1127692817 M * Johnsie Ask regardless... heh
1127692823 M * Johnsie Someone may take notice and answer.
1127692834 M * jorgy I did a "cp -r /vservers/v004 /vserver/v012" and also /etc/vservers/xxx
1127692840 M * Johnsie 95% of my questions get answered... I just have to wait sometimes. :)
1127692850 M * jorgy :-)
1127692854 M * Johnsie Uh-huh...go on.
1127692866 M * jorgy this is something that I would do in kernel 2.4 versions of vserver, and it worked okay.
1127692894 M * jorgy However, I get some *weird* things going on.  Like if I copy a file from the host operating system into /vserver/v012/root it doesn't show up there!  but if I go there in the host, it's still there!
1127692905 M * Johnsie Well, really, I don't think you can copy the /etc/vserver stuff...
1127692921 M * Johnsie Since each folder within /etc/vserver is specific to that VPS.
1127692934 M * jorgy also, the other *weird* thing is when I make a change in /vservers/v004/xxx the change also appears in v012.
1127692939 M * Johnsie You could copy things over, but you'd have to go reset some data specific to that VPS, etc.
1127692952 M * jorgy yeah?  I went in and changed nodename, IP,  in /etc/vservers/v012.
1127692954 M * Johnsie Yes, that'd be why... and I forget which file that is...
1127692971 M * daniel_hozac /etc/vservers/xxx has tons of symlinks.
1127692971 M * Johnsie Well, there's a lot of stuff symlinked to the vserver's root.
1127692986 M * jorgy symbolic links?   that sounds bad for a "cp" command.
1127692993 M * Johnsie Very. :)
1127692993 M * jorgy what is the recommend way?  tar?  dump/restore? rsync?
1127692999 M * Johnsie No...
1127693004 M * Johnsie Use vserver to rebuild. :)
1127693032 M * Johnsie Example:  mv /vserver/v004 (or whatever you're using) to /vserver/v004-backup
1127693033 M * jorgy well I'm getting bit by a but in current version of fedora core 4 and vyum.
1127693051 M * jorgy so I can't create a new one from scratch right now.
1127693054 M * Johnsie Then go and create your vserver with the vserver command, it'll make the skeleton...
1127693079 M * jorgy what would the argument to -m be to create the skeleton?
1127693080 M * Johnsie Then trash the contents in /vserver/v004 and move data from /vserver/v004-backup back into /vserver/v004.
1127693097 M * Johnsie Quite frankly, I don't know. ;)
1127693102 M * daniel_hozac -m skeleton
1127693109 M * jorgy k.
1127693114 M * jorgy which distro do you use?
1127693118 M * Johnsie Oh, the argument...hah
1127693125 M * Johnsie Thank you, daniel_hozac ... my brain is off.
1127693133 M * Johnsie I like Gentoo, personally.
1127693134 M * jorgy cool, thanks daniel
1127693144 M * jorgy that uses emerge, not apt or yum, right?
1127693149 M * Johnsie Yes.
1127693155 M * jorgy cool.
1127693175 M * jorgy I didn't realize that there were symbolic links in /etc/vservers/xxx.  that's probably my problem.
1127693182 M * daniel_hozac guaranteed.
1127693190 M * jorgy in any event, this is just a temporary thing until erico gets the vyum thing fixed anyway.
1127693211 M * Johnsie Yep.
1127693216 M * Johnsie I tried that once myself.
1127693219 M * Johnsie That's how I know...haha
1127693222 M * jorgy LOL
1127693249 M * jorgy wel lthanks johnsie and daniel.  I'm off to make some vserver  "-m skeleton".
1127693257 M * Johnsie It's just easier to rebuild using vserver and then copy data to its respective home.
1127693263 M * Johnsie You're welcome.
1127693265 M * Johnsie Good luck.
1127693268 M * jorgy thanks
1127693271 M * Johnsie Let us know how things turn out.
1127693275 M * jorgy will do.
1127693423 M * jorgy uh oh, got an error I saw recently on the mailing list:
1127693437 M * jorgy chroot-shopen("/etc/sysconfig/network"): No such file or directory
1127693437 M * jorgy chroot-shopen("etc/inittab"): No such file or directory
1127693446 M * jorgy do I need to worry about that?
1127693467 M * Johnsie Mmmmmmm, which distro are you using?
1127693472 M * jorgy FC4
1127693478 M * Johnsie Hmmmm.
1127693492 M * daniel_hozac how did you install the guest?
1127693503 M * jorgy oops. 
1127693514 M * jorgy I did a "-m skeleton -d fc4".
1127693529 M * jorgy what should I do for "-d" if I'm building just a skeleton, to copy stuff in after the fact?
1127693541 M * Johnsie Well, I never tried it in FC4...I'm a Gentoo die-hard guy, I guess, but I would -assume- that the IP configuration stuff resides on the host, which means you probably don't need /etc/sysconfig/network on the guest and /etc/inittab could be recreated.
1127693565 M * Johnsie But, I don't want to give you false information.
1127693570 M * Johnsie That's just my assumption.
1127693572 M * daniel_hozac leave -d out entirely.
1127693577 M * jorgy k.
1127693589 M * jorgy k, I'll try that (no -d)
1127693629 M * jorgy okay, no error this time.
1127693647 M * jorgy now I'm going to copy an existing vserver guest into there.  will "cp -rp /vservers/v004 /vserver/v012" work?
1127693657 M * Johnsie We'll have to put up another gold star on daniel_hozac's chart...haha
1127693673 M * jorgy * for daniel
1127693680 M * daniel_hozac heh.
1127693684 M * Johnsie Yes, that should work fine.
1127693687 M * jorgy k.
1127693688 M * jorgy brb
1127693815 M * jorgy bingo
1127693823 M * jorgy working perfectly, with no traces of v004 in it.
1127693828 M * Johnsie Yay!
1127693895 M * jorgy so the solution was to NOT cp /etc/vservers/v004 - create a skeleton, and then copy in the guest filesystem.
1127693905 M * daniel_hozac exactly.
1127693939 M * jorgy okay, next question... LOL
1127693948 M * jorgy I've seen various approaches to running bind in a vserver.
1127693967 M * jorgy the whole reason I'm using vserver is to limit damage if a service is compromised.
1127693989 M * jorgy for example, if postfix is "contaminated", I don't want all the other services (ldap, bind, whatever) to be as well.
1127694015 M * jorgy but if I make the changes to allow bind to run in a vserver, will it mean that if bind is ever rooted, my entire vserver host and guests will be vulnerable?
1127694040 M * daniel_hozac not really.
1127694057 M * daniel_hozac it could make it easier for the attacker to DoS your system post-breakin.
1127694086 M * jorgy by attacking network stacks of guests?
1127694107 M * daniel_hozac fork bombs and such.
1127694129 M * jorgy ahhh, I see.
1127694135 M * daniel_hozac but that could be averted with proper rlimits and CPU limiting.
1127694158 M * jorgy but it wouldn't let an attacker get a root shell on another guest (or the host) or even sniff tcp traffic to other guests?
1127694158 M * daniel_hozac (the only capability needed for regular BIND is CAP_SYS_RESOURCE)
1127694170 M * daniel_hozac not at all.
1127694178 M * jorgy okay, that's what I was afraid of.
1127694195 M * jorgy so I have to turn on CAP_SYS_RESOURCE to run bind - can I run stock bind, or do I need to compile from source?
1127694221 M * daniel_hozac that would be for stock bind.
1127694240 M * daniel_hozac if you compile it from source, i have a patch that disables the CAP_SYS_RESOURCE requirement.
1127694240 M * jorgy cool.  I prefer to do it with RPMs, it's easier to keep up with security patches, etc.
1127694245 M * daniel_hozac i agree.
1127694260 M * jorgy okay, I'll give that a whirl.
1127694264 M * jorgy thanks daniel!
1127694267 M * jorgy bye
1127694274 M * daniel_hozac bye
1127694277 Q * jorgy Quit: Chatzilla 0.9.67 [Fedora 1.0.6-1.2.fc4/20050909]
1127694303 J * revenger ~revenger@p54B01F26.dip0.t-ipconnect.de
1127694351 M * revenger is there any small distribution out there that works with vserver don't want to download 600 mb redhat now as explained in the howto..
1127694452 M * daniel_hozac you could do a net install of any of the common distros.
1127694507 M * daniel_hozac but basically any distro with a compiler and that works with a vanilla kernel would work with vserver.
1127694752 M * jayeola hi guys. how's this for a guide? http://deb.riseup.net/vserver/preparing/
1127694776 M * daniel_hozac jayeola: immediate question is, why isn't it on the wiki?
1127694784 M * jayeola um, dunno
1127694787 M * jayeola :P
1127694800 M * jayeola are there any steps in there missing or things that should not be there?
1127695548 Q * jayeola Quit: leaving
1127697388 N * Bertl_oO Bertl
1127697394 M * Bertl evening folks!
1127697568 M * Bertl daniel_hozac: reg. wiki, guess it's already linked there, at least I saw/checked this some time ago
1127697580 M * Bertl (not verified though)
1127697621 M * daniel_hozac ah, ok.
1127697653 M * Bertl daniel_hozac: btw, thanks for the support today, seems you had a busy one :)
1127697678 M * daniel_hozac hehe.
1127697745 M * Bertl I finally got the routing issues fixed (provider), so I should be available again :)
1127697792 M * daniel_hozac great!
1127697842 M * Bertl yeah, hope we can address ngnet soon ... looks good ...
1127698055 M * daniel_hozac cool, i guess a lot of people want to see that.
1127698235 M * Bertl the sys$command was fun to read ... :)
1127698250 M * Bertl you're on the ML too, no?
1127698279 M * daniel_hozac yeah, hehe.
1127698358 M * Bertl I once annoyed somebody by placing a file called '-rf' in his homedirectory  :)
1127698386 M * daniel_hozac hahaha
1127698438 M * daniel_hozac lol, i just read your replies to the body-less messsages.
1127698460 A * Bertl is in a funny mood right now :)
1127698567 P * stefani parting (is such sweet sorrow)
1127698631 M * Bertl daniel_hozac: I'm toying with the idea to make a kind of base kernel, including things like BME, split, various fixes, etc and base linux-vserver devel kernels and later stable kernels on that, what do you thing of that?
1127698694 M * Bertl do you know any useful/nice to have patches which are not in mainline but would be great to have for linux-vserver?
1127698785 M * daniel_hozac the base kernel idea makes sense. makes it more obvious what is directly vserver related.
1127698822 M * Bertl yeah, and I don't like to maintain BME for 2+ branches
1127698893 M * Bertl (and it looks like there will be more "BME"s in the future)
1127698977 M * daniel_hozac hmm, more extensions? such as?
1127699059 Q * litage Ping timeout: 480 seconds
1127699168 M * Bertl looks like http://archives.free.net.ph/thread/20050921.092132.62997b82.en.html#20050921.092132.62997b82
1127699213 M * Bertl (just an example)
1127699519 M * daniel_hozac ah.
1127699568 M * daniel_hozac i thought that was one of the most straight-forward kernel patches i've ever read...
1127699895 J * litage ~nick@203.201.97.26
1127700133 M * Bertl daniel_hozac: yeah, well, even Roman had a hard time to fine excuses :)
1127700168 M * daniel_hozac heh.
1127700188 Q * bnc Ping timeout: 480 seconds
1127700332 M * daniel_hozac well, i'm tired. good night all!
1127700344 M * Bertl excellent idea ...
1127700366 M * Bertl have a good whatever, everyone ... cya tomorrow!
1127700373 M * Bertl daniel_hozac: sweet dreams!
1127700382 N * Bertl Bertl_zZ
1127700428 J * erwan__taf ~erwan@ASte-Genev-Bois-151-1-65-47.w82-121.abo.wanadoo.fr
1127700831 Q * erwan_taf Ping timeout: 480 seconds
1127701321 Q * Aiken Quit: Leaving
1127702114 J * liquid_ ~liquid@p549745C7.dip.t-dialin.net
1127702556 Q * liquid Ping timeout: 480 seconds
1127706453 J * sebi ~sebi@C4dde.c.strato-dslnet.de
1127706558 Q * sebi_ Ping timeout: 480 seconds
1127711955 M * eyck Bertl: ooh, so you're starting your own patchset? :) great.
1127713842 J * Aiken ~james@tooax6-039.dialup.optusnet.com.au
1127714661 Q * erwan__taf Remote host closed the connection
1127717425 Q * monrad Quit: Leaving
1127720408 Q * Aiken Quit: Leaving
1127722567 J * prae ~prae@gut75-1-81-57-27-189.fbx.proxad.net
1127723060 J * hvd ~takeagues@fw-grz.hollomey.com
1127723239 Q * litage Ping timeout: 480 seconds
1127723554 Q * hvd Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org
1127723596 J * hvd ~takeagues@fw-grz.hollomey.com
1127724057 J * litage ~nick@203.201.99.114
1127724861 J * nox ~nox@noxlux.de
1127726679 J * yarihm ~yarihm@84-74-16-246.dclient.hispeed.ch
1127728979 J * TheSeer ~theseer@border.office.salesemotion.net
1127736774 J * entroposcope ~entroposc@24.148.139.16
1127737386 N * Bertl_zZ Bertl
1127737574 J * micah_ micah@micha.hampshire.edu
1127737993 Q * micah Ping timeout: 480 seconds
1127738021 M * Bertl morning folks!
1127738034 M * Bertl eyck: yeah, probably ...
1127738291 J * Homere ~homere@did75-2-81-57-120-45.fbx.proxad.net
1127738293 M * Homere hello
1127738330 M * Homere how is it possible for a script to determine if it's executed on a vserver or a real host ?
1127738426 M * Homere i'm working on a Plesk/ispconfig/.... clone for Debian
1127738431 M * Bertl welcome Homere!
1127738478 M * Homere and I need to choose between pureftpd or proftpd and bind version automatically
1127738497 M * Bertl depends on the setup ...
1127738568 M * Bertl if the system is configured to give away this information, you have several ways to read it ... if not, you can only 'try' to detect it by trial and error
1127738593 M * Bertl - /proc/self/vinfo
1127738628 M * Bertl - /proc/self/status
1127738640 J * neverdies maharaja@80.64.143.53
1127738643 M * neverdies hiho
1127738648 M * neverdies uhm - wrong config :)
1127738649 Q * neverdies Quit: 
1127738685 J * maharaja_ maharaja@ip52.ipax.at
1127738687 M * maharaja_ re again :)
1127738694 M * Bertl welcome maharaja_!
1127738697 M * maharaja_ :)
1127738726 M * maharaja_ Bertl: sometimes, vnamespace --new -- /usr/sbin/vserver ----nonamespace cluster1 start
1127738729 M * maharaja_ hangs
1127738734 M * Homere Bertl: thanks /proc/self/vinfo works 
1127738746 M * maharaja_ the status of this process ist D+ (using ps rax)
1127738762 M * maharaja_ do you know anything about that or how to debug this?
1127738790 M * maharaja_ util-vserver 0.30.208 - including your fix02 (if the recompile worked)
1127738842 M * Bertl Homere: yes, but be careful, a simple flag set, and it's gone ...
1127738858 M * maharaja_ the current situation is: i am writing some kind of wrapper for starting/stopping/managing vservers. cluster1 is up and running
1127738880 M * maharaja_ therefore, vserver cluster1 start (called by my script) should fail with "vserver allready running"
1127738907 M * Bertl D means it is doing device I/O, a) which process hangs, b) what kernel/setup do you use
1127738912 M * maharaja_ but vnamespace --new .... hangs after calling clone()
1127738966 M * Homere Bertl: ok, I'll use it until I find another solution
1127738996 M * Bertl Homere: yeah, most guests will have it ... I guess
1127739049 M * maharaja_ bertl: http://phpfi.com/79903
1127739142 M * Homere another question, what is important vserver hosting hardware ?
1127739156 M * Homere Memory, discs (RAID 5 or 10) ?
1127739196 M * Homere I plan to use AMD64X2, 4 to 8 GB of memory and RAID 10 (SATA 7200 RPM)
1127739252 M * Homere I've tested Lycos Vserver and it's very slow, I don't want to provide such bad service
1127739383 M * entroposcope #@!$!@ "stable" kernels
1127739417 M * entroposcope ;-)
1127739543 M * Bertl Homere: depending on what service and when you tested Lycos, they probably had about 100-120 guests on each node (maybe even more)
1127739575 M * Bertl Homere: memory is very important, goo disk I/O too ...
1127739602 M * Bertl Homere: SMP reduces latency and improves peak performance
1127739711 M * maharaja_ Bertl: my processes are still stuck in D+ and T+
1127739722 M * Bertl Homere: 64bit archs (like x86_64) have a real advantage regarding memory ...
1127739730 M * Homere I know
1127739736 M * Bertl maharaja_: could it be that you race them against eachother?
1127739745 M * Homere I'm using AMD64 and Opteron, it's really fast
1127739760 M * Homere I dont't want to use Pentium anymore :)
1127739813 M * Bertl maharaja_: testme.sh output please ...
1127740048 M * Bertl Homere: in general, tuning the kernel for your host system can improve overall performance drastically ...
1127740118 M * Homere I'm always using custom kernels without modules
1127740131 M * Homere modules are bad :)
1127740164 M * Bertl well, yes and no :)
1127740207 M * Bertl but you're probably better of with a monolithic kernel, that's right ...
1127740343 M * Homere I cut a server into units. 1U = 128 MB memory, 2GB disk and 1/32 CPU
1127740407 M * maharaja_ Bertl: testme.sh - http://phpfi.com/79911
1127740411 M * Bertl tx
1127740432 M * maharaja_ am i "allowed" to kill the processes?
1127740453 M * Bertl well, D processes can not really be killed ...
1127740474 M * Bertl the T means traced, you got that from the strace, I guess
1127740506 M * Bertl what does the -dev5 mean?
1127740525 J * monrad ~monrad@213083190134.sonofon.dk
1127740542 M * Bertl welcome monrad!
1127740549 M * monrad hi
1127740564 M * maharaja_ dev5 is the server name
1127740570 M * maharaja_ development server nr 5
1127740583 M * Bertl k
1127740640 M * Bertl maharaja_: so how do you get those processes into this state, could you describe it to me in more detail?
1127740741 Q * Homere Quit: This computer has gone to sleep
1127740753 J * Homere ~homere@did75-2-81-57-120-45.fbx.proxad.net
1127740773 M * maharaja_ Bertl: of course
1127740797 M * maharaja_ i am trying to build a managment script which doeas start/stop/create/remove/list available vservers
1127740838 M * maharaja_ during the tests, i started vserver cluster1 (vserver cluster1 start) and tried to issue another start by using my script
1127740867 M * Bertl okay, so you basically raced two starts against eachother, right?
1127740870 M * maharaja_ ./vmanage.sh start cluster1 invokes "vserver cluster1 start"
1127740888 M * maharaja_ mhm - i think i waited until the first start succeeded
1127740894 M * maharaja_ and was completet
1127740908 M * maharaja_ but i cannot say for sure that there has been no race condition
1127740938 M * Bertl k, and this 'happened' or is reproduceable?
1127740946 M * Homere is it possible to allocate more memory than available, 128 MB / vserver, 32 vservers = 4GB. What happends it I put 64 vservers on the host ?
1127740948 M * maharaja_ shouldn't "vserver cluster1 start" handle those race conditions?
1127740972 M * maharaja_ Bertl: i try to kill the running processes and will report with more details
1127740987 M * Bertl Homere: works just fine a) swap will be used and b) if no swap is available you'll get OOM kills (unless strict no overcommit)
1127741014 M * Bertl maharaja_: and another question, this is not on the 'hanging' machine, right?
1127741020 M * Homere ok, I understand how lycos works now :)
1127741046 M * maharaja_ Bertl: no - i can kill the processes with "killall -9 vnamespace"
1127741068 M * maharaja_ Bertl: but i noticed a strange error on stopping a vserver
1127741080 M * Bertl maharaja_: that would point to userspace synchronization waiting for 'something' to happen ...
1127741109 M * maharaja_ Bertl: http://phpfi.com/79913
1127741114 M * Bertl maharaja_: could you also upload the output of 'vserver-info - SYSINFO' please?
1127741164 M * maharaja_ Bertl: http://phpfi.com/79914
1127741187 M * Bertl Homere: the lycos setup also uses NFS, i.e. all guests are served from filers to nodes ...
1127741218 M * Homere yes, it's the case
1127741239 M * Bertl maharaja_: hmm, please try to trigger the D+ again ...
1127741298 M * maharaja_ Bertl: i try to do so - right now, i'm experiencing another behavior which should not occur (imho)
1127741313 M * Bertl let's hear ...
1127741315 M * maharaja_ Bertl: maybe this is the cause for the D+ situation in some cases
1127741325 M * maharaja_ Bertl: posting infos to phpfi.com
1127741331 M * Bertl k
1127741366 M * maharaja_ Bertl: how can i verify that i successfully recompiled the vserver tools _with_ your patch?
1127741391 M * maharaja_ Bertl: http://phpfi.com/79915 - please take a look at the amount of time vserver cluster1 start is taking
1127741411 M * Bertl good question ... probably with objdump :)
1127741461 M * Bertl well, looks like it is already running, what do you expect?
1127741530 M * Bertl or is it the delays you are worried about?
1127741544 M * maharaja_ i think that there shouldn't be any delays :)
1127741557 M * maharaja_ still investigating that D+ situation tou
1127741558 M * maharaja_ thou
1127741600 M * Bertl well, util-vserver does some things in unfortunate ways ... the 'running' check is one of those things ...
1127741630 M * maharaja_ again - is there any way to check if fix02 has been compiled into the util-vserver package i use?
1127741649 M * Bertl again, probably with objdump :))
1127741747 M * maharaja_ whats objdump? :)
1127741846 M * Bertl a tool to dump elf information and/or disassemble code ...
1127741973 J * RoadRunnR ~MrRoadRun@213.187.82.17
1127741986 M * Bertl maharaja_: the patch only affects code, not information or such ... maybe it should, but it doesn't 
1127741990 M * Bertl welcome RoadRunnR!
1127741997 M * RoadRunnR hi Bertl 
1127742054 M * RoadRunnR i just send a mail to the list about a BUG(), are there any additional information i should provide?
1127742174 M * Bertl RoadRunnR: ah, interesting ... please provide the output of:
1127742205 M * Bertl http://vserver.13thfloor.at/Stuff/SCRIPT/testme.sh
1127742240 M * RoadRunnR Linux-VServer Test [V0.13] Copyright (C) 2003-2005 H.Poetzl
1127742240 M * RoadRunnR chcontext is working.
1127742240 M * RoadRunnR chbind is working.
1127742240 M * RoadRunnR Linux 2.6.13+vserver x86_64/0.30.208/0.30.208 [Ea] (0)
1127742240 M * RoadRunnR VCI:  0002:0001 236 03000136
1127742240 M * RoadRunnR ---
1127742242 M * RoadRunnR [000]# succeeded.
1127742244 M * RoadRunnR [001]# succeeded.
1127742246 M * RoadRunnR [011]# succeeded.
1127742248 M * RoadRunnR [031]# succeeded.
1127742250 M * RoadRunnR [101]# succeeded.
1127742252 M * RoadRunnR [102]# succeeded.
1127742254 M * RoadRunnR [201]# succeeded.
1127742256 M * RoadRunnR [202]# succeeded.
1127742269 M * Bertl k, what does 2.6.13+vserver mean?
1127742304 M * RoadRunnR it's build using the debian make-kpkg, but without the debian patches
1127742320 M * Bertl k, which patches did you use?
1127742321 M * RoadRunnR it justs sets the uname a bit different
1127742349 M * RoadRunnR only patch-2.6.13.1-vs2.1.0-rc2.diff.bz2 
1127742367 M * Bertl ah, okay, so it's the development branch ... good ...
1127742428 M * RoadRunnR the box just restarted and i can now restart the vserver just fine, so i might be related to the fact that i changed the interface config of the vserver before doing the restart
1127742437 M * Bertl RoadRunnR: do you know how to trigger that?
1127742543 M * RoadRunnR no, just tried it and changing the interface stuff was not the problem
1127742561 M * RoadRunnR but i have the shell output from the failed restarts:
1127742564 M * RoadRunnR cspfe01:/etc/vservers/web01/interfaces# vserver web01 restart
1127742565 M * RoadRunnR Stopping periodic command scheduler: cron.
1127742565 M * RoadRunnR Stopping MTA: exim4.
1127742565 M * RoadRunnR Stopping internet superserver: inetd.
1127742566 M * RoadRunnR Stopping OpenBSD Secure Shell server: sshd.
1127742568 M * RoadRunnR Saving the System Clock time to the Hardware ClockCannot access the Hardware Clock via any known method.
1127742570 M * RoadRunnR Use the --debug option to see the details of our search for an access method.
1127742572 M * RoadRunnR Hardware Clock updated to Mon Sep 26 12:59:25 UTC 2005.
1127742574 M * RoadRunnR Stopping system log daemon: syslogd.
1127742576 M * RoadRunnR Stopping web server: Apache2Warning: DocumentRoot [/var/www/secure.connect-pay.com/htdocs/] does not exist
1127742578 M * RoadRunnR Warning: DocumentRoot [/var/www/www.connect-pay.com/htdocs/] does not exist
1127742580 M * RoadRunnR .
1127742582 M * RoadRunnR Sending all processes the TERM signal....
1127742584 M * RoadRunnR Sending all processes the KILL signal....
1127742586 M * RoadRunnR Unmounting remote and non-toplevel virtual filesystems....
1127742588 M * RoadRunnR Mounting root filesystem read-only...mount: permission denied
1127742590 M * RoadRunnR .
1127742592 M * RoadRunnR A timeout occured while waiting for the vserver to finish and it was
1127742596 M * RoadRunnR killed by sending a SIGKILL signal. Please investigate the reasons
1127742598 M * RoadRunnR and/or increase the timeout in apps/vshelper/sync-timeout.
1127742600 M * RoadRunnR Vserver '/etc/vservers/web01' still running unexpectedly; please investigate it manually...
1127742602 M * RoadRunnR RTNETLINK answers: Cannot assign requested address
1127742604 M * RoadRunnR RTNETLINK answers: Cannot assign requested address
1127742606 M * RoadRunnR /usr/lib/util-vserver/vserver.stop: line 110: 10252 Segmentation fault      mount -o remount,rw /proc
1127742608 M * RoadRunnR Message from syslogd@cspweb01 at Mon Sep 26 13:00:03 2005 ...
1127742610 M * RoadRunnR cspweb01 kernel: invalid operand: 0000 [1] SMP
1127742612 M * RoadRunnR Message from syslogd@cspweb01 at Mon Sep 26 13:00:06 2005 ...
1127742614 M * RoadRunnR cspweb01 kernel: invalid operand: 0000 [2] SMP
1127742616 M * RoadRunnR does that help?
1127742638 M * Bertl maybe, but it would be better to put such info on a pastebin ...
1127742652 M * RoadRunnR ups, sorry, next time
1127742682 M * RoadRunnR http://pastebin.com/374628
1127742694 M * Bertl what is running inside the guest after that?
1127742722 M * Bertl and the segfault? what segfaults there?
1127742787 M * RoadRunnR vps shows nothing, vserver-stat also claims the vserver is gone, but the box itself is "kind-of-dead", box is alive, reboot just hangs, only power-cycle helps
1127742830 M * Bertl okay, and you can trigger this somehow (i.e. within a reasonable amount of time)?
1127742885 M * RoadRunnR this was the first in time in the last 5 days that something like this happend, i'll try to trigger it again, but at the moment i'm not sure if i can trigger it
1127742892 M * Bertl the #
1127742897 M * Bertl mount -o remount,rw /proc
1127742922 M * Bertl seems to be there to 'flush' the procfs, we had that at some point (around 1.9.x IIRC)
1127742940 M * Bertl it looks to me like this could be related to the oops ...
1127742990 M * Bertl RoadRunnR: this is a testbox?
1127743009 M * RoadRunnR kind of, it's supposed to go into production shortly
1127743021 M * RoadRunnR within a week or so
1127743053 M * Bertl okay, you want to investigate this issue?
1127743083 M * RoadRunnR i'd like too
1127743132 M * Bertl okay, then let's check a few things and compile a new kernel (with some debug support)
1127743148 M * Bertl first:
1127743160 M * Bertl - what compiler and binutils do you use for the kernel?
1127743173 M * Bertl - is the guest 64bit or 32bit?
1127743206 M * RoadRunnR 64bit, gcc (GCC) 4.0.2 20050821 (prerelease) (Debian 4.0.1-6), GNU objdump 2.16.91 20050902 Debian GNU/Linux
1127743333 M * Bertl okay, let's downgrade that to 3.3.6 / 2.15.* first
1127743547 M * RoadRunnR gcc-3.3 is ok, but debian/sib only contains the 2.16 binutils, i would need to rebuild them first
1127743556 M * RoadRunnR ups, sid of course
1127743749 M * entroposcope hrm.  why does rpmbuild against util-vserver-0.30.208 want a xalan and xerces java xslt and xml stuff?
1127743757 M * entroposcope ah
1127743759 M * entroposcope for docs
1127743776 M * liquid_ ja hat geklappt
1127743787 M * nox fs/jfs/inode.c rejects when i try to apply patch-2.6.13-vs2.0.1-pre2.diff to 2.6.13.2 
1127743846 M * Bertl nox: okay, expect an update patch shortly ...
1127743874 M * nox Bertl: you are too great (:
1127743931 M * Bertl entroposcope: there is an option to disable doc building
1127744070 M * entroposcope ack
1127744079 A * entroposcope just didn't bother with the rpm
1127744150 P * RoadRunnR Leaving
1127744153 J * RoadRunnR ~MrRoadRun@213.187.82.17
1127744501 M * Bertl RoadRunnR: okay, if you got the 3.3.x gcc, make sure to enable the following kernel CONFIG options:
1127744525 M * Bertl CONFIG_VSERVER_DEBUG=y
1127744529 M * Bertl CONFIG_VSERVER_HISTORY=y
1127744536 M * Bertl CONFIG_VSERVER_HISTORY_SIZE=256
1127744554 M * Bertl leave the rest of your config as is ...
1127744814 M * RoadRunnR ok, rebuilding
1127745433 Q * eyck Read error: Connection reset by peer
1127746535 M * Bertl nox: http://vserver.13thfloor.at/Experimental/patch-2.6.13.2-vs2.0.1-pre2.diff.bz2
1127746580 M * nox lots of thx Bertl 
1127746611 M * Bertl you're welcome!
1127746661 M * maharaja_ bertl: do you know where http://phpfi.com/79934 comes from?
1127746717 M * Bertl I'd say it's the sync/waiting/rc-script process which is killed from the timeout 
1127746736 M * Bertl (or alternatively the vkill killing the rc stop)
1127746788 M * maharaja_ is there a way to debug this issue?
1127746798 M * maharaja_ or - what way do you recommend :)
1127746827 M * maharaja_ as it would be too tempting to answer my previous wquestion with 'yes, there is such a way' ;)
1127747120 J * eyck eyck@81.219.64.71
1127747829 M * Bertl wb eyck!
1127747854 M * Bertl maharaja_: it is most likely related to the runlevel scripts
1127747872 M * Bertl maharaja_: or some process (minilogd comes to my mind) not being stopped
1127747874 M * brc bertl
1127747878 M * brc good morning
1127747916 M * brc The sum of the RSS used for all vservers doesn't match the amount of memory used reported by "free". shouldn't them match ?
1127748031 M * Bertl depends ... where do you get the RSS sums from?
1127748081 M * brc  /proc/virtual/XID/limit -> RSS LINE first field.
1127748113 M * Bertl hmm, okay, that should match up, how much is it off?
1127748126 M * brc a lot
1127748137 M * brc maybe i am calculating it in the wrong way
1127748226 M * Bertl show me ...
1127748238 M * brc well
1127748246 M * brc if i have 1241 there
1127748254 J * stefani ~stefani@superquan.apl.washington.edu
1127748260 M * brc i do a (1241 * 4) / 1024
1127748313 M * Bertl ahem, 1241 pages a 4k each means 1241*4096 bytes ...
1127748339 M * brc i did this so long ago that i am all confused.. thinking.. hold :)
1127748358 M * Bertl free shows  used and buffers/caches
1127748372 M * Bertl so you have to substract buffers and chaces to get similar
1127748389 M * RoadRunnR Bertl: new kernel bootet, put the problem seems to have disappeared
1127748394 M * brc i come from this : 32MB=8192 64MB=16384 128MB=32768 256MB=65536
1127748400 M * Bertl brc /proc/meminfo is probably a better source ...
1127748420 M * brc ok lemme check
1127748423 M * Bertl RoadRunnR: we will try to trigger it, but honestly I suspect the gcc is to blame ...
1127748492 M * Bertl brc: 1241 is one guest, right, not the sum of all guests?
1127748513 M * brc just one guest
1127748532 M * brc hold
1127748542 M * lonewolff hey, i have been running debain vservers on debian hosts for about a month, but id like to offer my customers other distros, can yo have different distros in the vservers from the host distro?
1127748567 M * Bertl brc: well, don't  forget to substract the host's processes too
1127748573 M * Bertl lonewolff: simple, install them ...
1127748613 M * Bertl lonewolff: with debian as host system, it might be a little trickier, as most of the network based installs (based on apt-rpm and yum) do not work out of the box ...
1127748630 M * lonewolff ah 
1127748639 M * brc The sum is 237221
1127748641 M * Bertl lonewolff: but if you have, let's say a redhat or mandrage guest/tempalte, just isntall it
1127748681 M * lonewolff yeah, its just making the template that has me stumped atm
1127748760 M * Bertl if you are a provider, I'd suggest to install a few other distros (natively) and reduce/adjust them manually to become guest tempaltes
1127748776 A * entroposcope at that same spot atm
1127748802 M * lonewolff that makes sense i suppose, just instal on a system and copy / and get rid of the un-needed stuff :)
1127748822 M * Bertl on redhat/mandrake you can use debootstrap to get debian guests, as well as apt-rpm to get most rpm based distros via netinstall
1127748843 M * Bertl the apt-rpm never worked for me on debian (as host)
1127748874 M * brc Bertl: the sum is 237221, which means approx 948884 KB
1127748881 M * lonewolff hmm, so maybee i should use a different host os for the new server 
1127748919 J * Dark_Princess ~1034B640E@bzq-80-39-180.red.bezeqint.net
1127748929 M * Bertl welcome Dark_Princess! 8-)
1127748932 M * brc Bertl: how could i get that value  on meminfo ?
1127748942 M * Dark_Princess thnx u! : )
1127748951 M * Bertl brc: what is this the sum of?
1127748988 M * Dark_Princess hello to everybody
1127749015 M * brc Bertl: the sum of the first field on the RSS line of all /proc/virtual/xid/limit
1127749035 M * Bertl okay, you have to add the sum of all RSS values of host processes to that 
1127749063 M * brc but there is more free memory than it should be even without summing the host values
1127749075 M * brc my total memory is 1024mb
1127749123 M * Bertl hum, that's interesting ...
1127749131 M * Dark_Princess im gone
1127749134 M * Dark_Princess bye
1127749140 M * Bertl Dark_Princess: k, cya!
1127749157 M * Dark_Princess cya ; )
1127749159 P * Dark_Princess 
1127749180 M * brc Bertl: better to have more free memory than less :P
1127749200 M * Bertl yeah, but it means that some memory is accounted twice inside the guests
1127749217 M * brc Yeah
1127749219 M * brc i've done an other thing
1127749221 M * brc ps -auxw
1127749223 M * brc and summed the rss
1127749226 M * brc inside the context
1127749236 M * brc and it is always less than the one in /etc/virtualxid/limit
1127749240 M * brc smaller
1127749271 M * Bertl okay ... that sounds easy to reproduce ... will check it ...
1127749275 M * brc ok
1127749279 M * Bertl brc: kernel/version details?
1127749438 M * mnemoc Bertl: what 'uts' acronym means? 
1127749825 M * brc 2.6.11.3-vs1.9.3
1127749838 M * brc Sorry
1127749845 M * brc no it is right
1127749846 M * brc heheh
1127749864 M * brc kinda old ?? I did not upgrade it cause it was the first one to be 100% stable for me.
1127750030 M * Bertl mnemoc: good question, I have no answer what it stands for .. it is the uname related info ...
1127750048 M * Bertl brc: yeah, I'd suggest to give 2.0.1 a try ...
1127750062 M * mnemoc ust could be unix system type... but uts? :)
1127750064 M * Bertl (will be the version I'll test this with)
1127750118 M * brc i've had so manyu crashes in the past that i am afraid of upgrading :)
1127750171 M * Bertl heh ...
1127750478 Q * Homere Quit: This computer has gone to sleep
1127751228 M * entroposcope is there a way to get a loopback interface running inside a vserver?
1127751234 M * entroposcope or does the host's loopback work?
1127751424 J * Homere ~homere@did75-9-82-229-149-96.fbx.proxad.net
1127751572 M * Bertl entroposcope: loopback as in lo?
1127751580 M * entroposcope yes
1127751583 M * entroposcope as in 127.0.0.1
1127751590 M * entroposcope and lo
1127751595 M * Bertl that is working inside a guest (i.e. the host one is used)
1127751622 M * entroposcope ok
1127752151 Q * prae Quit: Execute Order 69 !
1127752261 M * Bertl brc: hmm, on 2.0.1 the accounting seems to be correct ...
1127752308 M * Bertl brc: (i.e. vserver-stat RSS == sum(ps auxwww) >= RSS current *4096
1127753228 M * entroposcope Bertl, regarding lo/127.0.0.1, I need to explicitly configure that on the host side of things to get the guest to bind to that interface?
1127753904 J * mef ~mef@targe.CS.Princeton.EDU
1127753908 M * mef hey bertl
1127754245 M * mef what do the VC_VXF_INFO_LOCK and VC_VXF_STATE_SETUP flags mean?
1127755416 M * mnemoc interfaces/if/mask = 255.255.255.0 thing and prefix = 24 equivalent?
1127755674 M * daniel_hozac mnemoc: yes.
1127755838 M * mnemoc does util-vserver does the conversion or it use ifconfig for mask and iproute for prefix?
1127755955 M * daniel_hozac i believe util-vserver does the conversion.
1127755964 M * daniel_hozac don't take my word for it though ;)
1127755984 M * daniel_hozac (and i would assume both ifconfig and iproute handle prefixes as well as netmasks)
1127756321 M * Bertl entroposcope: guests do not bind to any interfaces, just IPs
1127756488 M * Bertl mef: VXF_INFO_LOCK is not used anymore, VXF_STATE_SETUP means that the guest is in setup (administrative state), this is a one time flag
1127756553 M * mnemoc what 'scope' on interface config is?
1127756603 M * entroposcope ok. the interface (or alias) is created in the host, and IP bound to it, and then it's somehow available in the guest, per /etc/vservers/[vserver]/[ifnum]/dev,ip,prefix
1127756633 M * entroposcope ?
1127756667 M * Bertl yeah, kind of .. the guest is bound to a set of IPs ...
1127756669 M * entroposcope chbind used to lock guest
1127756685 M * Bertl interfaces carrying those IPs will be visible even if hide_netif is enabled
1127756697 M * entroposcope ok
1127756726 M * entroposcope and the /etc/vserver/.../interfaces/... stuff is where that part is configured, right?
1127756752 M * Bertl yes, exactly ...
1127756763 M * entroposcope my question was whether or not i had to explicitly configure lo and 127.0.0.1 there in order for a guest to have access to 127.0.0.1
1127756769 M * entroposcope and it looks like that's what i had to do
1127756773 M * entroposcope because that's what worked.  ;-)
1127756786 M * Bertl no, you do not have to configure 127.0.0.1
1127756796 M * Bertl (and you don't want to do taht)
1127756798 M * entroposcope well
1127756800 M * entroposcope hrm
1127756807 M * Bertl 127.0.0.1 is remapped to the first guest IP
1127756823 M * entroposcope when I didn't do that, I didn't see 127.0.0.1, nor could I ping that interface
1127756842 M * Bertl why would you want to?, btw it's a ping deficiency ...
1127756847 M * mef bertl: thanks
1127756861 M * entroposcope some apps want to bind to 127.0.0.1
1127756883 M * mef bertl: so I should just set VXF_STATE_SETUP in the mask and not as part of the flagword when doing vx_set_cflags().
1127756900 M * entroposcope I'm going to be running this in somethign like a hosting environment, and I'd prefer to avoid having to explain to random users that they can't use 127.0.0.1
1127756973 M * mef bertl: any progress on a token bucket based disk io controller?
1127757157 M * Bertl entroposcope: it's up to you, but usually no app binds to 127.0.0.1, all bind to localhost :)
1127757189 M * entroposcope heh.
1127757191 M * entroposcope I suppose
1127757198 M * Bertl mef: yes and no, you usually just clear the flag
1127757201 M * entroposcope so if I redefine localhost appropriate I don't ahve to care.
1127757219 M * Bertl yes, that's what most folks do ... (in /etc/hosts)
1127757228 M * entroposcope I'm a little confused by the differences i'm seeing between my gentoo host and the host I just put together
1127757232 J * prae ~benjamin@sherpadown.net
1127757237 M * entroposcope I think it must be between 2.0 and 2.1 vserver codebae
1127757240 M * entroposcope codebase
1127757242 M * Bertl mef: I had a deep look at it, but it seems mainline already got that under control ...
1127757289 M * Bertl entroposcope: yes, IIRC, gentoo uses the devel branch ...
1127757302 M * entroposcope yes, 2.1.x
1127757337 M * entroposcope i just patched my 2.6.12 kernel with patch-2.6.12.4-vs2.0.diff
1127757388 M * Bertl mef: the I/O scheduler, especially the cfq scheduler already allows for good I/O tuning ...
1127757422 M * entroposcope in 2.1 (or perhaps on gentoo), I can see all of my guest's interfaces -- all set up as aliases of eth0, named eth0:[vserver]
1127757433 M * entroposcope I know that part is somewhat arbitrary, in that I could name them anything
1127757442 M * entroposcope but they're visible and explicitly aliases
1127757455 M * entroposcope in 2.0, I don't see the guest's interface from the host
1127757491 M * Bertl you're sure that you use the same guest config? if so, then it's a bug of the gentoo kernel :)
1127757506 M * entroposcope oh, it's not the same guest config
1127757516 M * entroposcope but...
1127757528 M * Bertl well, then I assume that one guest has 'aliases' configured ... the other not :)
1127757545 M * entroposcope but IP is happening in the host, as you said.
1127757554 M * Bertl I know this multitude of options and choices is confusing, we need something simpler :)
1127757555 M * entroposcope guest just inherits whatever host tells it, via chbintg
1127757578 M * Bertl no, seriously, the issue is this:
1127757618 M * Bertl ifconfig was developed around 1980 :), it was replaced (well should have been) by iproute2 five or six years ago ...
1127757641 M * Bertl unfortunately most folks still use ifconfig, so they only get the legacy layer for that
1127757666 M * Bertl using 'ip addr ls' for example, you will find all ips configured properly on the hoest/guest
1127757670 M * entroposcope ah, ok
1127757672 M * entroposcope I see
1127757672 M * Bertl *host
1127757689 M * Bertl ifconfig on lists the aliased ones ...
1127757695 M * Bertl *only
1127757719 J * liquid__ ~liquid@p54973C6B.dip.t-dialin.net
1127757730 N * liquid__ liquid
1127757896 M * entroposcope where is  hide_netif set?
1127757930 M * Bertl it's in the flagword, 'flags' IIRC, see Flower Page for details
1127757941 M * entroposcope ok
1127757974 M * Bertl yep, flags it is ... contains all the cflags
1127758107 M * mef cfq works on a per vserver basis?
1127758126 Q * liquid_ Ping timeout: 480 seconds
1127758140 M * mef How would I limit one vserver from hogging the disk io bandwidth with a series of dd operations?
1127758202 M * Bertl no, the per vserver basis is not there ... but
1127758220 M * Bertl cfq can do balancing between 'taskgroups'
1127758252 J * oliwel ~mail-at-o@host-62-245-151-178.customer.m-online.net
1127758259 A * oliwel waves hello to the crowd
1127758266 M * Bertl mef: so basically assigning all tasks of one guest to one group, and giving all groups a lower priority than default, should do the trick (untested)
1127758271 M * Bertl welcome oliwel!
1127758277 M * oliwel Hi Bertl :)
1127758298 M * oliwel Nice to meet you - I hope you can help me (usually you can...)
1127758327 M * oliwel I have problems with drbd/vserver like mentioned in this post http://list.linux-vserver.org/archive/vserver/msg10413.html
1127758355 M * oliwel I need a workaround urgently because I must move two of my running servers to another host
1127758400 M * Bertl well, I already answered that one, no?
1127758494 M * oliwel You gave the tip to "unmount the device inside the contexts"
1127758499 M * Bertl oliwel: any part of my reply (http://list.linux-vserver.org/archive/vserver/msg10415.html) which wasn't clear?
1127758500 M * oliwel I dont know how to do that
1127758523 M * Bertl with umount? 
1127758585 M * oliwel But I dont see the devices inside this context
1127758593 M * Bertl but maybe let's take a step back and look at the whole picture .. what are you trying to accomplish/avoid?
1127758599 M * oliwel ok
1127758621 M * oliwel I have 3 Webserver, they have their www-tree on a drbd device
1127758628 M * brc Bertl: so you think there might be a problem on the memory limits? Since the accounting is incorret? or i am making a lot of confusionhere :)
1127758635 M * brc on vs1.9.3
1127758653 M * oliwel Memlimits was me yes, but this should be solvde now :)
1127758658 M * Bertl brc: on 2.0+ the accounting seems fine ... not going to test 1.9.3 right now :)
1127758707 M * oliwel Bertl: so I do three steps for starting the webserver
1127758717 M * brc Bertl: hehe ok
1127758720 M * oliwel Bertl: 1) set drbd to primary
1127758743 M * oliwel 2) monut the drbd to /data/myserver
1127758771 M * oliwel 3) start the vserver, this mounts the data into the vserver by mount bind, but only the one that I am using...
1127758777 M * oliwel all this works fine
1127758798 M * oliwel Now I do a "vserver www1 stop", umount /data/myserver, drbdadm secondary www1
1127758798 M * Bertl k, now you have 2 mounts of the drbd
1127758813 M * Bertl (inside the guest, and one outside)
1127758814 M * oliwel But stoping the vServer should also free the mount ?
1127758854 M * Bertl once the guest is completely stopped, the drbd should be released, yes
1127758857 M * oliwel ok but I am running a total of 5 vServer - so every one of these 5 context's has an instance of the drbd ount ?
1127758880 M * oliwel even if the drbd is outside the space of the vserver ?
1127758890 M * Bertl yes, if you start another guest (before unmounting the drbd) it will have one mount inside
1127758906 M * oliwel ok than this is my problem..
1127758910 M * Bertl oliwel: do you have a test system at hand?
1127758925 M * oliwel but how do I unmount this inside the context ? I cant see the mount there....
1127758931 M * Bertl we could do a few simple tests regarding this, which should pretty illustrate it ...
1127758943 M * oliwel No unfortunately this is my live system - I did all tests with one vServer....
1127758964 M * Bertl no laptop or so .. pity ...
1127758971 M * oliwel hmm wait - I think I can abuse my Fileserver at Home :)
1127758981 M * oliwel The Laptop doesnt have a drbd/vserver
1127758988 M * mef bertl: what's weird is that when I create a new context, set the ccaps, and then set the flags with flags.mask=VC_VXF_STATE_SETUP; and flags.flagword=0ull; then the call to vc_set_cflags() bumps me out of the context I just created.
1127759004 M * mef bertl: weird!
1127759008 M * Bertl oliwel: well,s trictly speaking we do not even need drbd ... but of course, it's better to test with ...
1127759066 M * Bertl mef: detailed example?
1127759082 M * oliwel Ok I will use the backup system - its worthless when I cant move the context so - a crash will cause 20 mns for walking there to reboot...
1127759104 M * Bertl so don't crash it :)
1127759114 M * oliwel Depends on you :=
1127759122 M * oliwel So what should I try ?
1127759133 M * Bertl first, make sure no guests are running ...
1127759149 M * Bertl (keep a guest available, a simple one, which can be started and stopped)
1127759173 M * oliwel Can I ue the system withput drbd ?
1127759174 M * Bertl then get something to mount ... e.g. your drbd device ...
1127759178 M * oliwel That would be simpler
1127759180 M * oliwel ok
1127759193 M * Bertl yes no problem, take a CD image or whatever to mount
1127759200 M * oliwel ok
1127759224 M * Bertl now let's mount and unmount it, just to make sure everything works as expected
1127759233 M * oliwel ok mom
1127759272 M * mef bertl: before I give you the detailed example, does vc_ctx_create() upon success put the current process into the context?  It seems to do so.
1127759273 M * oliwel ok I have a partition taht can be umount/monuted
1127759307 M * Bertl mef: yes, and depending on the flags, it will end up as init or just as process
1127759307 M * mef kernel code says so.
1127759308 M * mef ok
1127759337 M * Bertl oliwel: okay, you tried with that one ... let's assume you mount it on /mnt, is that fine?
1127759348 M * oliwel ist /mnt/backup
1127759363 M * Bertl okay, let's make another mount point called /mnt/test
1127759374 M * oliwel ok
1127759385 M * Bertl then lets do 'mount --bind /mnt/backup /mnt/test'
1127759399 M * Bertl and after that 'umount /mnt/backup'
1127759400 M * oliwel I assume I should mount the disk before ?
1127759402 M * oliwel ok
1127759407 M * Bertl yeah
1127759449 M * oliwel hmm tahts strange - I successfully un,onuted the original MP but the binded one is still there
1127759467 M * Bertl so, did you mount a device?
1127759470 M * oliwel so /mnt/backup is empty, /mnt/test contains the device
1127759478 M * oliwel yes - sda1 - a usb disk
1127759492 M * Bertl okay, could you remove it?
1127759498 M * oliwel yes
1127759511 M * Bertl what would happen to /mnt/test?
1127759519 M * oliwel umount /mnt/backup is successfull, the dir is emtoy
1127759544 M * oliwel mnt/test still shows the conetct of the hdd
1127759550 M * Bertl is it the only usb disk?
1127759561 M * oliwel yes
1127759570 M * Bertl what does lsmod show regarding module usage?
1127759578 M * Bertl (for usb-storage e.g.)
1127759585 M * oliwel I have the usb modules in the kernel :(
1127759593 M * oliwel running monoliths.
1127759605 M * Bertl hmm, okay .. le maybe a CD rom?
1127759622 M * Bertl (something you could eject)
1127759663 M * oliwel Do we need vserver kernel for this test ? the fileserver is not here in the room - I will test on my workstation ok ?
1127759686 M * Bertl the first few tests, yes, just a 2.6 kernel we need ...
1127759699 A * Bertl sounds like yoda :)
1127759732 A * oliwel gets a cd - by the power of the dark side .)
1127759887 M * oliwel ok I have a CD monuted the same way before - direct on cd1, via bind on cd2 and unmounted cd1
1127759925 M * oliwel same effect - cd1 is empty, cd2 contains the disk
1127760062 M * oliwel Bertl: huhu :)
1127760427 M * oliwel anybody here ?
1127760438 Q * oliwel Quit: Chatzilla 0.9.68.5 [SUSE 1.0.6-4.1/20050715]
1127760456 J * oliwel ~mail-at-o@host-62-245-151-178.customer.m-online.net
1127760640 M * Bertl oliwel: sorry, okay, now try to eject it with eject/button
1127760708 M * oliwel wont work 
1127760719 M * Bertl okay, you got an idea why?
1127760737 M * oliwel sure - its still monuted via the bind
1127760749 M * oliwel I played with vnamespace and I think I did it :)
1127760761 M * Bertl exactly ... so the same happens with namespaces
1127760763 M * oliwel  vnamespace -e 49162 umount /data/wwwoffice
1127760776 M * Bertl yeah, that's it ... 
1127760780 M * mnemoc is namespaces a vserver thing, or vanilla kernel implements it?
1127760791 M * Bertl vanilla kernel .. 
1127760796 M * oliwel works even if I cant see the monut inside the runnig server...but when I run this on every conext I can pull down the drbd
1127760819 M * Bertl mnemoc: vserver just adds the possibility to enter/move between those namespaces
1127760841 M * Bertl oliwel: you can see it in the namespace, not in the context :)
1127760842 M * oliwel Bertl: does the -c (cleanup) flag do the work for me ? On the list there was a note that it isnt working...
1127760863 M * oliwel BertL: How can I list a "namespace" ?
1127760880 M * Bertl the c) part is not implemented (by Enrico) it will probably happen with newer tools ...
1127760886 M * oliwel ahhh "vnamespace -e 49165 mount" - 
1127760893 M * Bertl oliwel: try vnamespace -e 49162 cat /proc/mounts
1127760920 M * oliwel ok so I must somewat "walk" through all my vservers and unmout the non uses mountpoints
1127760981 M * Bertl either that, or unmount them on startup
1127761006 M * Bertl that was the original idea mentioned with c)
1127761013 M * oliwel Bertl: Just to summrize - when I statzup a new vserver it copies the current namespace of the root server, thus it wil lknow all monuts actually done 
1127761030 M * Bertl all mounts currently there, yep
1127761037 M * Bertl (kind of snapshot)
1127761044 M * oliwel un,ount on startup - you mean when I start server A I simply unmount all devices not belonging to A
1127761061 M * oliwel ok I start the understand the problem
1127761076 M * Bertl oliwel: yes, inside the guest, you can simply remove all unwanted mounts (at startup)
1127761100 M * Bertl you could for example, replace the --bind mount, with a --move mount
1127761113 M * oliwel Any good idea how to do ? grep'ing through the proc ?
1127761116 M * Bertl (moving the already mounted device into the guest)
1127761162 M * Bertl oliwel: I'm open to improving this, but it need userspace support ... kernel cannot reliably decide which mounts are wanted and which not ...
1127761170 M * Bertl *needs
1127761178 M * oliwel yes thats clear
1127761185 M * Bertl walking /proc/mounts is a good option, I'd say ...
1127761204 M * Bertl (given that you have a criterion to decide what will be used)
1127761217 M * oliwel Bertl: Thanks for now - my girlfriend has "ebay" time - must share my laptop for some ten minutes....
1127761269 M * Bertl k, guess everything is clear now, right?
1127761286 M * Bertl oliwel: so send greetings to her! and have fun!
1127761321 M * Bertl okay, off for a while ... back later this evening ...
1127761329 N * Bertl Bertl_oO
1127762678 Q * oliwel Quit: Chatzilla 0.9.68.5 [SUSE 1.0.6-4.1/20050715]
1127762711 Q * nokoya Ping timeout: 480 seconds
1127763442 J * nokoya young@hi-230-82.tm.net.org.my
1127764052 J * Blissex ~Blissex@82-69-39-138.dsl.in-addr.zen.co.uk
1127765606 J * oliwel ~mail-at-o@host-62-245-151-178.customer.m-online.net
1127765665 A * oliwel waves hello to the crowd
1127765671 M * daniel_hozac wb.
1127765762 M * oliwel I cant start syslog inside a vserver - any ideas ?
1127765874 M * sannes .. looked in the syslog-ng.conf file?
1127765892 M * sannes it probably references /proc/kmsg which will not have access to inside a vserver ..
1127765911 M * sannes so if you uncomment that line and all related to it, it will probably work ..
1127765922 M * sannes using gentoo by any chance?
1127765935 M * daniel_hozac hmm, /proc/kmsg should be virtualized in recent versions, IIRC.
1127765997 M * Greek0 hi oliwel 
1127766003 M * sannes oh, nice... well it depends on versions then.. I don't know if vprocunhide unhides by defalt though..
1127766020 M * sannes s/defalt/default/
1127766051 M * Greek0 sannes: I think what daniel_hozac means is that it's not hidden by default, but rather patched in a way that no messages go to it if it's mounted inside of a guest
1127766097 M * Greek0 (no, I haven't checked the source for this, but I think I remember Bertl saying something in that direction)
1127766104 M * revenger is there any configuration example out there for a debian system?
1127766140 M * revenger I compiled the kernel and installed it ok.. but I installed the debian binaries afterwards and now I have no idea how to go on
1127766176 M * revenger the "new-howto" just doesn't fit anymore at the installation part
1127766185 M * Greek0 http://linux-vserver.org/Step-by-Step+Guide+2.6
1127766217 M * oliwel sorry was off ...I have removed kmesgm it complains about not opening /dev/log (inside teh vserver) because ist already used
1127766251 M * oliwel Strange: The server runs on one maschine, but not on another and I think all relevnat configs are the same
1127766290 M * Greek0 oliwel: syslogd should actually work fine, klogd probably not
1127766329 M * oliwel yeah it worked quite a while - updated teh servers last week and now have problems :/
1127766447 M * oliwel will move to another place - brb in 10 mins
1127766450 A * oliwel waves goodbye to the crowd
1127766452 Q * oliwel Quit: Chatzilla 0.9.68.5 [SUSE 1.0.6-4.1/20050715]
1127766682 M * revenger Greek0: don't think that is what I'm looking for..
1127766697 M * revenger the host system is a debian system and I already downloaded the redhat iso..
1127766711 M * revenger I'd like to go on at that point ..
1127766759 M * revenger that I also don't get "(Note: newvserver is a linuxconf entry and is NOT a command line utility.)"
1127766789 M * revenger newvserver is a script in /usr/sbin ok but what has that to do with "linuxconf"?
1127767150 M * Greek0 newvserver is deprecated
1127767168 M * Greek0 I don't know if there's a way to install a guest from a downloaded iso image
1127767218 M * Greek0 normally cd installs try to do too much, like trying to setup boot loader, partition harddrives, install kernels, ...
1127767278 M * Greek0 you only need to setup the filesystem.. I'd suggest looking into vserver-build(8), which is unfortunately not shipped in the debian packages
1127767305 M * revenger Greek0: do you know anyone who already installed it on debian?
1127767329 J * oliwel ~chatzilla@host-62-245-151-178.customer.m-online.net
1127767341 M * Greek0 I've got vservers running on a debian host
1127767355 M * mnemoc what does it come from? util-vserver-0.30.208 doesn't include vserver-build.8
1127767361 M * mnemoc s/what/where/
1127767362 M * Greek0 built with something like that: vserver DebianSid build -m debootstrap -- -d sid -m ftp://ftp.at.debian.org/debian/ (taken 1:1 from the step to step howto)
1127767368 M * Greek0 mnemoc: it's in the upstream tarball
1127767381 M * Greek0 no idea why opal hasn't included it. maybe FDL issues, didn't check
1127767406 M * mnemoc Greek0: i got it from here: http://www.13thfloor.at/vserver/s_rel26/v2.0/
1127767413 M * mnemoc Greek0: and than man page is _not_ there
1127767417 M * Greek0 maybe because he was stoned. really no idea. gotta check all those debian issues once I have more time again
1127767472 M * mnemoc btw, anyone using runit inside a vserver?
1127767502 M * revenger Greek0: just a general question is linux vserver the only vserver project out there? google found about 150.000 hits, that site has 6530 visits .. maybe that site isn't that old?
1127767519 M * Greek0 mnemoc: uhm. yep, just saw it. perhaps you've got to build the documentation first
1127767568 M * Greek0 revenger: no, util-vserver is around for quite some time now afaik. there are other virtualisation solutions, but none that carry "vserver" in their name AFAIK
1127767591 M * mnemoc :)
1127767755 M * Greek0 ok, I'm off to bed
1127767756 M * Greek0 cu
1127767879 M * revenger hmm.. isn't it 22:51?
1127768036 M * liquid hi
1127768126 M * liquid newbe question,   wenn using vyum i get the message about insecure and broken in chroot, how can i get rid of that?
1127768305 M * daniel_hozac patch yum.
1127768326 M * liquid how=
1127768328 M * liquid ?
1127768357 M * daniel_hozac there's a patch in the contribs directory in the util-vserver source, IIRC.
1127768601 M * liquid thx, will try that
1127769259 Q * oliwel Quit: Chatzilla 0.9.68.5.1 [SUSE 1.0.6-4.3/20050715]
1127769532 Q * Blissex Remote host closed the connection
1127769777 M * mnemoc rebootmgr init script loops over *.conf ..... is that still valid?
1127769791 M * daniel_hozac rebootmgr is very deprecated.
1127769872 M * mnemoc that's what i thought.... but my runit-based test vserver doesn't stop  and i thought i needed rebootmgr
1127769905 M * daniel_hozac reboot -f from within a guest should do it.
1127769926 M * mnemoc reboot = stop?
1127769930 M * mnemoc should be halt?
1127769968 M * daniel_hozac sure
1127769992 M * mnemoc sure reboot = stop or sure halt = stop?
1127770003 M * daniel_hozac reboot == reboot, halt == halt ;)
1127770007 M * mnemoc :)
1127770050 M * mnemoc what's the way to force the death of vserver's init?
1127770108 M * mnemoc it doesn't die from within
1127770110 M * daniel_hozac kill -9 <pid>?
1127770127 M * daniel_hozac (might want to use vkill for it)
1127770331 M * mnemoc vserver test start
1127770341 M * mnemoc # vserver test enter
1127770346 M * mnemoc everything ok
1127770348 M * mnemoc # init 0
1127770366 M * mnemoc - runit: leave stage: /etc/runit/2
1127770367 M * mnemoc - runit: enter stage: /etc/runit/3
1127770367 M * mnemoc Waiting for getties to stop...
1127770367 M * mnemoc Waiting for services to stop...
1127770367 M * mnemoc Shutdown...
1127770369 M * mnemoc - runit: warning: child failed: /etc/runit/3
1127770371 M * mnemoc - runit: leave stage: /etc/runit/3
1127770374 M * mnemoc - runit: sending KILL signal to all processes...
1127770376 M * mnemoc - runit: power off...
1127770379 M * mnemoc - runit: system halt.
1127770381 M * mnemoc but the thing still running! :(
1127770402 A * mnemoc put his head into his nasty flood
1127770460 Q * mef Remote host closed the connection
1127770865 M * mnemoc vkill worked :)
1127770885 M * mnemoc daniel_hozac: any idea why vserver stop may be triggering a ctrl-alt-supr ?
1127771236 J * revenger_ ~revenger@p54B00D9F.dip0.t-ipconnect.de
1127771656 Q * revenger Ping timeout: 480 seconds
1127771796 P * stefani I'm Parting (the water)
1127772267 J * Aiken ~james@tooax8-100.dialup.optusnet.com.au
1127773156 M * yarihm umm ... which was the page about vserver with the marihuana-background again?
1127773290 M * Aiken http://www.nongnu.org/util-vserver/doc/conf/configuration.html ?
1127773350 M * yarihm yeah, i just found it ... "the great flower page" :) thanks aiken
1127773361 Q * TheSeer Ping timeout: 480 seconds
1127773594 J * TheSeer ~theseer@border.office.salesemotion.net
1127774467 M * yarihm does anyone have an idea why a /etc/vservers/vservername/scripts/pre-start that goes like this will cause the start of the vserver to fail
1127774469 M * yarihm ip rule add from 10.0.0.20 table vservers
1127774469 M * yarihm ip route add default via 10.0.0.1 dev eth0 table vservers >/dev/null 2>&1
1127774469 M * yarihm ip route flush cache
1127774471 Q * TheSeer Ping timeout: 480 seconds
1127774577 Q * Homere Quit: This computer has gone to sleep
1127774608 M * daniel_hozac yarihm: append a exit 0?
1127774621 M * yarihm ok, i'll try
1127774866 J * Homere ~homere@did75-9-82-229-149-96.fbx.proxad.net
1127774874 Q * Homere Quit: 
1127775223 Q * Aiken Quit: Leaving
1127775437 J * Aiken ~james@tooax8-100.dialup.optusnet.com.au
1127775614 Q * nox Quit: If the world is without walls and fences -- who needs Windows and Gates ?
1127776161 J * nox ~nox@noxlux.de
1127776323 J * tuxcapoeira ~capoeira@85-124-167-78.work.xdsl-line.inode.at
1127777032 M * revenger_ can anyone help me with installing vserver? everything's installed so far and now i'm stuck
1127777461 Q * prae Quit: Pwet
1127778428 M * revenger_ /proc/uptime can not be accessed. 
1127778431 M * revenger_ hmm :/
1127778856 M * revenger_ anyone awake?