1182988812 P * stefani I'm Parting (the water)
1182989402 Q * rgl_ Quit: Leaving
1182989478 Q * edeschen Remote host closed the connection
1182989489 J * edeschen ~edeschen@proxy-sjc-1.cisco.com
1182990368 Q * Guest101 Quit: Guest101
1182990488 Q * Radiance Ping timeout: 480 seconds
1182990559 M * edeschen Hi, I have a question regarding file attributes (iunlink), but prepare yourself, 'cause I'm a newbie
1182990613 M * edeschen I'm seeing that if I use the "setattr --iunlink " command, it sets the correct property on the which is a hard link in my vserver
1182990656 M * edeschen I can see with showattr that the link attributes were properly updated (----UIX)
1182990665 M * edeschen everything good so far
1182990677 M * Bertl okay
1182990700 M * edeschen if I unmount the file partition that the whole system is on and remount, those attributes are then reset to -----uiX
1182990721 M * Bertl that is a bug, what kernel and what filesystem?
1182990732 M * edeschen so is this the expected behavior that unlinkable and immutable are reset after remount?
1182990738 M * Bertl (or reiser, without the attribs flag :)
1182990755 M * edeschen I'm using reiserfs 3.6 and vserver 2.0.1
1182990768 M * Bertl and you _did_ specify the attribs mount flag?
1182990774 M * edeschen yeah you guessed right I guess
1182990801 M * edeschen I'm stuck with an old kernel version and that is the only vserver patch I can put on
1182990821 M * Bertl how old is that kernel?
1182990827 M * slack101 Bertl, 200 people :P
1182990845 M * Bertl slack101: well, no, 200 guest with N people each :)
1182990853 M * edeschen kernel version is 2.6.14.3
1182990897 M * slack101 N people ?
1182990903 M * edeschen if that is a bug which was fixed, perhaps somebody can show me the fix and I can port it back
1182990914 M * edeschen or a patch would be great
1182990916 M * Bertl edeschen: yeah, that is quite old, I hope, with all the known exploits, it is quite secured behind a firewall or so
1182990931 M * Bertl edeschen: did you double check the attribs flag I mentioned?
1182990959 M * Bertl slack101: well, an unknown number of people per guest
1182990975 M * edeschen I did not specify an flags, perhaps my mistake, which flag should I use?
1182990988 M * Bertl edeschen: you can easily check with non-Linux-VServer commands/attributes
1182991017 M * Bertl edeschen: try chattr +i /some/file
1182991039 M * Bertl edeschen: and lsattr /some/file, then umount, and mount again
1182991059 M * slack101 Bertl, i seeeeee
1182991095 M * edeschen ok thanx Bertl, I'll try this
1182991098 M * edeschen really appreciated
1182991108 M * Bertl edeschen: you're welcome!
1182991494 J * DoberMann_ ~james@AToulouse-156-1-70-41.w90-16.abo.wanadoo.fr
1182991601 Q * DoberMann[ZZZzzz] Ping timeout: 480 seconds
1182991704 Q * [BiG^BrotheR] Quit: ][DreaM-ScripT][
1182991726 M * edeschen bertl: when using non-linux-vserver chattr on a file on the mounted reiserfs volume, I get an error: chattr: "Inappropriate ioctl for device while reading flags on X"
1182991746 M * edeschen bertl: I must be doing something wrong here
1182991750 M * Bertl so you are missing the proper mount option for reiser
1182991763 M * edeschen ah ok, I'll look for it then
1182991765 M * Bertl as I said, IIRC, it is called attribs
1182991787 M * Bertl (no Idea why reiser needs that :)
1182991790 M * edeschen bertl: sorry missed that, I'll try again
1182992956 Q * Johnnie Ping timeout: 480 seconds
1182993006 M * edeschen Bertl: looks like the "-o attrs" option for mount fix the problem, thanx Bertl, you're the man!
1182993029 M * Bertl great! have fun!
1182993037 M * edeschen thanx again
1182993367 M * Bertl okay, off slightly repartitioning my disk :)
1182993381 N * Bertl Bertl_oO
1182995837 Q * nou Ping timeout: 480 seconds
1182995975 Q * Piet Quit: Piet
1182996559 Q * lylix Read error: Connection reset by peer
1182996677 J * lylix ~eric@dynamic-acs-24-154-33-109.zoominternet.net
1182996869 Q * bzed Quit: Leaving
1182997758 N * Bertl_oO Bertl
1182997767 M * Bertl back now ..
1182998386 Q * JimBobDoe Ping timeout: 480 seconds
1182999099 J * nou Chaton@causse.larzac.fr.eu.org
1182999110 M * Bertl welcome nou!
1182999847 M * Bertl okay, off to bed now ... have a good one everyone!
1182999856 N * Bertl Bertl_zZ
1183000897 J * meandtheshell ~markus@85.127.102.87
1183001141 Q * onox Ping timeout: 480 seconds
1183002413 Q * coderanger_ Ping timeout: 480 seconds
1183003517 Q * weasel Remote host closed the connection
1183003611 J * weasel weasel@asteria.debian.or.at
1183003657 Q * nou Ping timeout: 480 seconds
1183004401 Q * toom Remote host closed the connection
1183004404 J * toom ~toom@coco.docisland.org
1183004515 Q * ktwilight Read error: Connection reset by peer
1183004552 J * ktwilight ~ktwilight@69.81-66-87.adsl-dyn.isp.belgacom.be
1183007137 J * nou Chaton@causse.larzac.fr.eu.org
1183007891 J * xeon_ ~node@d57-2-134.home.cgocable.net
1183008765 N * DoberMann_ DoberMann
1183008980 J * sharkjaw ~gab@158.36.45.236
1183010134 N * DoberMann DoberMann[PullA]
1183010402 Q * edeschen Remote host closed the connection
1183010412 J * edeschen ~edeschen@proxy-sjc-1.cisco.com
1183010731 J * DavidS ~david@p54812C8E.dip0.t-ipconnect.de
1183010786 J * dna ~naucki@p54BCF109.dip.t-dialin.net
1183011413 Q * dna Quit: Verlassend
1183011730 J * onox ~onox@kalfjeslab.demon.nl
1183013340 Q * meandtheshell Quit: Leaving.
1183013610 J * meandtheshell ~markus@85.127.102.189
1183014638 Q * virtuoso Read error: Connection reset by peer
1183014647 J * virtuoso ~s0t0na@80.253.205.251
1183015582 N * DoberMann[PullA] DoberMann
1183017721 M * arekm hm, /etc/vservers/guest/rlimits/resource{,.hard,.min,.soft} - soft and hard are known but what's min?
1183018052 M * arekm I'm setting as:10000 as.hard:15000 rss:5000 rss.hard:7000 but after entering vserver free shows tons of free memory, hmm
1183018508 M * DavidS can youuse it too?
1183018532 M * arekm hm, vlimit doesn't even show that as soft limit exists. (shows N/A)
1183018655 M * arekm [root@pepe /etc/vservers/test1/rlimits]# vlimit -c 100 -S --as 500000
1183018655 M * arekm [root@pepe /etc/vservers/test1/rlimits]# vlimit -c 100 -a -d | grep AS
1183018655 M * arekm AS N/A N/A 1000000
1183018657 M * arekm huh?
1183018828 M * DavidS (10:02:01) arekm: hm, /etc/vservers/guest/rlimits/resource{,.hard,.min,.**soft**}
1183018833 M * DavidS (10:07:32) arekm: I'm setting as:10000
1183018842 M * DavidS i detect a inconsistency here
1183018850 A * DavidS has never used limits imself
1183018855 M * DavidS himself
1183018869 M * arekm DavidS: there is resource file and resource.soft file - both seem to be for "soft" limit
1183018881 M * arekm anyway manual setting via vlimit -c 100 -S --as 500000 also doesn't work
1183019973 M * arekm looks like /proc/meminfo is not virtualized
1183020009 Q * ktwilight Read error: Connection reset by peer
1183020109 M * arekm because rss hard limit seems to be working. well, I can live without as soft limit :)
1183020138 J * ktwilight ~ktwilight@69.81-66-87.adsl-dyn.isp.belgacom.be
1183020339 M * arekm "The rss soft limit is shown inside the guest as the maximum available memory. " but where ;-/
1183020492 J * cedric ~cedric@80.70.39.67
1183020714 J * HeinMueck ~Miranda@host-88-217-199-211.customer.m-online.net
1183021353 M * arekm hm, there is vx_vsi_meminfo() and I guess it should do virtualization of /proc/meminfo, but somehow that doesn't work
1183021788 M * arekm ah, virt_mem flag is needed! :)
1183021800 M * arekm I wonder if virt_cpu works
1183022409 M * arekm no, not used in patch
1183022800 M * baldy vcontext: vc_ctx_create(): File exists... which file exist?
1183022960 M * baldy vsysctl: open("."): Permission denied
1183022961 M * baldy hmh
1183023030 J * lilalinux ~plasma@dslb-084-058-201-084.pools.arcor-ip.net
1183023215 M * baldy ahh works now
1183023485 N * DavidS DavidS|Vechta
1183023756 A * arekm set fill-rate:10 interval:50 tokens:100 tokens-max:200 tokens-min:5
1183023814 M * arekm works nicely :)
1183024524 Q * ||Cobra|| Remote host closed the connection
1183024670 M * arekm hm, where I can check how many tokens are currently in some context?
1183024751 M * arekm cpu 0: 3393 388 163971 4812 0 R- 500 5 500 2/100 1/8 0 0
1183024758 M * arekm I guess 500 is the thing here
1183024839 J * Piet hiddenserv@tor.noreply.org
1183024964 J * ||Cobra|| ~cob@pc-csa01.science.uva.nl
1183025099 M * arekm hm, interval2 and fill-rate2 are refering to the same bucket as interval and fill-rate?
1183025125 M * arekm or there are two buckets: one for normal work and other when advancing idle time?
1183025142 J * bonbons ~bonbons@ppp-111-215.adsl.restena.lu
1183025245 Q * bonbons
1183025251 J * bonbons ~bonbons@2001:5c0:85e2:0:20b:5dff:fec7:6b33
1183026169 J * rgl ~Rui@84.90.10.107
1183026175 M * rgl hello
1183026264 M * onox yo
1183026616 Q * eSa| Ping timeout: 480 seconds
1183026715 Q * Piet Remote host closed the connection
1183026728 J * Piet hiddenserv@tor.noreply.org
1183027083 Q * vserveraddict Ping timeout: 480 seconds
1183028080 J * [BiG^BrotheR] ~Dr-DreaM@84.23.96.253
1183028210 M * onox omg, ottawa is in canada
1183028430 N * Bertl_zZ Bertl
1183028437 M * Bertl morning folks!
1183028452 M * onox how you're do'ing?
1183028931 M * Bertl onox: fine, tx!
1183028957 M * Bertl arekm: one bucket, two different settings (they are switched when going idle)
1183029149 M * mattzerah hi Bertl
1183029348 M * mattzerah i think i'm going to use that kernel rpm as it is
1183029411 M * mattzerah i just want to do a recompile to make sure everything is good, and a test install on a fresh box to make sure it works as expected, then i'll move it to release -1
1183029411 M * mattzerah :)
1183029422 M * Bertl okay
1183029476 M * mattzerah i've re-compiled the dietlib and util-vserver into rpms as well so i'll move them to my server when i get a chance as well
1183029486 M * mattzerah everythnig seems to be working okayish :)
1183029569 J * ktwilight_ ~ktwilight@223.115-66-87.adsl-dyn.isp.belgacom.be
1183029575 M * mattzerah out of curiosity - the /etc/vserver/ config structure, is that specific to util-vserver ?
1183029597 M * mattzerah as far as getting guests working, you just need a utiltiy to talk to the kernel yes ?
1183029607 M * Bertl yep
1183029620 M * mattzerah and i'm guessing util-vserver uses that directory structure as its way of working out what to do
1183029632 M * Bertl vmcd for example should be able to do the talking :)
1183029644 M * mattzerah very cool
1183029650 M * mattzerah how is vmcd coming along ?
1183029655 M * mattzerah any idea ?
1183029661 M * Bertl it's perfect!
1183029669 M * mattzerah really? i thought it was still in devel
1183029671 M * mattzerah mmmm
1183029680 M * mattzerah i might check it out, compare the two :)
1183029683 M * Bertl but no replacement for util-vserver or vcd + friends
1183029701 M * mattzerah would they be able to work side-by-side - i guess only for different guests
1183029703 M * Bertl mattzerah: vcmd is just a hack tool to excercise the API
1183029710 M * mattzerah ahh, okay
1183029788 M * mattzerah i might try vcd out as well, just for a sticky beak :)
1183029978 Q * ktwilight Ping timeout: 480 seconds
1183031019 J * bulkBoy ~eSa|@ip-87-238-2-45.adsl.cheapnet.it
1183031119 M * Bertl wb bulkBoy!
1183031185 M * bulkBoy hi Bertl
1183031379 M * awk hello, nurse!
1183031594 J * weasel_ weasel@asteria.debian.or.at
1183031594 Q * weasel Read error: Connection reset by peer
1183031605 N * weasel_ weasel
1183031967 M * rgl Bertl, how was the conference?
1183031981 M * Bertl which one?
1183031998 M * rgl you ddn't make a presentation yesterday?
1183032023 M * Bertl you are talking of linuxsymposium, I presume?
1183032073 M * rgl someone told you were making a presentation, but I don't recall the place :-(
1183032111 M * Bertl ah, well, no presentation as far as I can remember :)
1183032181 M * rgl Bertl, http://www.linuxsymposium.org/2007/view_abstract.php?content_key=223 ?
1183032249 M * Bertl yep, I planned to give a presentation there, but the flight would have been roughly 1200 EUR and nobody there was able to help me funding that somehow
1183032305 Q * weasel Remote host closed the connection
1183032307 J * weasel weasel@asteria.debian.or.at
1183032311 M * rgl oh sux :(
1183032324 M * Bertl but my co-author is there ... so not that bad for the audience
1183032408 M * Bertl yeah, the bad part is that they expect a free software developer to cover travel and accomodation consts
1183032411 M * Bertl *costs
1183032452 Q * [BiG^BrotheR] Ping timeout: 480 seconds
1183032460 M * Bertl (I had expected them to know better ...)
1183032464 M * rgl which is odd :|
1183032505 M * rgl I gues they should support. but then again, I have nil experience on this :|
1183032550 M * rgl I hope they will unleash the presentations online, though.
1183032569 M * Bertl well, they didn't even bother to answer to my emails (although they kept sending me mails regarding coordination :)
1183032582 M * DavidS|Vechta sounds very lame
1183032606 M * Bertl yes, as I said, I'd expected more of 'the linux event' :)
1183032650 M * Bertl but probably the list of papers already gives an idea ...
1183032653 M * Bertl http://www.linuxsymposium.org/2007/speakers.php?types=TALK
1183032685 M * Bertl I see like 4-5 folks who are not Google, IBM, Intel ...
1183032706 J * pmenier ~pmenier@LNeuilly-152-22-72-5.w193-251.abo.wanadoo.fr
1183032802 J * [BiG^BrotheR] ~Dr-DreaM@84.23.96.253
1183033026 M * rgl commercial "players" :|
1183033123 M * DavidS|Vechta any quick ideas how to share one http port amongst multiple virtual hosts, each in its own vserver without loosing the client infos (like ip)?
1183033140 M * Bertl proxy?
1183033204 M * Bertl i.e. setup an http server on the host, which handles the vhosts and delegates the request ...
1183033219 M * DavidS|Vechta i have that running already, but all client logs only show the proxy ip .. which is not quite satisfactory ...
1183033349 M * Bertl then your setup is not complete
1183033367 M * Bertl i.e. your 'proxy' is not sending any client IP headers
1183033389 M * Bertl and/or your final servers are not using that info
1183033396 A * DavidS|Vechta has to dig deeper into apache's mod_proxy
1183034046 Q * Piet Ping timeout: 480 seconds
1183034081 M * rgl its good ideia to run the proxy on the host? why not on a guest?
1183034126 M * DavidS|Vechta sloppiness of wording i presume
1183034205 M * Bertl you can put it wherever you want :)
1183034376 M * rgl I see the host, has the master or the world, where nothing but essencial stuff for managing the guests is run. but probably you guys don't think like this?
1183034403 M * Guy- hi
1183034430 M * Guy- Bertl: I believe you wanted to make a comment about parent processes being able to send signals to their children across vserver exec, when I wasn't around?
1183034452 M * Bertl rgl: that would be the ideal setup, but reality tends to differ quite significantly :)
1183034474 M * Bertl Guy-: yep, basically your usage pattern was like this:
1183034482 M * DavidS|Vechta Bertl <<-- master of understatement ;)
1183034484 M * Bertl vserver - enter/exec something
1183034490 M * rgl Bertl, pulease, tell me were you have diverged? :D
1183034509 M * Bertl Guy-: then you send STOP/CONT
1183034562 M * Bertl this results in signals sent to/from the session leader, which would be across guest boundaries
1183034590 M * Bertl now the question is, why is that blocked? for several reasons:
1183034610 M * Bertl - first you don't want to allow guests to send signals to host processes
1183034629 A * arekm wonders if vserver stop work even if there are no free tokens in guest bucket? will something like forced kill -9 for guest happen after some timeout?
1183034634 M * Bertl - second, you don't want host scripts to 'accidentially' kill guest processes :)
1183034644 M * Guy- Bertl: I'm not sure I follow
1183034652 M * Bertl arekm: nope
1183034665 M * Guy- Bertl: this would not have involved a guest process sending a signal to a host process, but vice versa
1183034681 M * Bertl arekm: it is a good idea to set rate/interval so that a minimal amount of tokens is given over longer time
1183034682 M * Guy- Bertl: the session leader was my login shell on the host, no?
1183034715 M * Guy- Bertl: also, if host scripts 'accidentally' kill guest processes, that's a bug that should be fixed, not masked, imho :)
1183034721 M * Bertl Guy-: it is, but the setup is flawed, as the session leader is _outside_ and the processes are inside
1183034736 M * arekm Bertl: that's still problematic if there is bad customer (tm) who runs 1k of cpu intensive tasks to make vserver stop harder for the admin ;)
1183034739 M * Guy- yes, but that's just the way it's going to be with vserver exec
1183034745 M * Bertl Guy-: if you want this kind of setup to work, use a proxy, like ssh/rsh/telnet
1183034760 M * Guy- Bertl: it's really not about ctrl-z in vserver exec
1183034766 M * Bertl arekm: vkill hepls there :)
1183034790 M * Bertl Guy-: but?
1183034790 M * Guy- Bertl: what I'd like to do is run 'vserver foo exec somedaemon' under runit
1183034801 M * Bertl Guy-: so?
1183034825 M * Guy- Bertl: this involves a 'runsv' process spawning a child and execing vserver foo exec somedaemon
1183034843 M * Guy- Bertl: without vserver, I'd be able to tell the runsv process to send its child signals
1183034860 M * Guy- Bertl: like 'sv hup somedaemon' (this avoids the PID lotto stuff you see in initscripts)
1183034864 M * Bertl you have to adjust runit to send the singlas properly
1183034868 M * Bertl *signals
1183034888 M * Guy- well, yes, that's another way to address this problem, certainly
1183034901 M * arekm Bertl: right, more problematic in mostly automated env but well ..
1183034903 M * Guy- daniel_hozac wrote a simple wrapper called signal-relay that mostly does what I need
1183034909 M * Bertl Guy-: no, that is _the_ way to address this problem :)
1183034911 M * Guy- but obviously it can't relay the kill signal :)
1183034921 M * Bertl arekm: how so?
1183034970 M * arekm Bertl: well, another special case to handle :-) exec vserver stop, keep some timeout in other thread/process that will do vkill if vserver stop still hangs etc
1183034972 M * Guy- Bertl: I'm not so sure :) runit wants to be cross-platform, whereas vserver is, at least for now, a niche thing; I don't think the author of runit would want to support it explicitly
1183034978 M * Bertl Guy-: alternatively you can modify Linux-VServer to allow host->guest signals without further checking, but you should make sure that your scripts never send signals by accident :)
1183035003 M * Guy- Bertl: daniel_hozac came up with a patch that allowed such signals to be sent provided the sending process was the parent of the recipient
1183035016 M * Guy- Bertl: I thought about it for a while but I really can't see what's wrong with this
1183035049 M * Guy- Bertl: I think a parent process ought to be allowed to send signals to its children
1183035065 M * Guy- even if said children happen to be running in a different context now
1183035070 M * Bertl correct, but those are not children of that parent anymore
1183035084 M * Bertl they will, for example, be reaped by a guest side reaper
1183035110 M * Guy- well, vps axfu seems to think they are
1183035124 M * Bertl note: you can't have both, context isolation and parent-child relation
1183035132 M * arekm is interval == 1 considered crazy thing to use?
1183035150 M * Guy- Bertl: daniel's patch boils down to this:
1183035153 M * Guy- - if (!vx_check(vx_task_xid(t), VS_WATCH_P|VS_IDENT)) {
1183035155 M * Guy- + if (t->real_parent != current &&
1183035158 M * Guy- + !vx_check(vx_task_xid(t), VS_WATCH_P|VS_IDENT)) {
1183035164 M * Guy- are you saying this real_parent test is not going to work?
1183035212 M * Guy- also, surely the parent (running on the host) receives a SIGCHLD when the child (running in the guest) dies?
1183035389 J * ema ~ema@rtfm.galliera.it
1183035451 M * arekm another question, soft rss limit. From what I saw somewhere on wiki when process eats more memory than there is soft rss limit then it has some extra priority in what OOM killer will choose
1183035470 M * Bertl Guy-: that really depends ... the problem is, that 'vserver - exec' is already a gross hack
1183035481 M * arekm this means standard linux OOM killer right? not some vserver specific incarnation?
1183035498 M * Bertl Guy-: if you want proper context isolation, you should avoid that ...
1183035541 M * Bertl arekm: there is a Linux-VServer badness calculated, nothing else is adjusted
1183035562 M * Guy- Bertl: maybe I don't want "proper" isolation
1183035575 M * Guy- Bertl: just a chroot on steroids :)
1183035583 M * Bertl Guy-: then feel free to adjust the sources :)
1183035587 M * Guy- right :)
1183035597 M * Guy- Bertl: any thoughts on that patch?
1183035609 M * arekm Bertl: I mean that OOM killer will be run only if number of free memory pages is low on entire host not only in guest. is this correct?
1183035620 M * Bertl Guy-: well, if it works? :)
1183035634 M * Bertl arekm: yep
1183035699 M * arekm Bertl: ok. new question. why free shows some swap ? :-) (rss soft and hard limit sets, as hard, too)
1183035711 M * arekm I don't remember anything about swap on wiki hmm
1183035723 M * Guy- Bertl: I don't know if it works, I haven't tried it yet :) do you see anything fundamentally wrong with it?
1183035727 M * Bertl arekm: rss.hard-rss.soft is shown as swap
1183035776 M * arekm Bertl: ah
1183036096 M * Guy- Bertl: anyway, I just filed a wishlist item against runit, just in case :)
1183036107 M * Bertl can't hurt :)
1183036595 Q * sharkjaw Quit: Leaving
1183036763 Q * nou Ping timeout: 480 seconds
1183036843 J * nou Chaton@causse.larzac.fr.eu.org
1183037567 Q * Aiken Quit: Leaving
1183038305 M * arachnist hmm
1183038459 M * arachnist any ideas? http://phpfi.com/245553 util-vserver 0.30.213, debootstrap 1.0.0, host is gentoo
1183038492 M * Bertl missing arch
1183038502 M * arachnist how do i set arch?
1183038518 M * Bertl try to append -- --arch i386
1183038555 M * Bertl (or set the ARCH environment var)
1183038578 M * arachnist /usr/lib64/util-vserver/vserver-build: unrecognized option `--arch'
1183038593 M * Bertl you didn't miss the '--', did you?
1183038607 M * arachnist exporting ARCH did the trick
1183038662 M * arachnist and no, i didn't forget about --
1183038700 Q * [BiG^BrotheR] Quit: ][DreaM-ScripT][
1183038708 M * daniel_hozac then your getopt is probably broken.
1183038708 J * MonkeyDog ~satan@85.138.105.46
1183038728 M * Bertl wb MonkeyDog!
1183038744 M * MonkeyDog hello
1183039421 Q * xeon_
1183040269 M * Bertl okay, off for a nap ... back later ...
1183040277 N * Bertl Bertl_zZ
1183041658 J * vserveraddict ~vserverad@LPuteaux-151-41-11-129.w217-128.abo.wanadoo.fr
1183042129 J * hallyn_ ~xa@adsl-75-0-151-125.dsl.chcgil.sbcglobal.net
1183042484 Q * hallyn Ping timeout: 480 seconds
1183042605 M * onox Bertl_zZ.notify()
1183042619 M * daniel_hozac i doubt that works :P
1183042649 J * sauron ~zeus16384@12.20.4.100
1183042657 M * onox daniel_hozac: vserver-copy doesn't really work
1183042678 Q * HeinMueck Quit: Aah!
1183042688 M * daniel_hozac right, use vserver ... build -m clone
1183042759 M * arekm clone huh
1183042799 M * onox daniel_hozac: can I use --context, --interface, --hostname, and specify the name of the new guest?
1183042813 M * daniel_hozac you have to, yes :)
1183042927 M * onox how?
1183042931 M * onox --name Y?
1183042960 M * daniel_hozac vserver build -m clone --context.... -- --source
1183042979 M * onox thx
1183042980 M * onox btw
1183043003 M * onox what's the purpose of vserver-copy when build -m clone must be used
1183043005 M * onox ?
1183043013 M * daniel_hozac it's a legacy program.
1183043031 M * onox hmm
1183043034 M * daniel_hozac it operates on /etc/vservers/.conf guests.
1183043055 M * onox i have /etc/vservers//
1183043061 M * daniel_hozac you'll have to blame your distribution for packaging that stuff along with the rest of the package ;)
1183043074 M * onox Hollow, phreak!
1183043085 M * onox report immediately!
1183043136 Q * sauron
1183043364 J * DavidS ~david@p54812066.dip0.t-ipconnect.de
1183043369 M * rgl it was Hollow that went to linuxsymposium ?
1183043374 N * DavidS DavidS|Vecht1
1183043385 N * DavidS|Vecht1 DavidS
1183043386 M * onox don't know
1183043415 M * onox but Hollow and phreak must fix test failure + vserver-copy, otherwise i'm gonna cry like a baby
1183043462 M * daniel_hozac i don't think Hollow is at OLS.
1183043470 N * DavidS|Vechta Guest150
1183043470 N * DavidS DavidS|Vechta
1183043793 Q * Guest150 Ping timeout: 480 seconds
1183046149 J * dna ~naucki@p54BCD760.dip.t-dialin.net
1183046296 J * stefani ~stefani@tsipoor.banerian.org
1183046519 M * arekm context 1 is "read-only" ? I'm looking for a way to patch rc.shutdown to kill all processes (including these in different contexts) when doing shutdown/reboot of host machine
1183046556 Q * ensc Ping timeout: 480 seconds
1183046576 M * daniel_hozac have you seen the util-vserver initscript?
1183046623 J * Radiance 84b4506a87@halt.1984world.eu
1183046640 M * arekm that one I have just stops configured vservers
1183046762 M * daniel_hozac ah, Debian?
1183046822 M * arekm no, pld
1183046860 M * daniel_hozac well, you'd have you to ask them then.
1183046893 M * daniel_hozac the vanilla util-vserver initscript will kill all running guests.
1183046908 M * daniel_hozac or, rather, contexts.
1183046920 N * Bertl_zZ Bertl
1183046934 M * Bertl back now ...
1183046974 P * edeschen Leaving
1183047036 M * Bertl onox: I guess you already have everything you needed?
1183047078 N * DoberMann DoberMann[PullA]
1183047123 M * arekm daniel_hozac: ok, btw. thanks for help yesterday. I was today playing with hard scheduling - works nicely :)
1183047193 M * Bertl good to hear!
1183047672 Q * FloodServ Service unloaded
1183047695 M * cehteh anyone of you know if there is a way to mmap a file but tell the kernel that the backing file is dirty and i am only interested in writing it
1183047696 J * FloodServ services@services.oftc.net
1183047711 M * cehteh prevent initial page in
1183047762 M * Bertl cehteh: you mean like opening it write only?
1183047790 M * cehteh does that suffice?
1183047791 M * DavidS|Vechta O_TRUNC is probably not the proper solution?
1183047806 M * cehteh no pages in the middle of a file :)
1183047807 M * harry no comments on the newest grsec+vserver patch?
1183047808 M * harry odd...
1183047816 M * harry but i hope that's a good sign :)
1183047822 M * harry means i didn' make mistakes :)
1183047832 M * Bertl harry: usually no feedback is good feedback :)
1183047929 M * cehteh Bertl: i think i try that .. mhm strange hack
1183047958 M * harry in that case: wiiii :)
1183047970 A * harry off again...
1183048142 Q * bulkBoy Quit: Leaving
1183048338 Q * lilalinux Remote host closed the connection
1183048420 M * slack101 Bertl, what distro do you use for your host server ?
1183048452 J * eSa| ~kvirc@ip-87-238-2-45.adsl.cheapnet.it
1183048924 J * bzed zeimetz@ultra14.rbg.informatik.tu-darmstadt.de
1183049435 Q * vserveraddict Read error: Connection reset by peer
1183049977 Q * pmenier Quit: pmenier
1183050024 M * Bertl slack101: Mandriva
1183050746 Q * eSa| Quit: KVIrc 3.2.4 Anomalies http://www.kvirc.net/
1183050775 J * Federico3 ~fede@tor-irc.dnsbl.oftc.net
1183050885 M * Bertl welcome Federico3!
1183051429 Q * meandtheshell Quit: Leaving.
1183052092 M * Federico3 hi Bertl
1183052243 Q * cedric Quit: cedric
1183052449 J * sid3wind1 luser@bastard-operator.from-hell.be
1183052470 M * Bertl wb sid3wind1!
1183052532 Q * sid3windr Ping timeout: 480 seconds
1183052883 Q * sid3wind1 Read error: Connection reset by peer
1183052899 J * sid3windr luser@bastard-operator.from-hell.be
1183053395 Q * ema Quit: leaving
1183053457 Q * Federico3 Remote host closed the connection
1183053458 Q * FloodServ synthon.oftc.net services.oftc.net
1183053591 M * DavidS|Vechta just as a heads up: I'm currently working on a little puppet/vserver integration ... so if someone is looking into configuration automation, http://puppet.reductivelabs.com/ is the way to go :)
1183053609 M * Bertl nice
1183053763 J * FloodServ services@services.oftc.net
1183053945 J * ensc ~irc-ensc@p54B4EBA0.dip.t-dialin.net
1183054101 J * jakesol ~richardso@adsl-074-186-026-107.sip.mia.bellsouth.net
1183054143 M * jakesol hello, where can I change the binding to not run at INADDR_ANY
1183054162 M * sid3windr binding of?
1183054168 M * Bertl jakesol: hmm?
1183054190 M * jakesol binding of the ports
1183054213 M * daniel_hozac that depends entirely on the application...
1183054221 M * jakesol i have a vserver with 5 guests and installed a service at all guests and root
1183054229 M * jakesol its only listening at the root context
1183054240 M * Bertl inside a guest, sockets binding on INADDR_ANY will only 'bind' to guest IPs
1183054267 M * Bertl on the host, if you really must run that services, you have to restrict it to host 'only' ips
1183054313 M * jakesol where that change is made?
1183054321 M * Bertl which change?
1183054344 M * jakesol where I can let the other guests listen on that port
1183054364 M * DavidS|Vechta jakesol: in the config of the service running in the host
1183054369 M * Bertl hmm, looks like we have a communication problem here ...
1183054389 M * Bertl jakesol: let's go through this step by step
1183054394 M * sid3windr :)
1183054400 M * Bertl jakesol: what application or service?
1183054403 J * HeinMueck ~Miranda@dslb-088-065-244-031.pools.arcor-ip.net
1183054404 M * jakesol nrpe
1183054418 M * jakesol listening on port 5666
1183054427 M * Bertl okay, some nagios addon?
1183054441 M * jakesol yes, for monitoring resources
1183054455 M * Bertl okay, so that 'binds' to *:5666
1183054463 M * Bertl yes?
1183054471 M * jakesol yes and it uses the xinetd service
1183054495 M * Bertl okay, and you want to have it running on the host too, right?
1183054512 M * Bertl (not just inside the guests)
1183054515 M * jakesol i wanted at the root context and at the guest context
1183054516 M * arekm (or works as daemon, nrpe)
1183054535 M * Bertl jakesol: okay, we call the 'root context' HOST
1183054558 M * Bertl jakesol: your host probably has certain IP(s) not used by guests, yes?
1183054686 M * jakesol yes
1183054702 M * jakesol the host ip its different from the ips used by the guests
1183054732 M * Bertl okay, do you know how to configure xinetd to use only certain ips?
1183054779 M * jakesol the xinetd.conf of the HOST has the entry bind = 1.1.1.1 (the ip of the host)
1183054807 M * jakesol the xinetd.conf of the guest has no bind entry
1183054819 M * Bertl okay, that looks good
1183054825 Q * DavidS|Vechta Quit: Leaving.
1183054873 M * Bertl the bind is in the 'defaults' section, yes?
1183054896 M * jakesol yes
1183054906 M * Bertl okay, did you restart xinetd?
1183054926 M * jakesol when restarting xinetd at the guest i get the message bind failed (Address already in use (errno = 98)). service = nrpe
1183054941 M * Bertl forget the guest for now ... only the host is interesting
1183054952 M * jakesol ok
1183055018 J * esa bip@ip-87-238-2-45.adsl.cheapnet.it
1183055074 N * esa eSa|
1183055118 M * Bertl after the restart, let's check with 'lsof -ni :5666'
1183055164 M * jakesol i restarted xinetd at the HOST and got the following
1183055171 M * jakesol nrpe 7696 nagios 4u IPv4 978587069 TCP *:nrpe (LISTEN)
1183055193 M * Bertl okay, so the bind is ignored by xinetd
1183055205 M * Bertl try to add the bind line to the service itself
1183055219 M * Bertl ah, no, wait
1183055228 M * Bertl maybe first try to kill off the nrpe
1183055241 M * Bertl (might have stayed around)
1183055250 M * jakesol ok
1183055275 M * jakesol Shutting down Nagios NRPE daemon (nrpe): [ OK ]
1183055302 M * Bertl hmm, that sounds like a runlevel script?
1183055330 M * jakesol i ran /etc/xinet.d/nrpe stop
1183055335 M * Bertl ah, okay
1183055356 M * Bertl let's restart xinetd now and look again
1183055374 M * jakesol Stopping xinetd: [ OK ]
1183055374 M * jakesol Starting xinetd: [ OK ]
1183055404 M * jakesol the lsof out now has the ip
1183055411 M * jakesol xinetd 27705 root 5u IPv4 978636037 TCP 192.168.2.10:nrpe (LISTEN)
1183055425 M * Bertl excellent, so that means it is now restricted to the (host) ip
1183055440 M * Bertl now you can start the service inside the guests too
1183055452 M * Bertl (no restriction is required there)
1183055486 M * jakesol hum, now its working
1183055496 M * Bertl as I said :)
1183055513 Q * FloodServ synthon.oftc.net services.oftc.net
1183055549 M * jakesol tcp 0 0 host.domain.com:nrpe *:* LISTEN
1183055561 M * jakesol thats the netstat output from the guest
1183055574 M * Bertl yep, looks good
1183055613 M * Bertl what kernel/patches do you use?
1183055655 M * jakesol 2.6.14.3-vs2.0.1-rc5
1183055676 M * jakesol thanks for your help
1183055682 M * Bertl you might want to schedule an update sooner or later
1183055688 M * Bertl jakesol: you're welcome!
1183055693 M * jakesol thanks
1183055701 M * daniel_hozac sooner rather than later... that's ancient.
1183055705 M * daniel_hozac why such an old kernel?
1183055729 M * Bertl sooner == yesterday, later = this evening :)
1183055733 Q * eSa| Quit: Coyote finally caught me
1183055738 M * daniel_hozac hehe.
1183055745 J * esa bip@ip-87-238-2-45.adsl.cheapnet.it
1183055751 N * esa eSa|
1183055870 J * |eSa|| ~kvirc@ip-87-238-2-45.adsl.cheapnet.it
1183055997 Q * jakesol
1183056032 Q * |eSa||
1183056036 J * |eSa|| ~kvirc@ip-87-238-2-45.adsl.cheapnet.it
1183056639 Q * |eSa|| Quit: KVIrc 3.2.4 Anomalies http://www.kvirc.net/
1183056694 J * FloodServ services@services.oftc.net
1183057689 J * zLinux ~zLinux@88.213.31.27
1183057692 Q * dilinger Ping timeout: 480 seconds
1183058190 M * rgl humm, what are the prepatch versions of vserver?
1183058224 M * daniel_hozac what do you mean?
1183058228 M * rgl oh, its the same as in kernel.org... nm *G*
1183058229 J * mire ~mire@145-168-222-85.adsl.verat.net
1183058262 M * rgl like, alpha versions of the vserver patch?
1183058292 M * daniel_hozac usually beta.
1183058456 M * rgl the http://linux-vserver.org/ChangeLog-2.2 does not mention them. is that right, or the prepatch does not belong there?
1183058502 M * daniel_hozac you're referring to 2.6.21.5-vs2.2.0-rc3, yes?
1183058519 M * daniel_hozac the reason it's not on there is because i haven't looked at it enough to know what changed ;)
1183058589 M * rgl yup :D
1183058632 M * rgl ok :)
1183058788 M * rgl the http://kernel.org/pub/linux/kernel/v2.6/ChangeLog-2.6.21 refers to the whole changes from 2.6.20 to 2.6.21 ?
1183058795 M * daniel_hozac yes.
1183059078 M * rgl woah, the compressed source increased from 40M on 2.6.0 to 53M on 2.6.21.5 *G*
1183059093 M * rgl thats like. a lot! *G*
1183059094 M * daniel_hozac .tar.gz?
1183059097 M * rgl yup
1183059118 M * Bertl not really unexpected
1183059146 M * Bertl 90% of the kernel sources are drivers ... and guess what sells best? :)
1183059168 M * sid3windr megabytes
1183059197 M * sid3windr encarta 2000 (or so) had a total size of 600M but shipped split into 3 CD's
1183059201 M * sid3windr more CD's sells.
1183059267 J * dilinger ~dilinger@mail.queued.net
1183059270 M * rgl heh
1183059314 M * rgl Bertl, what sells best? storage/megabytes?
1183059327 M * sid3windr Bertl meant hardware support. :P
1183059355 M * Bertl wb dilinger!
1183059360 M * Bertl sid3windr: tx! :)
1183059375 M * rgl ah yes. more HW support is a good thing (I guess *G*)
1183059404 M * rgl but how about software support, like, vserver? ;-)
1183059420 M * Bertl doesn't sell at all :)
1183059422 A * rgl hides
1183059448 M * sid3windr =)
1183060275 J * gerrit ~gerrit@204.50.130.82
1183061283 Q * rgl Ping timeout: 480 seconds
1183061853 J * fb ~fback@red.fback.net
1183061880 M * fb good evening
1183061883 M * Bertl welcome fb!
1183061955 M * fb i've just noticed a strange problem with vserver
1183061970 M * fb three vcontext processes eats 100% cpu
1183061980 M * Bertl does it affect Linux-VServer too? (just kidding :)
1183061999 M * daniel_hozac fb: update to 0.30.213.
1183062004 M * Bertl fb: I assume you are running older tools (< 0.30.213)
1183062029 M * fb well, it used to work for about a year without touching anything :)
1183062054 M * Bertl and then you did what?
1183062055 M * sid3windr welcome to timebombing your software!
1183062152 M * fb updated vserver patch from 2.2.0-someRC to "stable" version
1183062186 M * daniel_hozac i doubt that's the trigger...
1183062192 M * fb thanks for the hint, i'll check with newer vserver tools
1183062221 M * Bertl FWIW, it's a known issue with older tools
1183062228 M * fb daniel_hozac: that's all i've changed
1183062259 M * daniel_hozac it's a user space bug.
1183062274 M * fb Bertl: upgrade tools and restart vservers, that's all?
1183062281 M * daniel_hozac no need to restart the guests.
1183062292 M * daniel_hozac just kill off the vcontext processes.
1183062300 M * daniel_hozac and they should be gone for good.
1183062332 M * fb i'm pretty sure they were not started with guest servers
1183062361 M * daniel_hozac what?
1183062406 M * fb when i started guest servers, they did not appear at vtop
1183062423 M * daniel_hozac nope.
1183062431 M * daniel_hozac they would appear after vserver ... enter.
1183062439 M * fb they appeared some time later, i'm sure about this
1183062440 M * daniel_hozac (which wasn't terminated nicely)
1183062479 M * fb and with many running guests there are "only" three cpu-consuming
1183062499 M * Bertl that's the tricky part about races :)
1183062516 M * daniel_hozac and as i said, only happens for vserver ... enter.
1183062531 M * daniel_hozac normally you wouldn't be doing that to all of your guests that often...
1183062552 M * Bertl fb: you can trust the man, he is writing those tools :)
1183062611 M * fb thanks for the hints :)
1183062651 J * rgl ~Rui@84.90.10.107
1183062651 M * daniel_hozac well, technically, Hollow wrote vlogin ;)
1183062724 Q * HeinMueck Quit: Aah!
1183062741 M * Hollow daniel_hozac: what was the cause for the race again?
1183062775 M * daniel_hozac for some reason vlogin wouldn't get signalled when the parent died/terminal vanished, and we weren't detecting EOF.
1183062799 M * Hollow ah, EOF, i rememeber :)
1183062826 M * Hollow i'm confused since i'm looking at this weired language called "scheme" ;)
1183062846 M * Bertl hehe, lol
1183062859 M * Hollow clearly too many braces.. :)
1183062862 M * daniel_hozac hehehe
1183062864 M * Bertl so you never wrote a gimp script-foo before?
1183062874 M * Hollow no, i have photoshop installed on my mac ;)
1183062892 M * Bertl and there you are using what script language?
1183062904 M * Hollow none, i rarely open photoshop
1183062914 M * Hollow and i have no graphics apps on linux
1183063203 M * slack101 anyonne have that virtuatables script anywhere ?
1183063243 M * slack101 looks like it down on both forums
1183063253 M * slack101 im sure someone has it saved somewhere though
1183063299 J * HeinMueck ~Miranda@dslb-088-065-244-031.pools.arcor-ip.net
1183063480 Q * HeinMueck
1183063557 J * coderanger_ ~laptop@wireless-88.media.mit.edu
1183064039 Q * coderanger_ Ping timeout: 480 seconds
1183064634 M * fb one more question, is it possible to use host's ipv6 interface inside guest with nodev option without additional patches?
1183064644 M * daniel_hozac no.
1183064664 M * daniel_hozac without the IPv6 patch, IPv6 is completely disabled within guests.
1183064720 M * fb do you think it's stable enough to use on production machine?
1183064748 M * daniel_hozac it should be, but i haven't had the luxury of using it yet.
1183064897 M * fb uh
1183064928 M * fb no patch against 2.6.19.7 with vserver-2.2.0.
1183064949 M * daniel_hozac latest should be fine, IIRC.
1183065000 M * fb daniel_hozac: i'll try, btw i was not able to compile latest release for 2.6.20.x
1183065090 M * daniel_hozac hmm?
1183065093 M * daniel_hozac why not?
1183065131 M * Bertl fb http://vserver.13thfloor.at/Experimental/patch-2.6.20.14-vs2.2.0.diff ?
1183065145 M * fb compile error, give me few moments i'll reproduce it
1183065166 M * fb Bertl: the one with grsec
1183065183 M * daniel_hozac hmm, i thought harry never did a release for 2.6.20.
1183065192 M * Bertl fb: ah, well, for that one harry is responsible ...
1183065193 M * fb 2.6.21.5
1183065199 M * fb sorry :)
1183065200 M * daniel_hozac so that's not 2.6.20.
1183065383 M * Bertl btw, do you know, does 2.6.21.5 now contain a fix for the sysfs issues?
1183065397 M * Bertl (I somehow lost completely track of that)
1183065419 M * daniel_hozac i don't think so, but i'm really not sure.
1183065700 J * [BiG^BrotheR] ~Dr-DreaM@84.23.96.253
1183065790 M * fb can i paste two lines here?
1183065815 M * daniel_hozac probably better to use paste.linux-vserver.org, but two lines is fine....
1183065822 M * fb security/commoncap.c: In function 'cap_netlink_send':
1183065822 M * fb security/commoncap.c:32: error: 'struct task_struct' has no member named 'gr_cap_rtnetlink'
1183065826 M * fb that's it
1183065841 M * Bertl harry: ping!
1183065870 M * daniel_hozac he finally got his feedback :)
1183065910 M * Bertl yep
1183065911 Q * dna Quit: Verlassend
1183066068 M * fb and ipv6 patch falis both on 2.6.19.7 and 2.6.21.5, not only on mentioned Makefile
1183066086 M * daniel_hozac 2.6.21 is expected...
1183066091 M * fb but i don't feel like i can repair this
1183066205 M * fb i can paste where it fails, if anybody interested
1183066229 M * Bertl put it on paste.linux-vserver.org and leave the url for harry
1183066287 M * fb he's also responsible for ipv6?
1183066312 M * Bertl daniel_hozac: FYI, sysfs patch still applies without issues
1183066320 M * daniel_hozac Bertl: how far off is a 2.3 patch with IPv6?
1183066326 M * daniel_hozac okay, that's... weird, i guess.
1183066331 M * daniel_hozac unless it's been solved some other way
1183066345 M * Bertl matter of a few days now, got a little distracted by some customers :)
1183066355 Q * mire Quit: Leaving
1183066360 M * daniel_hozac ah, okay. so i won't forward port it then :)
1183066374 M * Bertl daniel_hozac: the thing is this, I do not want to put the large ipv6 structures in 2.3
1183066389 M * Bertl so I have to replace the ipv4/ipv6 arrays by something else
1183066390 M * daniel_hozac the NB_IPV6ROOT?
1183066401 M * daniel_hozac right.
1183066403 M * Bertl my options here look like this:
1183066408 M * daniel_hozac hasn't that been the plan all along?
1183066450 M * Bertl a) have a configurable NB_IPV4 and NB_IPV6 for now, and just allocate them (only for now), replace that later with a proper hash
1183066478 M * Bertl b) allow only 4 ipv6 addresses _or_ 16 ipv4 addresses (again for now)
1183066506 M * Bertl c) implement a hacky hash which can handle variable sized/typed entries
1183066527 M * daniel_hozac (for now) wouldn't a simple list do the trick?
1183066529 M * Bertl my plan is still to go for c), so it took a little longer
1183066555 M * Bertl yeah, but adding a list just to rip it out a little later?
1183066584 M * daniel_hozac true.
1183066599 M * Bertl the problem with the hash is, we have to walk it for the src ip detection
1183066637 M * daniel_hozac ah yeah, and walking hashes isn't fun...
1183066647 M * Bertl so either we do a hash _and_ list, which would opt for a list as you suggested
1183066667 M * fb http://paste.linux-vserver.org/2877
1183066693 M * Bertl or we traverse the hash in a reasonably fast manner (which probably requires the hash to be resizeable)
1183066726 M * Bertl OTOH, we could generate a perfect hash or at least well structured one when adding/removing IPs
1183066734 M * daniel_hozac how about a small fixed-size hash for now, to become resizeable soon?
1183066758 M * Bertl yeah, something like we already use for xid/nid
1183066840 M * slack101 Bertl, did anyone ever send you those virtuatables scripts ? i cant find them anywehre and i guess they could work for what i need
1183066864 M * daniel_hozac they're on brc's forum.
1183066923 M * Bertl yeah, just seems the download is gone
1183066936 M * fb and for now i'll just use chroot-ed http server
1183066949 M * slack101 yea i could really use those hmmm
1183066961 M * daniel_hozac i should have it somewhere...
1183066962 M * slack101 instead of wasting time doing it in a web UI
1183066989 M * slack101 so basically it just relays everything from guest to host right ?
1183067016 M * slack101 so they will use the same iptables commands etc ?
1183067018 M * fb and finally i'll have access to debian / ubuntu mirror from ipv6 hosts :)
1183067026 Q * gerrit Ping timeout: 480 seconds
1183067078 M * daniel_hozac http://people.linux-vserver.org/~dhozac/t/vserver_virtuatables-0.1.tar.gz
1183067136 M * mugwump Bertl (or anyone who's worked with rlimit etc) does ulimit -m force processes to start paging out early when they hit the limit?
1183067253 M * Bertl hmm, I would expect it to kill the task :)
1183067261 M * mugwump it's supposed to limit rss
1183067278 M * mugwump on my desktop however it seems to be completely ineffective at it
1183067280 M * Bertl try it, shouldn't be too hard, no=
1183067288 M * Bertl s/=/?
1183067299 M * Bertl just allocate some memory and start dirtying the pages
1183067315 M * mugwump ok, I tried ulimit -m 200
1183067319 M * mugwump and then started emacs
1183067331 M * mugwump ps v lists the RSS as over 13MB
1183067352 Q * [BiG^BrotheR] autokilled: Running mirc script bot things in channels without permission scares people. Don't do it. Mail support@oftc.net if you have questions. (2007-06-28
1183067358 M * mugwump (I guess emacs stands for "eightEEN megs and constantly swapping" these days)
1183067484 M * mugwump ok, just tried setting the hard and the soft limit to 8MB and emacs is now using slightly less, but still more than 8MB
1183067507 M * mugwump I wonder if active executable pages don't count
1183067525 M * daniel_hozac and if you set it to 2 MiB?
1183067563 M * fb would you mind if i add to the mirror vserver patches?
1183067568 M * mugwump same thing
1183067577 J * gerrit ~gerrit@204.50.130.82
1183067591 M * daniel_hozac so the limit is not enforced?
1183067606 M * mugwump doesn't look like it. Of course this is a stock ubuntu kernel
1183067614 M * daniel_hozac (TBH i don't see where it would be either. RLIMIT_RSS only appears in proc and headers)
1183067636 M * daniel_hozac fb: they're GPL, do as you see fit...
1183067690 M * mugwump This works with rlimit rss doesn't it?
1183067693 M * fb daniel_hozac: yes, but even with gpl it seems to be, hm, polite? to ask the authors
1183067720 M * daniel_hozac mugwump: hmm?
1183067767 M * mugwump I thought that the vserver-rlimit stuff could limit rss per vserver
1183067772 M * daniel_hozac yes.
1183067786 M * mugwump ah, from getrlimit(2):
1183067812 M * mugwump RLIMIT_RSS: Specifies the limit (in pages) of the process’s resident set... This limit only has effect in Linux 2.4.x, x < 30, and there only affects calls to madvise() specifying MADV_WILLNEED
1183067821 M * daniel_hozac heh, okay.
1183068356 M * mugwump the bash(1) man page is awful on that front
1183068391 M * mugwump it seems to indicate (and I thought) that ulimit -v didn't affect anything other than the shell
1183068398 M * mugwump -v The maximum amount of virtual memory available to the shell
1183068408 M * Bertl btw, as we are deep in nostalgy ... does anybody know a version of ASClock which would work with recent gnome? preferable as panel applet
1183068438 M * mugwump there's a swallow applet
1183068476 M * harry fb: it's an old bug
1183068490 M * harry i fixed in my compile... but i forgot to fix it in the diff-sourcetree
1183068494 M * harry uploading a new one now...
1183068513 M * mugwump heh, no, it's a "Swallower Meta-Applet" ;)
1183068571 M * harry fb: it's fixed in the lastest one ;)
1183068649 M * mugwump so, anyway, ulimit -v is documented as above on bash(1)
1183068669 M * mugwump but an strace reveals it calling: setrlimit(RLIMIT_AS, {rlim_cur=10240*1024, rlim_max=10240*1024}) = 0
1183068671 M * fb harry: i suppose somebody else's responsible for ipv6 patch?
1183068702 M * Bertl fb: yep, that'd be bonbons :)
1183068762 M * fb Bertl: but i suppose rejected parts are because grsec, not the vserver itself ;)
1183068772 M * harry yes
1183068790 M * harry fb: that problem you got a few mins hours ago
1183068800 M * harry was entirely my fault (sorry for that)
1183068810 M * harry i'm rediffing the entire tree to see if there were any other probs...
1183068867 Q * FloodServ synthon.oftc.net services.oftc.net
1183068941 M * fb harry: to be clear, the patch (vserver-2.2 + grsec) against 2.6.21.5 applies cleanly, the paste comes from additional ipv6 patch.
1183068942 M * slack101 daniel_hozac, suggestion ...............copy the installation instruction from the forum and put it in the readme
1183068952 M * slack101 from the looks of it that couldnt be around to much longer
1183068963 M * harry all good now ;)
1183068972 J * FloodServ services@services.oftc.net
1183069007 M * harry fb: thats because the ipv6 is diffed with a non grsec vserver patch prolly
1183069098 M * fb and who should i ask for a grsec version?
1183069107 M * harry me ;)
1183069124 M * fb so here i ask for the one, please :)
1183069134 M * harry but... not today... i'll do it tomorrow if that's ok...
1183069183 M * fb i should be in a bed for a few hours now...
1183069206 M * harry where is the list with rejects?
1183069210 M * harry that i can take a look...
1183069220 M * fb harry: only one question, where to look for it?
1183069228 M * fb and don't horry with that
1183069230 M * harry where to look for what?
1183069248 M * fb for ip6 patch, when it's ready
1183069259 M * harry people.linux-vserver.org/~harry
1183069265 M * fb http://paste.linux-vserver.org/2877
1183069266 M * harry maybe a subdir ipv6 or so
1183069308 M * fb ok, and don't hurry, i can live with chroot-ed lighttpd on the host for now
1183069321 M * harry doesn't seem all that hard
1183069324 M * harry what patch did you use?
1183069333 M * harry iow: where are bonbons patches? :)
1183069338 M * harry people.linux-vserver.org/~bonbons?
1183069367 M * fb the one faq points to
1183069394 M * fb http://people.linux-vserver.org/~bonbons/ipv6/patch-2.6.20.4-vs2.2.0.ipv6-rc19.diff
1183069401 M * harry awtsj
1183069414 M * harry might need some serious tuning, but i'll see what i can do :)
1183069420 M * harry tomorrow :)
1183069429 M * fb and the same files fail when applied against 2.6.19.7
1183069448 J * Min464 ~103730197@IGLD-83-130-46-141.inter.net.il
1183069450 M * Min464 WOW! Free phone calls - http://callfree.point-serv.com/en/
1183069451 P * Min464
1183069464 M * Bertl WOW! stupid folks :)
1183069483 M * harry fb: i'll try my best :)
1183069487 M * Bertl okay, back shortly, have to reboot my frontend machine ...
1183069737 M * fb good night
1183069743 P * fb
1183070773 Q * ruskie Quit: Caught sigterm, terminating...
1183071407 J * Aiken ~james@ppp121-45-220-241.lns2.bne1.internode.on.net
1183071535 M * Bertl ah, udev sucks again, and as usual, it screws up big time :)
1183071637 M * onox what's going on?
1183071685 M * Bertl kernel changed some minor things and made symlinks for all the misc devices, udev doesn't know (in the version I have here)
1183071739 M * Bertl that means, no /dev/{null,zero,mem,kmem ...}
1183071764 M * Bertl you probably can imagine how that helps with booting ;)
1183071811 A * Bertl is so glad that we dropped the unfixable devfs ...
1183072105 M * onox udev sucks, devfs sucks even more?
1183072141 M * Bertl well, except for greg complaining that it is unfixable, I never had any issues with devfs ...
1183072169 M * onox and why the hell replaced null,zero,etc. with symlinks? :S
1183072188 M * onox insert "kernel folks"
1183072193 M * Bertl no idea, I guess it was some kernel cleanup moving them around
1183072215 M * onox they should clean up the featuritis of linus :p
1183072229 M * Bertl I would not be suprised if greg did that himseld (the moving around :)
1183072235 M * Bertl *himself
1183072238 M * onox lol :)
1183072243 M * onox hmm
1183072248 M * onox who's greg?
1183072269 M * Bertl Greg Kroah-Hartman
1183072284 M * onox still have no clue
1183072301 M * Bertl he is the kernel guy who 'invented' udev
1183072320 M * onox ah
1183072343 M * onox btw
1183072368 M * onox Bertl: does the vunify test during installation of util-vserver fails on your computer?
1183072378 M * onox s/fails/fail
1183072385 M * Bertl I hope not :)
1183072390 M * Guy- I also quite liked devfs
1183072486 M * harry all dev implementations suckt!
1183072489 M * harry suck
1183072566 M * harry for starters... they need to make devices unique!
1183072570 M * harry all of them
1183072576 M * harry eth0 should be eth0 forever
1183072590 M * harry if i put another card in there, it should be unique
1183072612 M * Bertl hmm ... maybe that is the time to mention that eth0 is not a 'device' handled by udev or devfs ... :)
1183072617 M * harry if i put 5 cards in there... they should all be unique, no matter what order modules are loaded, busses are scanned
1183072622 M * Bertl anyway, next try to get a newer kernel ...
1183072624 M * harry same goes for disks
1183072627 M * Bertl brb ...
1183072634 M * onox harry!
1183072636 M * harry for any devices
1183072638 M * onox the gentoo guru
1183072643 M * harry lol noooooooo
1183072647 A * harry hates gentoo
1183072648 M * onox vunify test fails
1183072664 M * onox harry: you are gentoo guru right
1183072666 M * onox ? :)
1183072686 P * MonkeyDog satan made me do it
1183072760 M * onox harry: am I right?
1183072889 M * Bertl hmm, went quite well this time ...
1183072903 M * Bertl onox: you are compiling what tool version?
1183072952 M * harry nono
1183072957 A * harry has never used gentoo
1183072959 M * harry never installed it
1183072962 M * harry or anything
1183072973 M * harry and i plan to keep it like that for at least a couple of months
1183072992 M * Bertl onox: you are probably looking for Hollow ...
1183073006 M * harry think so too ;)
1183073033 A * onox is confused
1183073089 M * onox I thought there were 2 gentoo users with (h|H).+
1183073388 M * harry i'm not one of them, onox :)
1183073392 A * harry off to bed now
1183073396 M * harry cya'll later
1183073398 M * slack101 that virtuatables thing uses php to write a daemon =\
1183073930 P * stefani I'm Parting (the water)