1263945720 M * urbee Bertl
1263945721 M * urbee still alive?
1263945767 M * urbee which CPU would u say is better 
1263945779 M * urbee E5405 QC 2,0ghz with 12mb cache
1263945785 M * urbee or X3320 2,5ghz with 6mb cache
1263945815 M * urbee i'm looking for better performance in running gameservers
1263945829 M * urbee on guest virtual servers ofcourse :)
1263945878 M * urbee my oppinion is that cache beats the shit out of those megahertz so i belive a better choice would be e5405
1263945992 M * fLoo well
1263946004 M * fLoo i am fine with an core i7 920 (8 cores, 6gb ddr3 ram)
1263946021 M * urbee thats 1cpu, 8cores?
1263946023 M * urbee damn :)
1263946036 M * urbee i'm behind time with my e5410's :p
1263946143 M * urbee core i7 920
1263946149 M * urbee this is actually 4cores?
1263946437 M * urbee lol
1263946440 M * urbee E7440 xeon
1263946444 M * urbee 16mb cache
1263946452 M * urbee &costs like 4k euro
1263946657 M * fzylogic Bertl: I figured out how to trigger the OOM killer system deadlock problem.  It seems related to the new pagefault OOM changes in mainline
1263947019 M * Bertl fzylogic: care to exlpain?
1263947051 M * Bertl urbee: I'm sure there are dozens of reviews online comparing various cpus
1263947067 M * fzylogic I wrote a little program to allocate an amount of memory between the rss.soft and rss.hard limits
1263947098 M * fzylogic slightly more than half the hard limit to be precise
1263947111 M * Bertl okay?
1263947128 M * fzylogic it then does a second malloc/memcpy of a smaller mount (10MB)
1263947143 M * fzylogic if you run 2 in a loop in 2 separate shells on the guest, you get a soft cpu lockup  :)
1263947171 M * Bertl okay, could you upload that program for me to test?
1263947193 M * fzylogic yeah, one sec
1263947308 M * fzylogic http://karategerbil.com/kernel_debug/memeater.c
1263947320 M * fzylogic lockup backtrace:
1263947320 M * fzylogic http://karategerbil.com/kernel_debug/soft_lockup.log
1263947332 M * fzylogic this is an otherwise unused host
1263947374 M * fzylogic and the rlimit settings:
1263947374 M * fzylogic http://karategerbil.com/kernel_debug/rlimits/
1263947465 M * fzylogic oh, that scanf() was for testing.  you can either take it out or run the loop with "< /dev/null"
1263947492 M * urbee Bertl: i was just wondering which one u would personally choose :)
1263947499 Q * cehteh Ping timeout: 480 seconds
1263947505 M * urbee i'm renting a machine and i need to decide
1263947625 M * Bertl well, I would have to do some research there too, so if you want to spend some money on me doing the research for you, that's fine :)
1263947792 M * fzylogic urbee: FWIW, we're running almost 200 E5405s with good results
1263947812 M * fzylogic though with an admittedly different workload
1263947946 M * urbee fzylogic: i'm pretty happy with e5410's too
1263947978 M * urbee Bertl: no need for that :p
1263948008 M * urbee gnite guys
1263948059 J * cehteh ~ct@pipapo.org
1263948461 M * Bertl fzylogic: so as argument I specify the soft limit in bytes, yes?
1263948473 M * fzylogic I've just been using the default in all my runs
1263948497 M * fzylogic but yes, that should work as well if you have different limits you want to tune for
1263948519 M * Bertl I'm currently going to test with 512 pages soft and 1024 hard
1263948529 M * fzylogic ok
1263948547 M * Bertl so I think I need to specify 4194304 bytes, yes?
1263948570 M * Bertl hmm, probably 2097152
1263948575 M * fzylogic it can be any value between hard and soft
1263948583 M * fzylogic well, between hard and hard/2
1263948610 M * Bertl okay, let's try that then, and I start it twice, yes?
1263948615 M * fzylogic yes
1263948628 M * fzylogic I've been doing the following in a pair of shells:
1263948629 M * fzylogic while true;do ./mem < /dev/null;done
1263948640 M * fzylogic by the second iteration, the machine has been locking
1263948660 M * fzylogic followed a minute later by the soft lockup messages on console
1263948701 N * micah_ micah
1263948723 M * Bertl hmm, here the first memeater process gets OOM killed
1263948867 M * Bertl should I try with a lower value?
1263948891 M * fzylogic yeah, I'd try that
1263948943 M * fzylogic it could be that the system's using enough memory that the first one pushes the guest over the edge before the second run can trigger the bug
1263949084 M * Bertl hmm, same with half the amount of memory
1263949096 M * Bertl (specified on the command line)
1263949129 M * fzylogic ohhh, yeah.  I was wrong  :)
1263949137 M * fzylogic the command-line picks chunk sizes, not upper limit
1263949151 M * fzylogic it's the memory variable that chooses the limit
1263949174 M * Bertl okay, so what do I need to change to match the 512/1024 limit?
1263949188 M * fzylogic int memory = 300000000
1263949208 M * fzylogic changing it to 2097152 would presumably match your setup
1263949226 M * Bertl let's try that, i.e. I'll make it an argument then :)
1263949247 M * fzylogic yeah, you can tell that code's not exactly written with end-users in mind  :)
1263949348 M * fzylogic our base install is too heavy to fix in the 512/1024 limits so I can't test that on my end
1263949353 M * fzylogic fit*
1263949600 M * Bertl hmm, I'm just setting them before I enter the guest (with vlimit)
1263949636 M * fzylogic our base install has apache, jabber, etc.
1263949640 M * fzylogic lots of cruft  :-/
1263949714 M * Bertl okay, but you can simply shut that down before setting the limits, no?
1263949753 M * Bertl anyway, same thing with specifying 2097152 for the mem part
1263949765 M * Bertl i.e. gets simply and cleanly killed off by OOM
1263949774 M * fzylogic ok
1263949777 M * fzylogic lemme try this...
1263949812 M * Bertl I can adjust the limit up to 1GB, although I'd prefer a smaller one
1263949889 Q * dowdle Remote host closed the connection
1263949931 M * fzylogic isn't 512 just a couple megabytes?
1263949946 M * fzylogic 73600 is only 256M
1263949968 M * Bertl yeah, just for a start, my test guest has just a shell running
1263950001 M * Bertl 512 is 2MB with a page size of 4k
1263950163 Q * hparker Quit: Quit
1263950388 M * fzylogic I multiplied your limits by 10 and triggered it
1263950389 Q * cehteh Ping timeout: 480 seconds
1263950405 M * fzylogic set the memory limit to 12000000
1263950412 M * fzylogic and the chunk size to 1000000
1263950424 M * Bertl okay, let me try
1263950573 J * cehteh ~ct@pipapo.org
1263950669 M * Bertl hmm, can't reproduce ... SMP required?
1263950688 M * fzylogic haven't tried on a non-smp...sounds possible
1263950718 M * fzylogic I can give you access to this host if you'd like
1263950735 M * Bertl nah, let me try with smp first
1263950983 M * Bertl hmm, even with 4 cpus, all I get is an OOM kill, let me upload my 'modifications' to your program and the test parameters
1263950990 M * fzylogic ok
1263951285 M * Bertl http://vserver.13thfloor.at/Experimental/TOOLS/memeater.txt
1263951329 M * fzylogic where's your modified source?
1263951334 M * Bertl same dir contains the source and the binaries, compiled with diet gcc -m32/64 -Wall -O2 -o memeater.32/64 memeater.c
1263951369 M * fzylogic ah, k
1263951779 M * fzylogic you sure it's not triggering the bug on your system?
1263951784 M * fzylogic it's not necessarily obvious at first
1263951796 M * fzylogic [  157.840998] BUG: soft lockup - CPU#2 stuck for 61s! [memeater.64:6491]
1263951796 M * fzylogic [  157.844038] CPU 2:
1263951796 M * fzylogic [  157.844038] Pid: 6491, comm: memeater.64 Not tainted 2.6.32.2-oomfix5-aufs2-vs2.3.0.36.28 #14 X7DBU
1263951796 M * fzylogic [  157.844038] RIP: 0010:[<ffffffff810ae85b>]  [<ffffffff810ae85b>] handle_mm_fault+0x6ab/0xb80
1263951812 M * fzylogic in fact, I thought it didn't work on mine until I tried to edit a file with nano
1263951818 M * fzylogic it hung opening the new filehandle
1263952468 M * fzylogic also, I just tested it against 2.6.31.5-vs2.3.0.36.21 and the bug is present there as well
1263952524 Q * cehteh Ping timeout: 480 seconds
1263952543 M * fzylogic and that's different hardware.  2.6.31.5 is on an opteron 1218HE
1263952721 M * Bertl I'm testing with 2.6.32.3 here, but let me test with 2.6.31.11/12
1263952746 M * Bertl and no, I didn't get a lockup, but I'm not sure I have the soft lockup checks enabled
1263952766 M * Bertl I'll check that too in a sec
1263952775 M * fzylogic that nano I ran after running memeater hung the terminal it started in
1263952803 Q * Piet_ Remote host closed the connection
1263952821 M * fzylogic and I need to head home so I'll check back in tomorrow.  email's jeremy@newdream.net if you have any questions before that
1263952832 M * Bertl okay, tx
1263952841 M * fzylogic thank you
1263952934 J * Piet_ ~Piet__@04ZAAAKNV.tor-irc.dnsbl.oftc.net
1263953069 J * cehteh ~ct@pipapo.org
1263953380 Q * infowolfe Quit: Leaving
1263959266 M * Bertl off to bed now ... have a good one everyone!
1263959270 N * Bertl Bertl_zZ
1263960066 J * SauLus_ ~SauLus@c192024.adsl.hansenet.de
1263960326 J * infowolfe ~infowolfe@c-71-236-152-35.hsd1.or.comcast.net
1263960475 Q * SauLus Ping timeout: 480 seconds
1263960475 N * SauLus_ SauLus
1263965471 Q * niki Quit: Leaving
1263966406 J * fzylogic_ ~fzylogic@wireless.fzylogic.net
1263966434 Q * fzylogic_ 
1263966446 J * fzylogic_ ~fzylogic@wireless.fzylogic.net
1263967245 Q * balbir Ping timeout: 480 seconds
1263967387 M * fzylogic_ Bertl_zZ: Turns out the memory values you pick are more important than I'd thought.  Here's a version that'll walk the available memory in arbitrary increments:
1263967388 M * fzylogic_ http://karategerbil.com/kernel_debug/memeater2.c
1263967413 M * fzylogic_ When you're back up, I'd be interested to see if you still can't trigger it  :)
1263967484 J * ghislain ~AQUEOS@adsl2.aqueos.com
1263967712 J * harobed ~sklein@arl57-1-82-231-110-14.fbx.proxad.net
1263968164 J * sharkjaw ~gab@90.149.121.45
1263968329 Q * fzylogic_ Quit: fzylogic_
1263970145 J * balbir ~balbir@122.248.161.59
1263970624 Q * ghislain Quit: Leaving.
1263970736 J * geos_one ~chatzilla@chello084115149052.4.graz.surfer.at
1263970930 Q * derjohn_foo Ping timeout: 480 seconds
1263971899 Q * harobed Ping timeout: 480 seconds
1263972467 J * derjohn_foo ~aj@213.238.45.2
1263973477 J * niki ~niki@cpe.fe4-0-120.0x50a6de52.kdnxd4.customer.tele.dk
1263976333 J * derjohn_mob ~aj@213.238.45.2
1263976333 Q * derjohn_foo Read error: Connection reset by peer
1263976448 J * {DRagON} ~lm@krasnet-gprs.etk.ru
1263976594 Q * {DRagON} 
1263976720 Q * balbir Ping timeout: 480 seconds
1263977364 J * cga ~cga@89-96-63-86.ip11.fastwebnet.it
1263977574 Q * cga 
1263977654 J * balbir ~balbir@122.248.163.1
1263978196 J * dna ~dna@170-198-103-86.dynamic.dsl.tng.de
1263978363 J * barismetin ~barismeti@zanzibar.inria.fr
1263978974 J * awk ~awk@gw1.security.web.za
1263978977 M * awk hi!
1263979045 M * awk hmm... guys trying to install a ubuntu system inside a centos host (parent)
1263979046 M * awk http://pastebin.com/m1e502ba1
1263979068 M * awk as you can see its complaining about debootstrap which I have downloaded and put in the directory it requested.. I even tried to rename it to 1.0.15
1263979929 J * kir ~kir@swsoft-msk-nat.sw.ru
1263980568 Q * orzel Ping timeout: 480 seconds
1263981428 Q * xdr Ping timeout: 480 seconds
1263981966 J * xdr ~xdr@h-238-149.A219.priv.bahnhof.se
1263982285 Q * dna Quit: Verlassend
1263983493 N * Bertl_zZ Bertl
1263983497 M * Bertl morning folks!
1263983523 M * Bertl awk: who requested to put it anywhere?
1263984684 Q * balbir Ping timeout: 480 seconds
1263985036 M * awk hmm, when I run the vserver build command it says checking for it cant find it and then it says please download and put in that uri directory
1263985038 M * awk ?
1263985246 M * Bertl it says, you should check where it is 'search the URL for the most recent *.deb ...'
1263985272 M * Bertl and then put that url into the file /etc/vservers/.defaults/apps/debootstrap/uri
1263985301 M * Bertl it doesn't tell you to download anything or to link files into some directory
1263985443 M * Bertl (and if you do that, it will be able to download the latest debootstrap and install quite fine)
1263985509 J * ktwilight ~keliew@151.83-240-81.adsl-dyn.isp.belgacom.be
1263985522 M * Bertl i.e. remove the downloaded file including the directory you created to put it in, and replace it by a file containing the url of the latest debootstrap for your platform
1263989473 J * SubZero ~SubZero@chello089076140236.chello.pl
1263990013 M * Bertl off for now ... bbl
1263990016 N * Bertl Bertl_oO
1263990640 J * AndrewLe1 ~andrew@u7.hlc.edu.tw
1263990641 Q * AndrewLee Read error: Connection reset by peer
1263990664 Q * awk Ping timeout: 480 seconds
1263990709 J * awk ~awk@gw1.security.web.za
1263991375 M * awk ahhh, sorry, my stupid
1263991382 M * awk missunderstood
1263991629 M * awk works tx
1263991895 Q * awk 
1263992410 J * balbir ~balbir@122.172.144.183
1263996895 Q * AndrewLe1 Ping timeout: 480 seconds
1263997801 Q * sharkjaw Remote host closed the connection
1263997817 J * thalunil ~thalunil@walledcity.de
1263998215 Q * niki Quit: Leaving
1263998460 Q * Piet_ Ping timeout: 480 seconds
1263998735 J * Piet_ ~Piet__@04ZAAALAL.tor-irc.dnsbl.oftc.net
1264000241 J * ghislain ~AQUEOS@adsl2.aqueos.com
1264000812 J * dna ~dna@170-198-103-86.dynamic.dsl.tng.de
1264001200 Q * SubZero Ping timeout: 480 seconds
1264001353 N * Piet_ Piet
1264001396 J * ktwilight_ ~keliew@6.169-247-81.adsl-dyn.isp.belgacom.be
1264001550 Q * ktwilight Ping timeout: 480 seconds
1264003231 Q * xdr Ping timeout: 480 seconds
1264003809 J * dowdle ~dowdle@scott.coe.montana.edu
1264005979 Q * Piet Remote host closed the connection
1264006130 J * Piet ~Piet__@04ZAAALEP.tor-irc.dnsbl.oftc.net
1264006279 J * hparker ~hparker@linux.homershut.net
1264007309 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d
1264007353 Q * jrklein Ping timeout: 480 seconds
1264007607 Q * balbir Read error: Operation timed out
1264007967 Q * barismetin Quit: Leaving...
1264008295 J * AndrewLee ~andrew@u7.hlc.edu.tw
1264008499 J * balbir ~balbir@122.172.55.187
1264010468 J * SubZero ~SubZero@chello089076140236.chello.pl
1264010718 Q * kir Quit: Leaving.
1264011095 Q * balbir Ping timeout: 480 seconds
1264011105 J * xdr ~xdr@h-238-149.A219.priv.bahnhof.se
1264011844 J * balbir ~balbir@122.172.55.187
1264012061 J * vServer_User ~vServer_U@host90-152-15-246.ipv4.regusnet.com
1264012405 Q * balbir Ping timeout: 480 seconds
1264012662 Q * vServer_User Remote host closed the connection
1264013233 J * ghislain1 ~AQUEOS@adsl2.aqueos.com
1264013462 J * vServer_User ~vServer_U@host90-152-15-246.ipv4.regusnet.com
1264013595 Q * ghislain Ping timeout: 480 seconds
1264013646 Q * vServer_User Read error: Operation timed out
1264013675 Q * derjohn_mob Ping timeout: 480 seconds
1264014655 J * hijacker_ ~hijacker@87-126-142-51.btc-net.bg
1264017342 Q * FireEgl Quit: Leaving...
1264017567 J * xdr_ ~xdr@h-238-149.A219.priv.bahnhof.se
1264017619 Q * xdr Read error: Connection reset by peer
1264017767 Q * fback Quit: leaving
1264017833 J * fback fback@red.fback.net
1264018360 Q * geos_one Read error: Connection reset by peer
1264018766 J * geos_one ~chatzilla@chello084115149052.4.graz.surfer.at
1264019543 J * derjohn_mob ~aj@c135217.adsl.hansenet.de
1264020009 Q * derjohn_mob Remote host closed the connection
1264020085 J * derjohn_mob ~aj@c135217.adsl.hansenet.de
1264020750 M * urbee sys-kernel/vserver-sources-2.2.0.7
1264020752 M * urbee thi sis kinda old isnt it?
1264020966 M * daniel_hozac yes.
1264022391 Q * geos_one Ping timeout: 480 seconds
1264022448 M * mnemoc well... it's still "officially" the last stable
1264022815 J * geos_one ~chatzilla@chello084115149052.4.graz.surfer.at
1264022923 Q * hijacker_ Quit: Leaving
1264023647 J * sharkjaw ~gab@90.149.121.45
1264024335 J * imcsk8 ~ichavero@148.229.1.11
1264024969 Q * bonbons Quit: Leaving
1264026621 M * Mr_Smoke urbee: it works really well though
1264026862 J * derjohn_foo ~aj@c193198.adsl.hansenet.de
1264026906 J * dna_ ~dna@170-198-103-86.dynamic.dsl.tng.de
1264027245 Q * dna Ping timeout: 480 seconds
1264027285 Q * derjohn_mob Ping timeout: 480 seconds
1264028676 Q * sharkjaw Ping timeout: 480 seconds
1264028896 Q * SubZero 
1264029597 Q * imcsk8 Quit: Leaving
1264031052 J * ghislain ~AQUEOS@adsl2.aqueos.com
1264031283 Q * ghislain 
1264031328 Q * ghislain1 Ping timeout: 480 seconds