question:  I have 4 VPSes... and only one is running... which is the one I'm going to hashify.  So, does it look at all vserver dirs or only the ones that are running?
1191020604 M * dowdle Or that step only creates a hash for other VPSes to unify against?
1191020607 M * daniel_hozac it only looks at the one you tell it to look.
1191020626 M * dowdle Nevermind... it says do this one line per vserver.
1191020661 J * mstrobert ~mstrobert@wkstn.wycliffe.ca
1191020705 M * baggins arekm: it looks so (from recent flames on pld-devel)
1191020842 M * Bertl arekm: could you try to capture one or two more of those history traces, and run the addresses listed at the end of each line starting with @ through 'addr2line -e vmlinux'?
1191020918 M * arekm Bertl: yes
1191020977 M * baggins daniel_hozac: as for umask patch, a simple grep mkdir /usr/lib/util-vserver/* gives a lot of mkdirs without -m755
1191020984 M * arekm Bertl: http://carme.pld-linux.org/~arekm/vmlinux :)
1191021009 M * arekm Bertl: while I'm going to capturing more traces
1191021022 M * baggins maybe a better solution would be to substitute all mkdir with a $MKDIR_P and set MKDIR_P in util-vserver-vars?
1191021072 M * daniel_hozac baggins: yes, and i don't see why any of them would be problematic.
1191021144 M * baggins mkdir -p "$VDIR/var/lib" maybe ?
1191021159 M * arekm baggins: tested vrpm, correct context is applied
1191021169 M * daniel_hozac your build method should set proper permissions on /var and /var/lib.
1191021193 M * arekm baggins: http://pastebin.com/m8c29857
1191021270 M * baggins arekm: looks correct?
1191021297 M * arekm baggins: yes, ip a in %post shows only 1 IP while outside there are 2 so it's as expected
1191021334 J * Punkie_ ~Punkie@home.pekelny.net
1191021380 Q * Punkie Ping timeout: 480 seconds
1191021392 N * Punkie_ Punkie
1191021528 M * arekm Bertl: addr2linux shows ??:0
1191021546 M * arekm tested 3 random addresses, huh?
1191021553 M * baggins arekm: rebuild kernel with -g ;)
1191021556 M * daniel_hozac do you have CONFIG_DEBUG_INFO enabled?
1191021624 M * Bertl arekm: if so, the addr2line might be broken
1191021657 M * arekm no debug info enabled. Will setting it to =y + make change addresses ? (in other words is checking addresses from old oops sane?)
1191021696 M * daniel_hozac it's probable.
1191021827 M * baggins it worked for me
1191022371 M * arekm [arekm@carme-pld linux-2.6.22]$ addr2line -e vmlinux 0xffff81013347f3a0
1191022371 M * arekm ??:0
1191022388 M * Bertl try with gdb
1191022417 M * arekm and the command in gdb is?
1191022488 M * Bertl # gdb vmlinux 
1191022494 M * Bertl (gdb) l *<addr>
1191022507 M * arekm (gdb) l *0xffff81013347f3a0
1191022507 M * arekm No source file for address 0xffff81013347f3a0.
1191022509 M * arekm hm
1191022668 M * Bertl okay, I'm too tired right now to take this apart without address mappings
1191022687 M * Bertl but I'm confident that we can track this down with a little more time
1191022697 M * Bertl (and testing :)
1191022721 M * Bertl arekm: what is that guest doing in that 5 minutes?
1191022746 M * Bertl and why is it giving up references at the end?
1191022797 M * arekm Bertl: 1 user, compiling kernel, sshd, lighttpd, some other services
1191022812 M * Bertl arekm: that machine is NUMA, right?
1191022844 M * arekm [root@carme ~]# dmesg|grep -i NUMA
1191022845 M * arekm No NUMA configuration found
1191022846 M * arekm no
1191022913 M * Bertl hmm ... so probably just the parallelism, preemption enabled?
1191022937 M * Bertl no, should be listed on the traces
1191022985 M * Bertl okay, I'll call it a day, but I have a few ideas where to look ... can we continue testing tomorrow?
1191023045 M * arekm voluntary preempt only
1191023089 M * arekm we can play tommorow, uh, today, after some sleep :)
1191023105 M * Bertl okay, then let's do that ... thanks for your time!
1191023126 M * Bertl have a good one everyone! cya later ...
1191023133 N * Bertl Bertl_zZ
1191023201 M * arekm that l * thing doesn't work due to something
1191023202 M * arekm (gdb) print &__kstrtab___netdev_alloc_skb
1191023202 M * arekm $6 = (char (*)[19]) 0xffffffff805870c0
1191023202 M * arekm (gdb) l *0xffffffff805870c0
1191023202 M * arekm No source file for address 0xffffffff805870c0.
1191023645 A * arekm sleeps
1191023646 Q * arekm Quit: leaving