1160094000 Q * meandtheshell Quit: exit (0);
1160094044 J * _node node@c-69-143-148-254.hsd1.md.comcast.net
1160094459 Q * dreamind Quit: dreamind
1160096296 Q * duckx Ping timeout: 480 seconds
1160096444 J * duckx ~Duck@tox.dyndns.org
1160096690 Q * virtuoso Remote host closed the connection
1160096702 J * virtuoso ~s0t0na@shisha.spb.ru
1160096709 N * Bertl_oO Bertl
1160096714 M * Bertl evening folks!
1160096744 M * Bertl gex-: that sounds strange .. details?
1160096831 M * derjohn wb Bertl!
1160096844 M * Bertl hey derjohn! how's going?
1160096873 M * derjohn full terror .. leaving for italy for some days and still undone versver guests ;/
1160096936 M * derjohn might you have a hint for me about an nfs mounted in a vserver and "Value too large for defined data type: access to / failed" ?
1160096967 M * Bertl hmm, nfs tagged on any end?
1160096978 M * derjohn my host is 64 bit, the guest 32 bit ...
1160096986 M * derjohn tagged? hmmm
1160097007 M * Bertl what nfs version, what protocol?
1160097021 M * derjohn the filesystems (on lvm) are mounted tagged but the share I am talking about is nfs - untagged
1160097041 M * derjohn no good idea ?
1160097116 M * Bertl well, the question is: has one of the involved kernels nfs tagging (kernel option) enabled and the other not?
1160097228 M * derjohn guest CONFIG_TAG_NFSD is not set, NFS server CONFIG_TAG_NFSD=y
1160097231 M * derjohn bingo
1160097321 M * derjohn Bertl, is there a workaround without recompiling one of the kernels?
1160097362 M * Bertl nope, that change is too intrusive to have it runtime configurable (for now)
1160097380 M * derjohn which way should I go?
1160097383 M * Bertl the only option would be to switch to nfsv4 if compiled in
1160097386 M * derjohn use tagging or not?
1160097406 M * Bertl depends on what you want, the tagging makes the nfs xid aware
1160097424 M * derjohn but what I two guests want to use the same share ?
1160097446 M * derjohn (and they have not same XID)
1160097519 M * derjohn well CONFIG_NFS_V4=y ... i'll try to mount that particular mount V4
1160097630 Q * micah Remote host closed the connection
1160097633 J * micah ~micah@micah.riseup.net
1160097757 J * _suka ~suka@bi01p1.co.us.ibm.com
1160097779 Q * _suka 
1160097864 J * s0undt3ch_ ~s0undt3ch@bl7-244-7.dsl.telepac.pt
1160098056 M * derjohn Bertl, what would happen if I mount one mount via NFS with tagging into two guests? could both use the share at the same time?
1160098114 M * Bertl yep, that's the idea behind it
1160098159 M * derjohn but if A writes a file with with XID 100 and B was XID 150 ...
1160098173 M * derjohn B should be able to even see it?
1160098262 Q * s0undt3ch Ping timeout: 480 seconds
1160098263 N * s0undt3ch_ s0undt3ch
1160098272 M * derjohn wb s0undt3ch !
1160098287 M * Bertl derjohn: on the client it is handled like 'normal' xid tagging
1160098366 M * derjohn Bertl, means: A cant see files that belong to B ? or do i just mistake something ?
1160098481 N * hardwire hardwire-afk
1160098562 M * Bertl derjohn: on the client, they are protected against cross context access
1160098579 M * Bertl but you cannot use the 'same' file in different hosts
1160098586 M * Bertl s/hosts/guests/
1160098807 M * derjohn so the file's xid is a numeric value as a property of the file. this one is transferred via net. hmmmm ... till today I used one share mounted on two differnt hosts.
1160098855 M * derjohn on those hosts i did a bindmount into some location of a guest. both guests were able to access the shared storage.
1160098871 M * derjohn read: the same files.
1160098930 M * derjohn do i have to use tagxid as mount option in the nfs part ???
1160099018 M * derjohn interesting. I can access the files via shell, but apache has problems to deliver them.
1160099025 M * Bertl yep
1160099047 M * derjohn what is differntly in the apache part?
1160099050 M * Bertl (that was regarding the mount option)
1160099055 M * derjohn k
1160099063 M * Bertl not 100% here, I'm a little ill
1160099078 M * derjohn oh , sad !
1160099087 M * derjohn have cold ?
1160099090 M * Bertl flu, nothing tragical
1160099104 M * derjohn ruesselpest ?
1160099129 M * Bertl apache probably uses sendfile
1160099142 M * Bertl try if https is fine but http fails
1160099154 M * Bertl sendfile over nfs is not implemented yet (iirc)
1160099160 M * derjohn not ssl enabled there ...
1160099190 M * Bertl well, alternatively you can disable the sendfile manually
1160099205 M * derjohn what if dont set "tagxid" while mounting ?
1160099218 M * Bertl it shouldn't hurt
1160099231 M * Bertl just no tagging info will be transferred
1160099252 M * derjohn remember: i still have one guest without nfs tagging 
1160099294 M * derjohn hm, of, but why the Value too large for defined data type: access to / failed ?
1160099311 M * derjohn I mean I didnt mount with tagxid
1160099395 M * Bertl no, the tagging option (kernel) changes the rpc/nfs structures
1160099439 M * derjohn so: If I recompile with the option, but do not (!) mount with tagxid, the sendfile stuff will be happy ?
1160099651 M * Bertl I don't remember if sendfile is supported (on NFS) at all
1160099678 M * Bertl but yes, if both sides have tagging enabled, but do not use tagxid, it will work fine too
1160100167 M * ntrs just a short report
1160100177 M * ntrs for about 50 hours now the guests that I have limited to a single cpu with cpusets have not had a single mysql hang yet.
1160100180 M * derjohn hi ntrs ! I hope good news ?
1160100206 M * ntrs what does that tell us? mysql threading is broken on multi cpu systems?
1160100242 M * Bertl well, or something in linux-vserver causes broken behaviour
1160100267 M * Bertl what would clarify that would be putting one guest on a single dual cpu machine
1160100288 M * Bertl without linux-vserver on it, that is, just a chroot
1160100586 M * ntrs I will wait another week or so and will keep an eye on the guests with the limitation. Will report again soon.
1160100606 M * Bertl okay, great, thanks!
1160100884 Q * ensc Killed (NickServ (GHOST command used by ensc_))
1160100894 J * ensc ~irc-ensc@p54B4D325.dip.t-dialin.net
1160100950 M * gex- Bertl, whops, didn't notice your reply, sorry :)
1160100967 M * Bertl np
1160100990 M * gex- i tried 'vserver --debug v01 start'
1160100997 M * gex- it SIGSEGV's while running the following:
1160101016 M * gex- ++ /bin/nice -n 0 /usr/sbin/chbind --silent --nid 1800 --ip 1.2.3.4/24 -- /usr/lib/util-vserver/exec-ulimit /etc/vservers/v08/ulimits /usr/sbin/vcontext --create --silent --xid 1800 -- /usr/sbin/vnamespace --set -- /usr/sbin/vlimit --dir /etc/vservers/v08/rlimits --missingok -- /usr/sbin/vsched --xid self --fill-rate 0 --interval 32 --tokens 500 --tokens_min 200 --tokens_max 1000 --priority-bias dummy -- /usr/sbin/vuname --xid self --dir /etc/vservers/v08/uts -
1160101029 M * gex- 1.2.3.4 being the real ip address... whops, and vserver --debug v08 start is what i did
1160101076 M * Bertl that line looks incomplete
1160101097 M * Bertl maybe use paste.linux-vserver.org
1160101135 M * gex- hm probably too long for here - http://paste.linux-vserver.org/447
1160101180 M * Bertl so the runlevel script segfaults
1160101183 M * Bertl try the 
1160101188 M * Bertl following
1160101213 M * Bertl chroot into the guest's root dir and execute /bin/bash
1160101244 M * gex- ok, i'm there :)
1160101256 M * gex- then try to run the init script manually i suppose?
1160101274 M * Bertl no
1160101295 M * Bertl what shell does the rc script use?
1160101309 M * gex- /bin/bash
1160101322 M * Bertl so executing that one works fine, yes?
1160101326 M * gex- yes
1160101352 M * Bertl okay, what tool version are you using? i.e. could you upload the output of vserver-info - SYSINFO too?
1160101379 M * gex- 0.29.5 is what i'm using
1160101418 M * gex- err... 0.30.210, sorry, checked on the wrong machine
1160101440 M * Bertl okay, is it a problem to try 211-rc2?
1160101447 M * gex- output is on http://paste.linux-vserver.org/448
1160101506 M * gex- hm, using the dhozac rpm repo here, guess i could do a manual build tho
1160101537 M * Bertl not necessary
1160101554 M * Bertl an alternative is to add a few lines to the rc script (for a test)
1160101589 M * gex- ah ok... what do i need to add?
1160101701 M * derjohn ntrs, dou you have a mysql 4.1 for debian etch somewhere?
1160101709 M * gex- wait, what is this...
1160101724 M * gex- i just tried to chroot to the vserver directory again, and now i get a SIGSEGV there
1160101761 M * ntrs derjohn, no, sorry.
1160101772 M * derjohn k
1160101784 M * Bertl gex-: hmm, strange
1160101798 M * Bertl gex-: what changed in the meanwhile?
1160101815 M * Bertl (btw, the additions would be here: http://paste.linux-vserver.org/449)
1160101830 M * Bertl but that's not really relevant when you get it with a simple chroot
1160101847 M * gex- nothing really, i just exited from the chroot, did the vserver-info command you gave me above, then tried to chroot again
1160101857 M * gex- crash appears, according to strace, after "open("/etc/ld.so.preload", O_RDONLY)    = -1 ENOENT (No such file or directory)"
1160101884 M * gex- executes /bin/bash, tries to open /etc/ld.so.preload and SIGSEGVs... so probably a bash problem after all
1160101928 M * Bertl looks like, could be nptl related
1160101964 M * gex- hmm... any clues as to what i could do to fix?
1160102024 M * Bertl update glibc or show an old 2.4 kernel version
1160102070 M * gex- hm, glibc is the most current one fedora core 1 has to offer... show an old 2.4 kernel version? how would i go about that?
1160102101 M * Bertl you can easily fake that with the uts functions of linux-vserver
1160102138 M * Bertl but this should work for a test too:
1160102142 M * Bertl export LD_ASSUME_KERNEL=2.4.19#
1160102150 M * Bertl without the #
1160102177 M * gex- hmm... now i get: chroot: error while loading shared libraries: libc.so.6: cannot open shared object file: No such file or directory
1160102247 M * gex- that's coming from the 'main' server tho, after setting LD_ASSUME_KERNEL=2.4.19 even 'ls' doesn't work any more
1160102267 M * Bertl not outside, that's for inside :)
1160102295 M * gex- hm, how do i run that from the inside when i can't start the vserver tho? ;)
1160102332 M * Bertl well, for the guest start, put that version into the uts config
1160102341 M * Bertl it will then show up immediately
1160102367 M * Bertl that was more for chroot() when it 'worked'
1160102375 M * Bertl to make the rest working too
1160102456 M * gex- gosh, am i stupid today? :) what's the URL of the GFP again? i'm too blind to find it in the linux-vserver site apparently heh
1160102484 M * Bertl http://www.nongnu.org/util-vserver/doc/conf/configuration.html
1160102521 M * Bertl uts/release is what you want
1160102538 M * gex- ah thanks... so i just put the export LD_ASSUME_KERNEL=2.4.19 in there?
1160102550 M * Bertl nope, just 2.4.19
1160102556 M * gex- oh ok
1160102568 M * gex- hm, no... still same sigsegv :(
1160102571 Q * samueltc Ping timeout: 480 seconds
1160102578 Q * _node 
1160102587 J * _node node@c-69-143-148-254.hsd1.md.comcast.net
1160102622 M * Bertl gex-: what is the host distro?
1160102660 M * gex- fedora core 5
1160102704 M * Bertl well, as the chroot() causes the same segfault (sometimes)
1160102740 M * Bertl the issue is not vserver related, but, it is very likely glibc related (i.e. the guest glibc is too confused)
1160102772 M * Bertl the only thing I can suggest to try is to switch to 'plain' init style
1160102794 M * Bertl so that a complete guest transition happens with a static init
1160102814 M * Bertl keep the uts entry though, as it will probably be required
1160102862 M * gex- shouldn't i be seeing the '2.4.19' somewhere in the debug output? (vserver --debug v08 start)
1160102984 M * Bertl not directly, it's only listed as dir
1160103047 M * gex- ah ok, was just wondering :) hm... could you explain how to switch to plain init style maybe? always had it working this way, so no experience with that
1160103071 Q * _node Ping timeout: 480 seconds
1160103123 M * Bertl echo plain >/etc/vservers/<vserver-name>/apps/init/style
1160103137 M * gex- oh ok, that's simple ;)
1160103159 M * Bertl yeah, it will not give any output (from the startup) anymore
1160103173 M * Bertl and it will start the init inside the guest
1160103182 M * Bertl (well, at least try to :)
1160103212 M * gex- hm, got no output indeed, but according to vserver v08 status, it's still stopped... vps aux shows no processes in the context, too
1160103235 M * gex- when i try to run the start command again, now i get 'vshelper.init: can not determine xid of vserver 'v08'; returned value was '''
1160103390 M * Bertl try with --debug again
1160103409 M * Bertl btw, are you 100% sure that the guest works?
1160103440 M * Bertl I mean it looks a little like damaged libraries to me
1160103467 M * gex- it's running on the old machine fine, all i did was stop it there, tar zcfv the directory and tar zxfv it on the new guest, so pretty sure...
1160103483 M * Bertl tar?
1160103495 M * Bertl not such a good idea
1160103500 M * gex- uh? why?
1160103521 M * Bertl because without the proper options, you will get messed up uid/gids
1160103542 M * Bertl do you have rsync at hand?
1160103588 M * gex- not _yet_ :)
1160103624 M * Bertl okay, when you have :) try rsync -azx --numeric-ids
1160103838 M * Bertl rsync also knows --dry-run (for testing)
1160103856 M * Bertl together with --verbose or --progress you'll get an idea what it will do
1160103895 M * gex- syncing now
1160103964 Q * AndrewLee Ping timeout: 480 seconds
1160103991 M * Bertl what kernel version is the host kernel there?
1160104033 M * gex- 2.6.17 on the new box
1160104038 M * Bertl nah, the old one
1160104065 M * Bertl you might want to mimic the kernel version of that one 
1160104112 M * gex- 2.4.26 there
1160104143 M * gex- arg...
1160104168 M * Bertl c or v?
1160104173 M * gex- place where the two machines are is just losing connectivity
1160104179 M * gex- :D
1160104201 M * Bertl space? exclipse with the sun?
1160104211 M * gex- SSH@Router 1 #sh cpu
1160104212 M * gex- 99 percent busy, from 8 sec ago
1160104217 M * gex- more like... load :p
1160104484 M * gex- seems to recover tho, and rsync almost done :)
1160104596 M * gex- and done, didn't change anything tho, so don't seem to be bad libs
1160104629 M * gex- also tried with 2.4.26 as kernel version
1160104674 M * Bertl what about the --debug run?
1160104700 M * gex- +++ /usr/sbin/vserver-info /etc/vservers/v08 CONTEXT false
1160104700 M * gex- ++ xid=
1160104700 M * gex- ++ warning 'vshelper.init: can not determine xid of vserver '\''v08'\''; returned value was '\'''\'''
1160104717 M * Bertl what xid has that guest?
1160104720 M * gex- 1800
1160104735 M * Bertl what does vserver-stat say?
1160104747 M * Bertl (regarding that xid)
1160104757 M * gex- doesn't appear 
1160104787 M * Bertl very strange
1160104814 M * gex- when i try to run '/usr/sbin/vserver-info /etc/vservers/v08 CONTEXT' i get 1800 back fine
1160104845 J * AndrewLee ~andrew@tnlug.linux.org.tw
1160104920 M * gex- i see it's trying to execute /sbin/init though, and no SIGSEGV there
1160104961 M * Bertl please give the .211-rc2 a try
1160105004 M * Bertl IIRC, there was some issue with /bin/env being shadowed by /usr/bin/env
1160105043 M * Bertl (or the other way round)
1160105096 M * Bertl http://www.mail-archive.com/vserver@list.linux-vserver.org/msg09276.html
1160105156 M * gex- hm, either i'm blind once again or it's not in the regular download area of util-vserver?
1160105605 M * Bertl no idea where it was moved now :(
1160105678 M * Bertl ah, here it is:
1160105679 M * Bertl http://people.linux-vserver.org/~dhozac/p/uv/experimental/util-vserver-0.30.211-rc2.tar.bz2
1160105896 M * gex- let me build...
1160106289 M * gex- updated, same problem though
1160106336 J * essobi_ kstone@216.26.137.75
1160106346 M * essobi_ Howdy peebles.
1160106377 M * essobi_ I'm trying to use vserver-copy and it's whining about /etc/vserver/vs.conf not existing...
1160106385 M * essobi_ what's this file for and how can I populate it?
1160106424 M * Bertl vserver-copy is legacy stuff
1160106435 M * Bertl it's neither maintained nor supported anymore, IIRC
1160106471 M * essobi_ o
1160106487 M * essobi_ What's the proper way to copy a vserver then?
1160106500 M * Bertl gex-: okay, try the _ENV change then please
1160106526 M * Bertl essobi_: simple, create a skeleton config, then copy/rsync/dump/extract the template
1160106562 M * gex- Bertl: it's already set to /bin/env (which is not the symlink)
1160106564 M * essobi_ That doesn't sound too simple.
1160106589 M * Bertl gex-: okay, then I have absolutely no idea what your guest dislikes
1160106618 M * Bertl gex-: if you want, you can pack a minimized version together (into a dump/tar) and I have a look at it tomorrow
1160106635 M * gex- hm, what is this... just did dmesg
1160106637 M * gex- vxW: xid=1800 did hit the barrier.
1160106638 M * gex- vxW: xid=1 did hit the barrier.
1160106642 M * gex- what does that mean?
1160106656 M * Bertl that means that you have barrier flags set inside the guest?
1160106689 M * Bertl maybe broken tools on the original host?
1160106719 M * gex- i wouldn't think so actually, the message seems to be old though, from yesterday morning
1160106790 M * Bertl essobi_: well, you can file a feature request to the tools, I would really appreciate a copy/clone method for the new config
1160106802 M * Bertl (but it wasn't done yet)
1160106826 M * essobi_ Roger that.
1160106847 M * Bertl gex-: so if you cut that guest down to roughly 40megs, with the issues still appearing, I can take a closer look at it ...
1160106872 M * Bertl (probably bash + libraries should be enough for that to happen)
1160106912 M * gex- so tar /bin/bash and /lib?
1160106941 M * Bertl try to create a reduced copy of that guest and _make sure_ that this reduced versions exposes the issues ... I'm off to bed now (6am here) but I'll look into it when I get up ...
1160106969 M * Bertl (just leave an url here or send it via email)
1160106982 M * gex- 6am here too (germany), and i got a train to catch in 5 hours, yikes :) i'll probably do this when i get back from my weekend trip
1160106996 M * Bertl okay, no problem!
1160107011 M * gex- thanks a bunch for your help though, really appreciate it!
1160107022 M * Bertl you're welcome! have a good one everyone! cya tomorrow!
1160107031 N * Bertl Bertl_zZ
1160107031 M * gex- g'night :)
1160109819 M * daniel_hozac essobi_: FYI, vserver <name> build -m skeleton ... -- rsync -Hav /vservers/<source>/ /vservers/<name>/ works fine.
1160109925 M * daniel_hozac gex-: you probably should've gotten 0.30.211-rc3 instead ;)
1160109966 M * daniel_hozac (doesn't really change anything with regard to your problem, but it fixes some other bugs)
1160111039 M * essobi_ -m skeleton being the vserver you want to copy?
1160111057 M * essobi_ Oh oh
1160111058 M * essobi_ I see..
1160113906 Q * Piet Quit: :tiuQ
1160117291 Q * Aiken Ping timeout: 480 seconds
1160117316 Q * Loki|muh Ping timeout: 480 seconds
1160117612 J * Loki|muh loki@satanix.de
1160120266 J * ruskie ~ruskie@ruskie.user.oftc.net
1160120460 J * Aiken ~james@tooax6-005.dialup.optusnet.com.au
1160120741 J * meandtheshell ~markus@85-124-175-102.dynamic.xdsl-line.inode.at
1160120960 J * eGnarF ~bartek@bk.crystone.se
1160120970 M * eGnarF Hi!
1160121014 M * eGnarF I'm trying to limit the maximum allowed memory that a guest server may use, but it doesn't seem to work..
1160121039 M * eGnarF I tried vlimit -c GuestServer --rss 256M
1160121058 M * eGnarF But the programs in the guest can still use all memory in the server.
1160121094 J * prae ~Benjamin@5-63.206-83.static-ip.oleane.fr
1160121163 M * Borg- GuestServer? you meen you specifiy a xid of guest server?
1160121205 M * eGnarF Borg-: I specify it's name. the same I use when I do a vserver GuestServer start
1160121214 M * Borg- its wrong
1160121217 M * Borg- use xid
1160121231 M * Borg- vlimit --help says that exacly
1160121238 J * trippeh atomt@x.vx.no
1160121249 M * eGnarF Ok.. Lemme test.
1160121272 J * dna_ ~naucki@p54BCE72A.dip.t-dialin.net
1160121401 M * eGnarF Borg-: I did vlimit -c 1051 --rss 256M
1160121411 M * eGnarF Borg-: But this is what I get: 1051     7   1.7G   1.6G  10m59s51   2m11s92   1d22h32 test64
1160121433 M * eGnarF Borg-: And my test program can allocate all 4GB of ram...
1160121481 M * Borg- do vlimit -c 1051 -a
1160121487 M * Borg- what it says in RSS ?
1160121502 M * eGnarF Borg-: RSS         N/A                 N/A                 0x10000000
1160121505 M * Borg- hmm
1160121512 M * Borg- no limits then.. dunno why. 
1160121526 M * eGnarF What's the 0x10xx?
1160121529 M * Borg- wait for someone who used limits successfuly.. I didnt tried them yet
1160122119 M * matti Morning.
1160122429 J * dna___ ~naucki@p54BCE72A.dip.t-dialin.net
1160122856 Q * dna_ Ping timeout: 480 seconds
1160123083 J * dna_ ~naucki@p54BCE72A.dip.t-dialin.net
1160123409 M * eyck      !ss
1160123496 Q * dna___ Ping timeout: 480 seconds
1160123507 J * dna___ ~naucki@p54BCE138.dip.t-dialin.net
1160123670 Q * shedi Quit: Leaving
1160123713 M * trippeh Hm, I've lost the test patches (eg for 2.6.18) url
1160123741 Q * dna_ Ping timeout: 480 seconds
1160124001 J * dna_ ~naucki@p54BCE138.dip.t-dialin.net
1160124166 Q * dna___ Ping timeout: 480 seconds
1160124966 J * glut ~glut@no.suid.pl
1160125360 Q * derjohn2 Quit: Verlassend
1160125851 Q * DreamerC Quit: leaving
1160125873 J * lilalinux ~plasma@dslb-084-058-217-009.pools.arcor-ip.net
1160127665 M * doener trippeh: http://vserver.13thfloor.at/Experimental -- that one?
1160127700 M * trippeh doener: Yup =)
1160127702 M * trippeh Thanks
1160127913 J * DreamerC ~dreamerc@59-115-50-96.dynamic.hinet.net
1160128562 Q * mcp Read error: Connection reset by peer
1160128603 J * mcp ~hightower@wolk-project.de
1160129346 J * FireEgl FireEgl@Sebastian.Atlantica.US
1160130514 P * independence 
1160131147 J * bubulak_ ~bubulak@whisky.pendo.sk
1160131346 Q * bubulak Ping timeout: 480 seconds
1160131797 Q * DreamerC Quit: leaving
1160132399 J * Ralph ~Avrupacaf@81.215.182.13
1160132420 Q * Ralph 
1160135298 Q * lilalinux Remote host closed the connection
1160135483 J * lilalinux ~plasma@dslb-084-058-217-009.pools.arcor-ip.net
1160135613 J * DreamerC ~dreamerc@59-115-50-96.dynamic.hinet.net
1160136000 J * Ben81 ~Ben81@tipi0e.lri.fr
1160136456 J * shedi ~siggi@dsl-149-109-85.hive.is
1160137414 J * oli_ ice@p54A3B0AD.dip0.t-ipconnect.de
1160137416 M * oli_ hi all
1160137505 M * oli_ someone here who could help me by a gentoo guest?
1160137642 J * mugwump_ ~samv@watts.utsl.gen.nz
1160137642 Q * mugwump Read error: Connection reset by peer
1160137692 J * Thorsten ~Thorsten@dslb-084-058-144-091.pools.arcor-ip.net
1160137722 Q * ay Quit: leaving
1160138625 J * Radiance 08f6a0b00d@halt.1984world.eu
1160138901 Q * Aiken Ping timeout: 480 seconds
1160139102 M * lilalinux oli_: not until they know, what you need ;)
1160139200 M * oli_ ok
1160139227 M * oli_ i have a problem, beacaue i have2 gentoo guests and they run agetty, this causes a very high load on the vserver
1160139241 M * oli_ i'll paste a message output
1160139286 M * oli_ http://paste.linux-vserver.org/450
1160139307 M * oli_ this comes each second two times
1160139542 M * doener remove agetty ;)
1160139587 M * doener if some other ebuilds depend on it, at least remove the corresponding entries from inittab
1160139768 J * ComplexMind ~mark@87.124.129.7
1160139809 Q * ComplexMind Remote host closed the connection
1160139879 J * ComplexMind ~mark@87.124.129.7
1160140213 Q * ComplexMind Remote host closed the connection
1160140231 J * ComplexMind ~mark@87.124.129.7
1160140361 M * ComplexMind can somebody tell me the best way to mount/umount a dedicated / filesystem for a vserver at start/stop time?
1160140402 M * ComplexMind eg say I have an lv dedicated to a single vserver... how would I hook that into the configuration for that vserver so it mounted automatically
1160140499 M * Thorsten /etc/vserver/foo/fstab ?
1160140522 M * ComplexMind so you can specify a / filesystem there? 
1160140537 M * Thorsten Mmm, no?
1160140550 M * ComplexMind I haven't tried it :)
1160140566 M * ComplexMind should that work for a / filesystem dedicated to a vserver?
1160140585 M * Thorsten You can mount it via autofs
1160140629 M * Thorsten So it will be mounted whenever someone tries to access it
1160140635 M * ComplexMind so I would need to write a couple of custom start/stop scripts?
1160140650 M * Thorsten Mmm, why?
1160140688 M * ComplexMind well, the idea here is to be able to mount a dedicated vserver / partition from shared storage, so that vservers can be moved around easily
1160140741 M * ComplexMind so when a vserver is started I need to mount that partition automatically, and umount when stopping
1160140828 M * Thorsten In theory, everything you have to do is tell autofs that as soon as someone tries to access something in /var/autofs/vserver1 mount /dev/whatever and to unmount it if it hasn't be used for x seconds.
1160140846 M * Thorsten So vserver foo start and vserver foo stop (wait x seconds) should be enough
1160140943 M * Thorsten You can however also write a 2 line shell script
1160141136 M * ComplexMind perhaps I should be using /etc/vservers/<vsname>/apps/init/mtab?
1160141161 M * ComplexMind or is that for something different...?
1160141227 M * Thorsten sorry, don't know
1160141295 M * ComplexMind ok, well thanks - Im sure it can be done quite easily with some little scripts hooked into the start/stop, it's just be nice if I could specifiy the mount in the config somewhere and have it mount "on demand"
1160142256 N * Bertl_zZ Bertl
1160142262 M * Bertl morning folks!
1160142269 M * phedny hi Bertl :)
1160142270 M * Thorsten hi
1160142277 J * hkhl ~hkhl@85.183.76.2
1160142357 M * Bertl eGnarF: care to elaborate on your issues once again?
1160142367 M * Bertl daniel_hozac: ping?
1160142371 M * Thorsten Is there already a vserver patch for 2.6.18 available?
1160142376 M * Bertl yep
1160142437 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.18-vs2.0.2.1-t8.diff
1160142444 M * Thorsten thx Bertl
1160142463 M * Bertl or devel
1160142472 M * Bertl http://vserver.13thfloor.at/Experimental/patch-2.6.18-vs2.1.1-rc35-t1.diff
1160142503 M * eGnarF Bertl: Sure..
1160142548 M * eGnarF Bertl: I've tried to set RSS limit on a guest, using the following command: vlimit -c 1051 --rss 256M
1160142573 M * eGnarF Bertl: vlimit -a -c 1051 says: RSS         N/A                 N/A                 0x10000000
1160142600 M * eGnarF Bertl: And I can still allocate all 4GB of host server memory using a simple test program.
1160142614 M * Bertl allocate means?
1160142652 M * eGnarF Bertl: Setting the rss size to something really small (--rss 1000) seems to kida work. When I set that value all programs that try to use memory get killed.
1160142677 M * eGnarF Bertl: Allocate means doing a malloc() and then filling that block with random crap.
1160142683 M * Bertl do you have an idea how linux memory management works?
1160142693 M * eGnarF Bertl: not much.
1160142705 M * Bertl how much main memory do you have on the host?
1160142709 M * eGnarF Bertl: 4GB
1160142802 M * Bertl okay, so the following should happen with 256M RSS hard limit
1160142826 M * Bertl - allocation (malloc or calloc) of 4GB will succeed 
1160142847 M * Bertl - filling up that memory with random data will work until you reach the 256M
1160142866 M * Bertl - at this point, you will get into an OOM state where two things can happen
1160142884 M * Bertl - OOM killer goes and kills your process
1160142896 M * Loki|muh murder!
1160142898 M * Bertl - memory is swapped out (which doesn't account as RSS)
1160142906 M * eGnarF That I understand. But the OOM killer doesn't start murdering.
1160142908 M * Bertl RSS = Resident Set Size
1160142938 M * Bertl eGnarF: what kernel version do you have?
1160142984 M * Bertl eGnarF: and could you provide your test tool's source code?
1160143129 J * Nysis ~Public@dslb-088-073-047-163.pools.arcor-ip.net
1160143159 M * eGnarF Bertl: 2.6.17.7
1160143231 M * eGnarF Bertl: does --rss limit the total amount of memory for the entire guest or just for single processes?
1160143255 M * eGnarF Bertl: here's the test-program I've used: http://pastebin.ca/193006
1160143270 M * eGnarF Bertl: I also tested it on an actual guest that's using way too much ram.
1160143316 M * Bertl what does the /proc/virtual/<xid>/limit show for such an over limit guest?
1160143335 M * Bertl and what linux-vserver version do you use on that kernel?
1160143362 M * eGnarF Bertl: all of it, or just the RSS-line?
1160143378 M * Bertl all of it, but please use paste.linux-vserver.org
1160143422 M * eGnarF Bertl: vserver-version is 2.0.2
1160143494 M * eGnarF Bertl: http://paste.linux-vserver.org/451
1160143529 M * eGnarF Bertl: util-vserver versions i tested with are 0.30.208 and 0.30.210
1160143537 M * meandtheshell eGnarF: http://list.linux-vserver.org/archive/vserver/msg10545.html might help understanding ... 
1160143595 M * Bertl eGnarF: I think the problem is that your limits are much too high
1160143614 M * Bertl basically your RSS limit is at 268435456 PAGES
1160143619 M * eGnarF Bertl: Ah.
1160143631 M * Bertl that means 1099511627776 on x86
1160143644 M * Bertl or 1TB
1160143668 M * meandtheshell a damn Supercomputer Cluster :)
1160143668 M * eGnarF 256MB shoud be 65536 with 4k pages.
1160143676 M * eGnarF I'll try that.
1160143709 M * Bertl also I'd suggest to update to 2.6.17.13 + vs2.0.2.1
1160143715 M * Bertl (or later)
1160143759 J * ierk ~bg@priv.blacknet.de
1160143771 M * Bertl welcome ierk!
1160143774 M * ierk hi
1160143797 M * ierk I've a probably dump question, but my google-foo is failing me.
1160143806 M * Bertl np, let's hear
1160143816 M * eGnarF hmm.
1160143817 M * MooingLemur foo-fu
1160143828 M * ierk I'm running a vserver with the tools/kernel from debian-backports and would like to get halt/reboot to work in the vserver
1160143840 M * ierk but i only get /dev/initctl: No such file or directory
1160143855 M * Bertl ierk: that's quite simple, your guest is sysv init style based
1160143860 M * ierk creating the named pipe does not help, as nothing is opening it
1160143865 M * Bertl so no real init is started inside
1160143868 M * ierk it's debian, too. so yes
1160143885 M * Bertl now the reboot/halt are nothing but tools to signal init (by default)
1160143895 M * Bertl (which will fail here, naturally)
1160143910 M * Bertl you have two options here (plus a few variations)
1160143929 M * Bertl - use reboot -f / halt -f to bypass init and signal the kernel directly
1160143949 M * Bertl - use the 'plain' init style and start a guest init
1160143979 M * Bertl of course you can write a 'reboot' wrapper or alias which calls reboot -f
1160143980 M * eGnarF Bertl: When I set rss to 65536, this is what I got on my test-guest qith 400mb allocated: RSS:           546            5800           65536           9
1160143988 M * ierk bertl: by using plain, the normal tools would work?
1160143992 M * eGnarF Bertl: so I'm guessing much of my memory went to swap?
1160144009 M * eGnarF Bertl: Changing the AS to 65535 seems to give me the results I want  tho.
1160144025 M * Bertl yes, the AS is the address space allocated
1160144036 M * Bertl the RSS is just the pages in RAM
1160144072 M * Bertl ierk: yes, but you will lose the startup/shutdown output, and use slighly more resource (i.e. the additional init process)
1160144295 M * ierk Bertl: and now it acts fishy. if i reboot my vserver, it kicks me out of my ssh session, instead of back to the host (I've used versver test enter)
1160144313 M * eGnarF Bertl: one more question. On one guest vserver-stat is showing 0.9GB while /proc../limits shows only 30764 pages...  See http://paste.linux-vserver.org/452
1160144359 M * eGnarF Bertl: One thing tho, the host is running amd64. But I think the page-sizes are the same on that and x86?
1160144391 M * Bertl ierk: that's normal
1160144422 M * Bertl ierk: you ssh client basically enters the guest, so on guest reboot, that process is killed
1160144450 M * ierk Bertl: ok, than I'm fine with it. thanks a lot.
1160144496 M * Bertl eGnarF: depends, 'normal' pages are 4k and huge pages are 16k IIRC
1160144517 M * Bertl ierk: typically you would ssh directly to the guest
1160144557 M * eGnarF Bertl: so, I can't set a 256MB memory limit and be sure it stays 256mb since the guest might use huge pages for some stuff?
1160144563 M * ierk Bertl: shure, but I'm still testing. so vserver test enter seamed simpler
1160144597 J * Piet hiddenserv@tor.noreply.org
1160144608 M * Bertl eGnarF: no, the page size is typically constant (in the accounting) but it depends on the kernel config what the actual page size is
1160144633 M * Bertl eGnarF: there is a tool somewhere to figure the page size
1160144656 M * eGnarF Bertl: Does it say what the size is on the kernels .config?
1160144696 M * Bertl yes, but it's easier to check on the running kernel
1160144773 M * eGnarF Bertl: You don't happen to remember what the tool is called?
1160144838 M * Bertl http://oldwiki.linux-vserver.org/Resource+Limitshttp://oldwiki.linux-vserver.org/Resource+Limits
1160145092 M * eGnarF Bertl: The tool says 4k pages, but the numbers returned from vserver-stat and /proc/..limits/ don't match up..
1160145148 M * Bertl that could be for several reasons
1160145166 M * Bertl a) the vserver-stat is a very old tool and it basically sums up ps auxwww
1160145192 M * Bertl b) for some time (maybe not fixed in your tools) the vserver-stat had an off by 1024 bug :)
1160145221 M * Bertl c) the sum of all processes is not necessarily equal to the accounted amount
1160145257 M * Bertl bottomline, vserver-stat would benefit from an overhaul, and the kernel accounted values can be considered _the_ reference
1160145431 M * ComplexMind I'm trying to build a vserver inside a dedicated filesystem.  Is there a way to get around the toplevel directory exists error without hacking the util-vserver build functions?
1160145484 M * ComplexMind I'd rather not have to build the vserver and then copy onto a sperate filesystem...
1160145634 M * eGnarF Bertl: hey, thanks. I think I'm getting it to do what I want now =)
1160145648 M * ntrs is there any reason security wise to use 2.6.18 instead of staying with 2.6.17.13? Is there any major benefit of upgrading?
1160145671 M * Bertl security wise no
1160145799 M * ComplexMind please ignore my last question :P
1160145829 M * ComplexMind there were a couple of files in the filesystem...
1160145986 Q * brc_ Ping timeout: 480 seconds
1160146047 J * Piet_ hiddenserv@tor.noreply.org
1160146300 M * eGnarF Bertl: what does the ANON line in /proc/.../limits show?
1160146304 Q * Nysis Read error: Connection reset by peer
1160146321 Q * Piet Remote host closed the connection
1160146404 M * eGnarF Bertl: because i malloced and filled 1GB of ram, and the RSS-line didn't move a bit, but the ANON line was quite a bit larger than before.
1160146636 M * Bertl anon is anonymous memory
1160146653 M * Bertl i.e. memory not backed by a file or shared between processes
1160146674 M * Bertl I would say that memory directly went into swap then
1160146687 M * eGnarF Bertl: Ok.. I need a better memory tester then =)
1160146714 M * eGnarF Bertl: It seems odd it went to swap tho. Since I have about 2GB free ram at the moment.
1160146736 M * Bertl depends on the configured swappiness
1160146744 M * eGnarF and when I run the program, 'free' register 1GB more ram usage on the host.
1160146788 M * Bertl but rss doesn't change a bit?
1160146798 M * eGnarF Bertl: well, not much. Let me pastebin it for you.
1160146827 M * Bertl also check/upload the ps auxww entry for that pid
1160146925 M * eGnarF Bertl: http://pastebin.ca/193058
1160147062 M * eGnarF I need to run.. I'll be back in 30 minutes or so.
1160147072 M * Bertl hmm, could you add some labeling to the ps output
1160147079 M * Bertl it's quite different to mine
1160147125 M * Bertl np
1160147143 M * matti Hi Bertl, how are you? Feeling a bit better now?
1160147179 M * Bertl well, let's say I'm feeling something ...
1160147563 Q * meandtheshell Remote host closed the connection
1160147633 J * meandtheshell ~markus@85-124-175-102.dynamic.xdsl-line.inode.at
1160147660 N * Piet_ Piet
1160147795 J * lilalinux_ ~plasma@80.69.35.186
1160148236 Q * lilalinux Ping timeout: 480 seconds
1160148444 J * DreamerC_ ~dreamerc@61-217-246-141.dynamic.hinet.net
1160148462 Q * bubulak_ Remote host closed the connection
1160148503 Q * coocoon Ping timeout: 480 seconds
1160148804 J * bubulak ~bubulak@whisky.pendo.sk
1160148851 Q * DreamerC Ping timeout: 480 seconds
1160149150 J * coocoon ~coocoon@dslb-084-056-198-200.pools.arcor-ip.net
1160149245 J * Rich_Estill ~restill@c-24-11-195-139.hsd1.mi.comcast.net
1160149498 Q * bubulak Remote host closed the connection
1160149541 Q * Ben81 Quit: Leaving
1160149878 J * bubulak ~bubulak@whisky.pendo.sk
1160149995 J * stefani ~stefani@tsipoor.banerian.org
1160150013 M * Bertl okay, guess I'm off to bed again ...
1160150019 N * Bertl Bertl_zZ
1160150104 M * eGnarF Bertl_zZ: here's the labeled output http://pastebin.ca/193109
1160150202 P * Thorsten Leaving
1160150401 Q * bubulak Quit: Lost terminal
1160150553 J * bubulak ~bubulak@whisky.pendo.sk
1160151676 Q * prae Quit: Quitte
1160152071 Q * coocoon Quit: KVIrc 3.2.0 'Realia'
1160152335 Q * ComplexMind Remote host closed the connection
1160153224 Q * matled Read error: Connection reset by peer
1160153279 J * bogus_ ~bogusano@fengor.net
1160153279 Q * bogus Read error: Connection reset by peer
1160153381 J * brc_ ~bruce@201.19.147.226
1160153409 Q * brc_ 
1160153413 J * brc_ ~bruce@201.19.147.226
1160153564 J * matled ~matled@85.131.246.184
1160154337 M * daniel_hozac Bertl_zZ: pong, sorry for the delay...
1160154471 Q * virtuoso Ping timeout: 480 seconds
1160155194 J * magnus_remove ~mange@h102n2fls31o823.telia.com
1160155242 M * magnus_remove I like debian etch beta 3, Remove Azeem or what the hell its name is in #linux/#debian or im gonna get mad
1160155275 M * daniel_hozac huh?
1160155277 M * magnus_remove What cudoz do i have.. for one, gadmintools
1160155290 M * Loki|muh hmmmm
1160155296 M * daniel_hozac i think you're in the wrong channel :)
1160155310 M * magnus_remove I want all the bots out of #debian
1160155322 M * daniel_hozac and how does bitching at us help with that?
1160155331 N * magnus_remove Magnus
1160155341 N * Magnus Mange
1160155389 M * Mange daniel_hozac: ill get mad and shit and thats never good :)
1160155421 M * daniel_hozac and why does that affect us/how can we affect that?
1160155442 M * Mange Its ok, Debian will kick Azeem anyhow.. no worries
1160155484 M * Mange It doesnt affect me... i like the way we have always coded give or take a few cvs/etc
1160155508 M * Mange who are you  daniel.. whats the workload ?
1160155519 M * daniel_hozac hmm?
1160155535 M * Loki|muh hmmmm, I would say its a problem of too much C2 ;)
1160155546 M * Mange much weird patches to apply that you have to wedge in there ?
1160155567 M * Mange Loki|muh: no, C02 is the bong! :)
1160155583 M * daniel_hozac Loki|muh: is that C4's predecessor? :)
1160155596 M * Mange Ps, Loki games Pwns
1160155644 M * daniel_hozac just making sure... Mange: you realize you're in #vserver, right?
1160155652 M * Mange We want to sponsor some devel on loki, seen it or faking Loki son ?
1160155676 M * Loki|muh I thought of Ethanol ;)
1160155703 M * Mange Ah, so then atleast youre a bit funny (I dont like booring)
1160155705 M * Mange :)
1160155779 Q * hkhl Read error: Connection reset by peer
1160155781 M * Mange weasel::Sweden yearly attach billions of goodness.. now i want a bigscreentv damnit!!! :)
1160155829 M * tokkee Has this channel been renamed to #misc-talk?
1160155836 M * Mange AsstittetittietittieASS.. repeat over and over
1160155846 M * daniel_hozac tokkee: seems like it.
1160155855 N * hardwire-afk hardwire
1160155870 M * Mange tokkee: did you have anything important to say ?
1160155878 M * Mange or just rambled on
1160155893 M * tokkee Mange: Nobody seems to be interested in what you're talking about...
1160155905 M * Mange daniel had anything important to say ?
1160155918 M * tokkee ... besides that nobody seems to be able to make sense of it.
1160155942 M * Mange tokkee: i just hope you are good enough not to be useless my friend
1160155964 M * tokkee I am goddamn useless...
1160155986 M * Mange Wehn an idiot rules a channel theyll want people with lesser skills..
1160156011 M * Mange Are youi of lesser will ?
1160156026 M * Mange Perhaps with a slice of Dill ?
1160156090 M * Mange Do you know why people think my code PWNZ ?
1160156106 M * daniel_hozac no, nor did we know you write code.
1160156118 M * Mange  what you write ?
1160156118 M * daniel_hozac because, you know, we've never seen you before.
1160156132 M * Mange Im lookin into writing myself
1160156140 M * meandtheshell Mange: Please stop. Stay on channel topic.
1160156148 M * Mange so, im not the B00ngcoder ?
1160156156 M * Loki|muh *plonk* ;)
1160156160 M * tokkee .oO( people should go to bed after drinking too much... )
1160156167 M * Mange topic's booring, i hate booring
1160156177 M * daniel_hozac tokkee: or at least a ditch or something.
1160156180 M * Mange www.gadmintools.org ?
1160156204 M * Mange See, lameness doesnt count, coolness does.. its always been like that
1160156207 M * tokkee daniel_hozac: ack
1160156211 M * daniel_hozac so what do you guys think, should i use my powers?
1160156216 M * tokkee Could somebody please kick this guy?
1160156219 M * tokkee :-)
1160156219 M * Mange YEs! :)
1160156223 M * meandtheshell daniel_hozac: yes - please
1160156224 M * daniel_hozac thank you, that's all i needed :)
1160156229 F * ChanServ +o daniel_hozac
1160156237 M * Mange cool
1160156239 K Mange daniel_hozac please stop
1160156239 J * Mange ~mange@h102n2fls31o823.telia.com
1160156242 M * meandtheshell rofl
1160156249 M * Mange heh
1160156249 M * meandtheshell daniel_hozac: thank you daniel
1160156259 M * Mange I appreciate it
1160156355 M * Mange I think youre somewhat crazy but if you do that again itll take ages before back onTarnets..
1160156373 M * tokkee +b helps a lot...
1160156379 M * Borg- Mange: you are script kiddie?
1160156381 M * Mange Vitamins ?
1160156383 M * daniel_hozac tokkee: i believe in second chances.
1160156395 M * tokkee daniel_hozac: He did have his.
1160156402 M * Mange Borg-: nope, im coding for you.. www.gadmintools.org
1160156434 M * Mange But i will be angry if booted a whole bunch of times
1160156441 M * node Mange: so?
1160156449 M * Mange Borg-: ?
1160156481 M * ntrs mange seems to be an AI bot
1160156500 M * ntrs i think a ban is in order
1160156508 M * daniel_hozac ntrs: surely AI would be able to create understandable sentences?
1160156512 M * Mange so, you dont like a 30 yearold in your channel.. with all the suggestions and the knowhow ?
1160156521 M * ntrs a bad AI bot, that is.
1160156544 M * Loki|muh yeah but he sounds drunk. or stoned ;)
1160156550 M * daniel_hozac or both!
1160156559 M * Mange 213.114.20.44: have you a hearing aid ?
1160156562 M * Borg- Mange: are you script kiddie?
1160156581 M * Mange Severity::Lamers
1160156589 M * Borg- bot
1160156593 M * Mange Moron
1160156623 M * Mange Toool: looser: request tiites and a large drink :)
1160156638 M * Mange Borg-: Cmon man
1160156656 M * Mange Im having "Thor" pretty good beer
1160156665 F * daniel_hozac +b *!*@h102n2fls31o823.telia.com
1160156668 K Mange daniel_hozac Mange
1160156680 F * daniel_hozac -o daniel_hozac
1160156681 M * Loki|muh too bad - it was getting funny *g*
1160156688 M * daniel_hozac hehe.
1160156700 M * Borg- Loki|muh: hehe yeah ;)
1160156840 M * node daniel_hozac: so i was looking at the vserver kernel patch last night, and i noticed that security_ops in security/security.c is still a global resource
1160156852 M * node which means that LSM can't be vserver-aware
1160156873 M * daniel_hozac ... unless you fix it :)
1160156883 M * node okay, just wondering if there was any work already done on it
1160156899 M * node so, i was also thinking, where is an appropriate place to push security_ops?
1160156910 M * node perhaps into struct vx_info?
1160157022 M * daniel_hozac would seem like the most logical place to put it.
1160157040 M * node alright
1160158184 Q * shedi Quit: Leaving
1160158367 M * micah daniel_hozac: it seems like etch general freeze is in 12 days, and it takes 10 days for packages to move from unstable to testing to be included, which means any change will need to be in in 2 days :P
1160158392 M * daniel_hozac hmm, ok.
1160158394 M * daniel_hozac mugwump_: ping!
1160158410 M * micah that detail was just brought to my attention :(
1160158440 M * daniel_hozac i'm fairly confident that current HEAD is as bug-free as it'll get, but i still have to read through the diff to make sure i didn't do any overly stupid things...
1160158515 M * micah daniel_hozac: I think its worse
1160158533 M * daniel_hozac ?
1160158540 M * daniel_hozac i.e. it's already too late?
1160158547 M * micah daniel_hozac: getting things in today is the only way to be safe, as it can take one day for dinstall, and one day for brittney on either end
1160158554 M * micah although it might work in less time
1160158561 M * daniel_hozac ah.
1160158565 M * daniel_hozac mugwump_: superping! then :)
1160158588 M * micah unless urgency medium is specified (which changes things from 10 days to 5), but we'd need to have a reason for it (although we can probably find one)
1160158628 M * tokkee micah: You got about 45 minutes until dinstall ;-/
1160158681 M * micah thats bad
1160159216 M * daniel_hozac so what's required to get urgency medium? :)
1160159238 M * daniel_hozac (i'm still just 35% through the diff, i don't think i'll make it in half an hour...)
1160160545 Q * FireEgl Ping timeout: 480 seconds
1160161209 J * virtuoso ~s0t0na@shisha.spb.ru
1160161579 M * matti :)
1160161665 M * hardwire :0
1160161695 M * hardwire :P
1160161715 M * hardwire monkey tounge monkey oooh monkey smile
1160162059 J * shedi ~siggi@inferno.lhi.is
1160162695 J * samuel ~samuel@adsl-68-122-125-205.dsl.pltn13.pacbell.net
1160162929 Q * cdrx Quit: Leaving
1160165474 Q * oli_ Quit: Leaving
1160165993 Q * hardwire Ping timeout: 480 seconds
1160166213 J * bonbons ~bonbons@83.222.36.111
1160166232 M * micah daniel_hozac: there isn't any hard rules, we likely can do medium without problems
1160166240 M * micah although there is still a chance...
1160166841 J * hardwire ~hardwire@89-208-58-66.gci.net
1160167158 N * Bertl_zZ Bertl
1160167165 M * Bertl hey folks!
1160167545 M * Rich_Estill Hey B.  
1160167556 M * Rich_Estill Well, been fun.  HAve a good weekend.
1160167559 P * Rich_Estill Leaving
1160168313 M * Bertl daniel_hozac: I decided to include your pidtype patch in 2.1.1, could you rerun your magic scripts on the patched kernel to eliminate the (now) obsolete includes?
1160168840 M * daniel_hozac Bertl: sure, i was just thinking about that the other day.
1160168892 M * daniel_hozac did you apply my headers patch?
1160168915 M * Bertl 85 or 84 or both?
1160168935 M * daniel_hozac umm, the one for devel.
1160168942 M * daniel_hozac (85 i think)
1160168948 M * daniel_hozac yes.
1160168980 M * Bertl I'll apply it
1160168986 M * daniel_hozac well, it doesn't really matter.
1160168994 M * daniel_hozac i was just wondering which tree to run it on ;)
1160169019 M * Bertl I can upload the current branch if you like
1160169022 M * daniel_hozac ok, sure.
1160169259 Q * transacid Quit: irssi neustarten
1160169319 Q * bonbons Quit: Leaving
1160169472 J * derjohn2 ~aj@227-153.0-85.cust.bluewin.ch
1160169543 Q * kaner Remote host closed the connection
1160169591 M * derjohn2 Bertl, i was not able to get rid of the NFS problem. even not after compiling the kernel with nfs_tagging stuff.
1160169615 M * Bertl strange, do you have a setup where this can be easily reproduced?
1160169834 M * derjohn2 hm, hm, i am not sure.
1160170150 Q * mire Quit: Leaving
1160170255 J * kaner kaner@strace.org
1160170580 J * are|lunch ~are@62.112.159.81
1160170697 N * are|lunch _are_
1160170891 Q * samuel Ping timeout: 480 seconds
1160171055 M * derjohn2 Bertl, 1st I have to make sure that the problem is not a pebkac.
1160171090 M * _are_ Hi
1160171151 M * derjohn2 hi Adrian!
1160171579 J * transacid ~transacid@transacid.de
1160171660 M * Bertl daniel_hozac: http://vserver.13thfloor.at/Experimental/patch-2.6.18-vs2.1.1-rc36.diff
1160172136 M * derjohn2 Bertl, might it have to do with 32 bit guest on a 64 bit host?
1160172255 M * fs Bertl: in case you missed my report here, I tested 2.6.18+vs2.0.2.1-t8 with hardcpu=y and hardcpu_idle=y
1160172278 M * fs the problem of slugish and jumpy mouse in X11 persists
1160172303 Q * meandtheshell Quit: exit (0);
1160172307 M * fs if the host has load, everything is fine, but: if the guest has load, the mouse jumps
1160172321 M * fs if the guest has load and the host does not, the mouse hangs too
1160172344 M * fs if I run the cpuhog in both the guest and the host, everything is fine again
1160172381 M * fs so this last point is different in comparison to the kernel without hardcpu and hardcpu_idle set
1160172403 M * Bertl strange
1160172448 M * fs indeed
1160172473 M * Bertl how many virtual cpus are there?
1160172480 M * hardwire isn't that weird
1160172486 M * hardwire virtual cpu's
1160172489 M * hardwire I think its weird
1160172494 M * fs I tested on a single CPU box
1160172499 M * fs no hyperthreading
1160172536 M * Bertl btw, that cpu hog you used (which showed it better) is that still what you use?
1160172597 M * fs itÄs the cpuhog.c you asked me to use
1160172600 M * fs it'
1160172601 M * fs s
1160172623 M * Bertl okay, give me a minute, I've updated my laptop today
1160172662 M * fs it's an amd64 box if that matters
1160172693 M * Bertl powerpc here, but if it is a sheduler issue, it should not matter
1160172848 J * coocoon ~coocoon@dslb-084-056-198-200.pools.arcor-ip.net
1160172848 M * Bertl cool, I can reproduce that here ...
1160173008 M * Bertl let's talk a little about what could be the reason for this behaviour
1160173042 M * Bertl fs: so we have the following situation: (correct me if I'm wrong somewhere)
1160173056 M * daniel_hozac Bertl: http://people.linux-vserver.org/~dhozac/p/k/delta-headers-fix86.diff
1160173066 M * Bertl - cpuhog in a guest makes X stutter
1160173080 M * Bertl - cpuhog on the host makes X behave as alwas
1160173084 M * Bertl *always
1160173111 M * fs yes exactly
1160173121 M * Bertl the X stuttering is visible especially on mouse moves
1160173129 M * fs yes
1160173149 M * Bertl a mouse move is basically a hardware interrupt, causing some visual update when procesed
1160173221 M * Bertl so as the hardware irq happens, it gets handled, but the userspace process has some delay until rescheduled
1160173240 M * Bertl that's what can be easily seen on the mousepointer
1160173245 Q * micah Remote host closed the connection
1160173255 J * micah ~micah@micah.riseup.net
1160173283 M * fs to be specific: I have usb keyboard and mouse; typing on the keyboard is fine, but moving the mouse is jumpy and pasting a selection happens after a dozen of seconds
1160173309 N * hardwire hardwire-afk
1160173320 M * fs so you can paste a selection, and hit enter several times before the selection actually shows up in the terminal
1160173327 M * Bertl indeed, same here, keyboard is fine, mouse jumpy
1160173365 M * Bertl so it's kind of mouse specific .. hmm hmm ...
1160173416 M * Bertl what really confuses me is that there should be no difference between guest or not guest process, as long as the scheduler is not activated
1160173427 M * fs hm
1160173450 M * Bertl but obviously there is
1160173474 M * Bertl how long does a test compile/run take you?
1160173527 M * Bertl because I have an idea what could affect this 
1160173548 M * fs around half an hour
1160173556 Q * derjohn2 Quit: Verlassend
1160173559 M * Bertl so long, on an amd64?
1160173570 M * fs if I deactivate all other images beside the vserver-amd64 one
1160173581 M * fs well, it has everything activated as module
1160173594 M * daniel_hozac and you're doing complete rebuilds?
1160173595 M * Bertl ah, you are building the kernel from scratch every time?
1160173613 M * fs I do build the debian kernels, from scratch but with ccache
1160173618 M * fs it might take less
1160173641 M * Bertl okay, let's get a test kernel tree, with reduced modules and such
1160173654 M * Bertl also build it once and keep the built tree around
1160173670 M * Bertl a change and compile should then take roughly 2 minutes
1160173718 M * Bertl I'll try to provide substantial data what causes the misbehaviour and then we can try to fix it
1160173825 M * fs ok
1160174041 Q * coocoon Quit: KVIrc 3.2.0 'Realia'
1160174337 M * fs I am doing a build with the status quo right now
1160174359 M * ntrs Bertl, what's in rc36, should I go up to 36 right away or is 35 still good?
1160174384 M * Bertl rc36 is just a status for daniel_hozac, no need to jump on it
1160174423 M * ntrs ok
1160174660 M * daniel_hozac Bertl: btw, what's the cvirt file at the top of the tree? cscope?
1160174832 M * Bertl garbage, just ignore it
1160174968 M * fs bertl: ok I am on sync, just point me to an url with a patch or something when you are done
1160175170 M * Bertl I have two theories ATM, one is that the task slice is miscalculated in the guest case, giving a much longer task slice to the process than desired
1160175191 M * Bertl the other one is that the need rescheduling when in polling is not working as expected
1160175222 M * Bertl I'm now thinking about a good way to get data to test that
1160175258 M * Bertl we basically have a complete scheduling monitor built into the kernel, but no userspace tools yet
1160175324 M * fs hmm =)
1160175435 Q * gerrit Remote host closed the connection
1160175534 J * gerrit ~gerrit@c-67-160-146-170.hsd1.or.comcast.net
1160175590 M * fs Bertl: well, I am falling asleep now, I might be more of a help tomorrow 
1160175608 P * stefani I'm Parting (the water)
1160175767 Q * gerrit Remote host closed the connection
1160175769 Q * lilalinux_ Remote host closed the connection
1160175956 M * Bertl fs: okay, np
1160176061 J * samuel ~samuel@adsl-68-122-125-205.dsl.pltn13.pacbell.net
1160176220 Q * dna_ Quit: Verlassend
1160176559 M * Bertl guess it's the right time to implement the userspace part of the scheduling monitor :)
1160177027 M * doener evening!
1160177063 M * doener Bertl: scheduling monitor? what does that mean/need/foo?
1160177088 M * Bertl do you want to help?
1160177188 M * doener yeah, I need to focus on something...
1160177195 M * doener (don't ask why)
1160177202 M * Bertl okay :)
1160177215 M * Bertl cscope: file search, monitor
1160177243 M * Bertl check the monitor.h and monitor.c files
1160177255 M * doener 2.6.18-vs2.1.1-rcX?
1160177265 M * Bertl yeah, or older ones too, the .h descibes the data structures
1160177277 M * Bertl it's there since half a year or so
1160177323 M * doener I'm on my laptop atm, which only has vanilla trees atm ;)
1160177332 M * Bertl what is missing (i.e. what we need to implement)
1160177367 M * Bertl doener: then get the latest rc36 :)
1160177372 M * doener got it
1160177408 M * Bertl .. is the following:
1160177422 M * Bertl - a syscall command interface to retrieve the data
1160177430 M * Bertl   (that's what I'm working on right now)
1160177444 M * Bertl - an userspace tool which collects the data for processing
1160177457 M * Bertl - a tool which can read/interpret the data (i.e. dump it somehow)
1160177496 M * Bertl best would be if you could focus on the last part, which should be fairly independant of the rest
1160177528 M * doener hm, 300MB /tmp is not enough for cscope...
1160177545 M * Bertl include/config/vserver/monitor.h
1160177549 M * Bertl kernel/vserver/monitor.c
1160177609 M * Bertl the data collecting tool will simply copy the data from the kernel to the disk as fast as possible
1160178888 M * matti Hello again Bertl ;)
1160178926 M * Bertl hey
1160178938 M * matti "_
1160178942 M * matti Ops.
1160178943 M * matti :)
1160178950 M * matti And goodnight ;)