1156985306 P * Skram 1156985306 J * Skram ~Mark@hermes.sentiensystems.com 1156985642 J * MrX ~urk@219.95.12.208 1156990286 J * ekc2 ~EKC@netblock-66-245-252-180.dslextreme.com 1156992026 M * ekc2 What is the best way to implement per-vserver bandwidth quotas? (both egress and ingress) 1156992099 M * Skram hmmm 1156992125 M * Skram Since each VServer has its own IP, you could try ipfm 1156992133 M * Skram (runs on the host of course) 1156992147 M * Skram openvcp.org uses it... 1156992172 M * ekc2 interesting. never heard of it. reading about it now. 1156992239 M * ekc2 it looks like ipfm just logs bandwidth consumption 1156992287 M * ekc2 is there an equivalent of the vser cpu token bucket scheduler for vserver network packets? 1156992621 M * Skram im sorry, i was in a ruch and didnt see the quotas part 1156992622 M * Skram hrmm 1156992642 M * Skram Well, I had a system going at one point which used iptables and mysql to log packets and bytes per ip 1156992669 M * Skram i did it by IP as each VPS on my hosts has a dedicated WAN IP, but you could probably do it via interface, etc. 1156992694 M * Skram but as far as limiting.. i dont think there is anything implemented.. one would have to write their own scripts, etc. 1156992752 M * ekc2 yeah. i was hoping someone had already figured this out. there should be a way to take advantage of the exponential backoff feature of tcp to throttle bandwidth by dropping packets or delaying them 1156992792 M * Skram you could do shaping or whatever with lartc, no? 1156992818 M * Skram I once looked into it, but not being a network-guru and somewhat busy, i put that at the end of my todo list 1156992889 M * Johnnie I don't think it'd be too hard to disable the interface or something if you used rrdtool to analyze traffic through iptables or ipfm or something. 1156992913 M * Skram right 1156992915 M * ekc2 bingo. never heard of that, either. looks complicated. this might do it, though. 1156992919 M * Johnnie I'm sure one could write a cron script to poll every X amount of minutes to see how much traffic has passed to that IP/interface and choke it off accordingly. 1156992938 M * Skram you use iptables or ipfm or cacct to account for traffic, then have some script to check and decide whether to not accept anymore traffic, etc. 1156992957 M * Johnnie Well, Cacti relies on rrdtool. 1156992976 M * Johnnie There's also ipac-ng. 1156992978 M * Skram Never messed with cacti.. you could also use rrdcollect(d).. pretty cool 1156992991 M * ekc2 yeah, but i don't want to cut of traffic to the vserver. I want to throttle the bandwidth from, say, 10mbit to 1mbit 1156993002 M * Skram use rrdcollect with /proc/virtual//cacct (i think) 1156993018 M * Skram lartc is the answer then.. and I know nothing more about it :) 1156993022 M * Johnnie Then, if you're using iproute2, I'm sure you could manage to do some means of throttling. 1156993035 M * Johnnie Well, at least control and polling... 1156993044 M * Johnnie The throttling would inevitably come from elsewhere. 1156993050 M * Johnnie i.e. Skram's suggestion 1156993089 M * ekc2 yes, latrc looks like the answer. 1156993110 M * Skram ekc2: do make a wiki entry with your findings.. I am sure mroe than just you and I are interested 1156993127 M * Skram I am going to post my iptables/mysql stuff when I get a chance to clean it up and make better graphs 1156993194 M * ekc2 will do. i wish i didn't have to read all of this latrc documentation, though :) a lartc vserver post-start script would be nice 1156993219 M * Skram yeah 1156993228 M * Johnnie My consultation fee is $75/hr., ekc2. 1156993229 M * Johnnie LOL 1156993230 M * Skram lartc seemed overwhelming to me 1156993237 M * ekc2 lol 1156993253 M * Skram Johnnie is cheap. For me to do manual labor, which is the cheapest, I start at $100 1156993255 M * Johnnie I never really did any traffic shaping. 1156993264 M * Skram :P 1156993273 M * Johnnie I've relied on other devices to do that, not the Linux kernel. 1156993298 M * ekc2 do you use userspace tools like trickle? 1156993329 M * Skram not yet 1156993340 M * Johnnie Chapters 9 & 12 look like the focus of your reading on the LATRC HOWTO document. 1156993376 M * ekc2 I know there's some expensive hardware to do traffic shapping. but, this lartc capability in the kernel is impressive 1156993393 M * Johnnie And hey, look at that, I just learned something...there really is a network token bucket filter. 1156993447 M * Johnnie Check out section 9.2.2 ... that about sums it up. 1156993460 M * ekc2 hmmm. you're right. a "hierarchial token bucket" and a "token bucket filter" 1156993490 M * Johnnie That'll be $75. 1156993494 M * Johnnie Kidding...haha 1156993525 M * Johnnie man tc 1156993547 M * ekc2 it looks like i have to use connmark to mark packets to-from each vserver and then setup a seperate "token bucket" for each vserver 1156993657 M * ekc2 so far I have (for throttling the entire server): tc qdisc add dev bond0 root handle 1: htb default 60 1156993667 M * ekc2 tc class add dev bond0 parent 1: classid 1:60 htb rate 2kbit ceil 10kbit prio 0 1156993679 M * ekc2 however, wget doesn't seem to be throttled 1156993687 M * ekc2 perhaps tc doesn't work with bonding 1156993712 M * Johnnie iproute2 might be handy in that department. 1156994008 Q * DreamerC Quit: leaving 1156994027 J * DreamerC ~dreamerc@59-115-51-158.dynamic.hinet.net 1156994068 M * ekc2 it looks like I need to use the Intermediate Queueing Device (IMQ) and iproute2 instead of using physical network devices. This also also ingress bandwidth throttling (whereas vanilla lartc is only for egress) 1156994118 M * Johnnie Interesting. 1156994129 M * ekc2 someone at planetlab has written a vserver bandwidth throttling script: http://build.planet-lab.org/build/planetlab-3_2_1/SOURCES/resman-0.03/scripts/bwlimit 1156994203 M * ekc2 however, it doesn't work for ingress and bonded ethernet 1156994342 M * Johnnie Well, data typically moves at a synchronous rate. 1156994618 Q * olilo Quit: bbiab 1156995788 Q * ekc2 Remote host closed the connection 1156995811 J * ekc2 ~EKC@netblock-66-245-252-180.dslextreme.com 1156996428 J * olilo hiddenserv@tor.noreply.org 1156997707 Q * olilo Quit: bbiab 1156998155 J * olilo hiddenserv@tor.noreply.org 1156998257 Q * olilo 1156998417 Q * FireEgl Quit: Bye... 1156998498 Q * ekc2 Remote host closed the connection 1156999368 J * olilo hiddenserv@tor.noreply.org 1157001184 M * eyck was it planned that vserver-utils are no longer working: can not change context: migrate kernel feature missing and 'compat' API disabled: Function not implemented ? 1157001235 M * eyck (with 0.30.210-10), while 0.30.209 works fine and dandy 1157003108 J * yarihm ~yarihm@whitehead2.nine.ch 1157003175 J * balbir ~balbir@59.144.51.152 1157003470 Q * MrX Ping timeout: 480 seconds 1157004146 J * MrX ~urk@219.95.13.176 1157005529 Q * PowerKe Quit: Oops, wrong button 1157007515 Q * cdrx Ping timeout: 480 seconds 1157007568 J * dna ~naucki@61-224-dsl.kielnet.net 1157008248 Q * Aiken Ping timeout: 480 seconds 1157009130 J * ||Cobra|| ~cob@pc-csa01.science.uva.nl 1157009262 Q * starlein Ping timeout: 480 seconds 1157009297 J * meandtheshell ~markus@85-124-233-41.work.xdsl-line.inode.at 1157010090 M * weasel Bertl_zZ: will you ever start signing your patches? 1157010481 J * cdrx ~legoater@242.32.96-84.rev.gaoland.net 1157010742 J * Savvy ~SavvyChat@pppbur7-110.isnet.net.tr 1157012490 M * eyck probably as soon as 1.2.11 is released 1157012519 M * Savvy hi, is there a manual orf howto about quota in guests on a shared file system 1157012520 M * Savvy ? 1157014301 J * Milf ~Miranda@ipsio237.ipsi.fraunhofer.de 1157014328 M * Milf Anyone else having ICQ problems? 1157014710 J * coocoon ~coocoon@p54A0689F.dip.t-dialin.net 1157014725 M * coocoon morning 1157014746 M * dna Milf: 1157014747 M * dna Authentication Error! 1157014747 M * dna \\x00\\x1a 1157014747 M * dna http://www.aol.com?ccode=US&lang=en 1157014750 M * dna ;) 1157014805 M * Milf Ok, so you've got problems too? 1157014846 M * dna yes ;) 1157015622 Q * michal_ Ping timeout: 480 seconds 1157015910 J * michal_ ~michal@www.rsbac.org 1157016241 Q * olilo Remote host closed the connection 1157016304 J * olilo hiddenserv@tor.noreply.org 1157016906 Q * shedi Quit: Leaving 1157017320 Q * cdrx Ping timeout: 480 seconds 1157018741 M * michal_ anybody alive? 1157018757 M * michal_ i'm wondering why i'm getting 1157018758 M * michal_ chbind: vc_net_create(): Invalid argument 1157018767 M * michal_ when trying to start new vserver 1157019208 J * cdrx ~legoater@cap31-3-82-227-199-249.fbx.proxad.net 1157019269 M * michal_ seems to be some bond0 interface problems i gues 1157020367 M * michal_ nvm, working now :] 1157021425 M * derjohn michal_, nvm ? 1157021473 M * michal_ nevermind :] 1157021476 M * derjohn michal_, i would guess you compiled the kernel without network legacy api. did ayn vserver run on that kernel already? bondX works, my server "james bond" run guests ... 1157021482 M * derjohn oh ;) 1157021512 M * derjohn Milf, you here? ehlo !; ) 1157021547 Q * Belu_zZz Read error: Connection reset by peer 1157021609 M * michal_ everything is owkring now 1157021617 M * michal_ jsut vserver does not support dynamic xid anymore 1157021627 M * michal_ so i have set a static one, and it's flying :] 1157021642 Q * cryptronic Read error: Connection reset by peer 1157022819 M * nayco_work hello, all ! 1157023085 M * daniel_hozac eyck: yes, util-vserver will have to be compiled with --enable-apis=NOLEGACY to work on your ancient kernels, it's in NEWS :) 1157023115 M * daniel_hozac Savvy: AFAIK it's not implemented yet because no one has tested it. 1157023151 M * daniel_hozac michal_: that's what you get when you disable dynamic xids in the kernel ;) 1157023617 Q * Savvy Ping timeout: 480 seconds 1157024326 J * shedi ~siggi@dsl-149-109-85.hive.is 1157025089 Q * nokoya Quit: changing servers 1157025217 Q * bubulak Remote host closed the connection 1157025219 J * nokoya ~young@hi-230-82.tm.net.org.my 1157025640 M * eyck what do you mean by 'acient', they're brand new... 1157025653 M * eyck damn, bad keyboard :( 1157025671 M * eyck daniel_hozac: thanks, have you got any influence on util-vserver debs? 1157025695 Q * nayco_work Quit: leaving 1157025762 M * daniel_hozac no, but i'm sure if you ask micah nicely you can get the debs to use that option. 1157025968 M * eyck yeah, unfortunately for some strange reason, when I ask nicely, people get offended and stuff.. 1157026026 M * daniel_hozac lol 1157026465 J * bubulak ~bubulak@whisky.pendo.sk 1157027909 Q * olilo Remote host closed the connection 1157028395 J * Belu_zZz B.Lukas@mail.openvcp.org 1157028419 M * matti Hi eyck. 1157028429 J * cryptronic crypt@mail.openvcp.org 1157028492 A * Belu_zZz is away (iŽll be back later...) 1157029713 M * eyck hi matti 1157030653 Q * yarihm Ping timeout: 480 seconds 1157031112 M * mnemoc hi, OT: do you know any tool to analize the flow of a program (C) without executing it? 1157031290 N * Bertl_zZ Bertl 1157031295 M * Bertl morning folks! 1157031308 M * Bertl hey eyck! how is it going? 1157031310 M * mnemoc wb Bertl 1157031327 M * Bertl hey mnemoc! 1157031482 M * matti Hi Bertl. 1157031509 M * Bertl ah, matti! :] 1157031866 M * Bertl ntrs: ping? 1157031885 M * ntrs Yes? 1157031896 M * Bertl ah, here are the patches to test, when you get around 1157031898 M * Bertl http://vserver.13thfloor.at/Stuff/delta-rc31.2ra.diff 1157031904 M * Bertl http://vserver.13thfloor.at/Stuff/delta-rc31.2rb.diff 1157031918 M * Bertl i.e. unmodified devel + 2ra 1157031926 M * Bertl and if that doesn't help 1157031934 M * Bertl unmodified devel + 2rb 1157032070 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1157032134 M * eyck Bertl: fine, thank you, I crave sunlight. 1157032165 M * Bertl it's quite sunny here :) 1157032175 J * coocoon ~coocoon@p54A0689F.dip.t-dialin.net 1157032282 M * ntrs Bertl, were you able to contact ensc? What is the current status of the utils, will someone else take over? 1157032343 M * Bertl nope, ensc is not responding to my queries, messages or emails 1157032379 M * ntrs So, I guess someone else will take over the tools? 1157032384 M * eyck so, this means I'm taking over the tools? 1157032388 M * Bertl daniel_hozac: maybe you could try once again? 1157032394 M * eyck prepare for rewrite then! 1157032413 M * Bertl eyck: back to the stone age? :) 1157032430 M * eyck when real men were real men, 1157032436 M * Bertl lol 1157032449 M * eyck and furry creatures from alpha centauri were furry creatures from alfa centauri 1157032502 M * Bertl okay, I have to leave now ... have a lot of stuff to do today ... will be back in the evening ... 1157032536 M * Bertl ntrs: please keep daniel_hozac updated about the patch/test results ... 1157032550 M * ntrs ok 1157032554 N * Bertl Bertl_oO 1157032814 J * ybanafa ~admin@82.114.176.176 1157033231 M * ybanafa Hi All .. :) 1157033413 M * daniel_hozac welcome ybanafa! 1157034099 P * ybanafa Leaving 1157035245 M * Skram Hi all 1157035266 M * Skram eyck: you were the one trying to do bandwidth shaping/limiting/accounting "last night", right? 1157035294 M * daniel_hozac no, that was ekc2. 1157035485 M * tanjix when i try to start postfix in a vserver i get "fatal: could not find any loopback addresses" 1157035491 M * tanjix any ideas? 1157036273 M * daniel_hozac strace? it seems postfix can be problematic, http://linux-vserver.org/ProblematicPrograms 1157036280 M * daniel_hozac that particular issue isn't mentioned though. 1157037013 M * tanjix daniel_hozac: On 3000+ vservers Postfix runs well - only on one machine it will not. That's why I ask ;) 1157037051 M * daniel_hozac thus "can be" :) 1157037139 M * eyck Skram: nope, my bandwidthd shaping/limiting/accounting works fine. 1157037582 Q * ||Cobra|| Remote host closed the connection 1157038097 J * root ~root@proxy2.laroche.univ-nantes.fr 1157038144 M * root hello ! 1157038146 M * daniel_hozac you shouldn't IRC as root :) 1157038151 N * root nayco 1157038152 J * robig ~robig@83.221.253.78 1157038157 M * robig hello 1157038163 M * daniel_hozac hi 1157038183 M * nayco test 1157038188 M * nayco ah :-) 1157038344 J * stefani ~stefani@tsipoor.banerian.org 1157038579 M * robig can sb. tell me the url of the wiki? 1157038591 Q * Milf Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org 1157038616 M * coocoon robig: http://wiki.linux-vserver.org/Welcome_to_Linux-VServer.org 1157038621 M * robig the url at linux-vserver.org/Documentation is wrong 1157038655 M * matti robig: You can change it :) 1157038735 M * matti Oh my. 1157038741 M * matti New WiKi looks fancy ;] 1157038765 M * matti Well, maybe not new, but still fancy ;] 1157038767 M * robig ok :D dome 1157038768 M * robig done 1157038791 M * daniel_hozac robig: what? which URL? you realize linux-vserver.org _is_ the wiki, right? 1157038819 M * robig :) lol 1157038905 M * matti daniel_hozac: It uses WiKi as content manager only, but you got the point. 1157038907 M * matti :) 1157039089 M * daniel_hozac robig: that change is incorrect... 1157039114 M * robig so I'm sorry 1157039128 M * matti Hm. 1157039138 M * matti So, where is the true lvs WiKi? 1157039139 M * matti ;] 1157039139 M * daniel_hozac linux-vserver.org is the wiki, wiki.linux-vserver.org is the new wiki, that strahlungsfrei wiki i never understood the purpose of... 1157039151 M * matti Hm... 1157039172 M * matti We should use Track. 1157039177 M * matti As madwifi is. 1157039177 M * robig so remove that link, and publish the new wiki ;) 1157039190 M * matti They have really nice page lately. 1157039205 M * daniel_hozac the new wiki will go live along with the release of 2.0.2. 1157039212 M * matti robig: Hooo... New WiKi is a deep secret. 1157039218 M * matti robig: You already know too much. 1157039219 M * matti ;] 1157039226 M * matti ;p 1157039289 M * robig I've just vergotten.. what wiki? what is a wiki?... ;) 1157039309 M * matti Right... 1157039310 M * matti ;p 1157039319 M * matti You cannot escape now! 1157039319 M * matti ;] 1157039368 M * robig why not? are you using a firewall? ;) 1157039409 M * matti robig: As fas as I know, there's no life outside firewall, so ;p 1157039410 M * matti ;] 1157039416 M * matti s/fas/far/ 1157039540 M * robig I've read about private networking in your (old ;-) ) wiki.. that method creates one virtual lan interface per vserver, I'm right? 1157039579 M * daniel_hozac does anyone know who maintained the vserver.strahlungsfrei.de site? 1157039784 M * daniel_hozac robig: you'll have to link me. 1157039911 M * robig what do you mean? 1157039933 M * daniel_hozac i mean i have no idea what you're talking about. 1157040055 M * robig okay. I've found this http://oldwiki.linux-vserver.org/HowtoPrivateNetworking, and wanted to know how it works 1157040078 M * robig it will create a new net device for each vsevrver? 1157040084 M * robig -v 1157040097 M * ntrs daniel_hozac, unmodified devel + 2ra did not help. 1157040105 M * ntrs daniel_hozac, will try 2rb now 1157040109 M * daniel_hozac ok. 1157040163 M * daniel_hozac robig: no, it'll use the same device (dummy0) for all of them. 1157040193 M * robig that sounds good 1157040663 J * bonbons ~bonbons@83.222.36.236 1157040933 M * robig daniel_hozac: no other devices line dummy0.* ? 1157040986 M * daniel_hozac well, if you use the name file, you'll get dummy0:X. they're just aliases though, and not different interfaces. 1157041019 M * robig shh. ok thanks 1157041027 M * robig -s+a 1157041740 J * TuxedoMask ~Andrico@lncswibas01-pool7-a34.lncswi.tds.net 1157041757 Q * robig Ping timeout: 480 seconds 1157042494 J * Viper0482 ~Viper0482@p549773F9.dip.t-dialin.net 1157042691 Q * cdrx Quit: Leaving 1157044124 M * ntrs daniel_hozac, unmodified devel + 2rb works fine for over an hour. 1157044129 J * brc_ ~bruce@201.19.202.127 1157044147 M * ntrs I will keep it up and will report later again 1157044157 M * TuxedoMask Ok. 1157044171 N * TuxedoMask Andrico 1157044176 M * daniel_hozac ntrs: that's great, albeit very strange. 1157044262 M * ntrs daniel_hozac, well, I am just reporting. 1157044641 M * daniel_hozac ntrs: what arch is that on? 1157045453 M * daniel_hozac ntrs: could you possibly run objdump -d vmlinux --start-address=0x
--stop-address=0x
and paste the result on paste.linux-vserver.org? 1157045538 M * daniel_hozac (of the original devel kernel) 1157045544 M * daniel_hozac if you still have it, i mean. 1157045872 M * ntrs daniel_hozac, this is on i386 1157045908 M * ntrs what original devel kernel? I only have vmlinux from the 2rb kernel 1157045944 M * daniel_hozac ah, i thought you might've kept separate trees :) 1157046131 M * daniel_hozac if that indeed fixes it, could you try reverting 2rb and adding http://people.linux-vserver.org/~dhozac/patches/kernel/delta-xchg-test01.diff ? 1157046751 Q * shedi Quit: Leaving 1157047223 Q * matti Ping timeout: 480 seconds 1157047370 J * matti matti@linux.gentoo.pl 1157047715 J * FireEgl FireEgl@Sebastian.Atlantica.US 1157048055 Q * Andrico Quit: Ok, ich bin raus... 1157048136 J * yarihm ~yarihm@84-75-123-221.dclient.hispeed.ch 1157049519 J * ntrs_ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1157049519 Q * ntrs Read error: Connection reset by peer 1157049757 P * stefani I'm Parting (the water) 1157050559 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1157050687 Q * Viper0482 Remote host closed the connection 1157052050 J * ntrs__ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1157052051 Q * ntrs_ Read error: Connection reset by peer 1157053323 J * cdrx ~legoater@cap31-3-82-227-199-249.fbx.proxad.net 1157053587 Q * gerrit Ping timeout: 480 seconds 1157053792 J * rhodes ~rhodes@hc652a837.dhcp.vt.edu 1157053828 M * rhodes Can someone help with a vserver network config issue? 1157053851 M * daniel_hozac maybe, what's the problem? 1157053920 M * rhodes I have a Debian system, kernel 2.6.16.16 and have defined a vserver. Both are on the same subnet, I see the virtual interface from both system, but no traffic is routed back into the vserver 1157053970 M * rhodes If I tcpdump the traffic from the primary host, I see packets out and back in on the virtIP if I tcpdump traffic from the vserver I only see traffic out 1157054004 M * daniel_hozac both? i.e. host and the guest? 1157054045 M * daniel_hozac could you describe your network setup? 1157054143 M * rhodes The network is a standard class C network x.x.x.x/24. The host and guest IPs are x.x.x.55 and x.x.x.54. 1157054175 N * Bertl_oO Bertl 1157054182 M * Bertl evening! *short visit* 1157054182 M * rhodes The tcpdump from the host see outbound and inbound response traffic on the virtual IP, but it does not get to the guest 1157054208 M * Bertl rhodes: sounds like either your router does not route the guest ip, or some firewalling route blocks it 1157054209 M * daniel_hozac what are you tcpdump'ing? everything? 1157054224 M * Bertl rhodes: try the following thing (on the host) 1157054233 M * rhodes I tcpdump with a filter of the virtIP, the tests are simply ICMP pings 1157054237 M * Bertl ping -I www.google.com 1157054247 M * Bertl and then the same with -I 1157054268 M * Bertl if the second one fails (on the host) then it is _not_ guest setup related 1157054339 M * rhodes That solved it. Was this because the gateway/switch didn't have MAC for the VirtIP 1157054388 M * Bertl in general, note: Linux-VServer networking happens on the host, the guest is just a subset of the host (network wise) 1157054499 M * rhodes Meaning this virtual IP setup is no different that binding virtual IPs to a single interface (ie multiple hosted web sites on one system with different IPs) 1157054530 M * Bertl precisely, the only difference is the ip isolation (i.e. sockets are isolated between guests) 1157054558 M * Bertl means, an app inside a limited guest cannot bind any ip outside it's set 1157054563 M * Bertl *its 1157054601 M * rhodes On a more general Q, is there an expectation/effort that vserver will be integrated into the Linux kernel project? 1157054658 M * Bertl yes, and no, there is an efford getting virtualization into mainline (Linux-VServer style) but there are several parties out there, and a subset has to be found for integration 1157054714 M * Bertl daniel_hozac: hmm, what is the idea behind your patch? 1157054753 M * rhodes We're considering using vserver in a production deployment of a Java servlet hosting system. Using vserver to prevent a misbehaving app/client from interfering with other hosted applets. Does this sound risky? 1157054781 M * daniel_hozac i was thinking that maybe the (void) is what screws it up (gcc optimizes it away?), but it shouldn't really change anything, should it? 1157054851 M * daniel_hozac (well, vxis->vxi will still be set, but that shouldn't matter) 1157055114 J * shedi ~siggi@inferno.lhi.is 1157055279 M * Bertl rhodes: no, sounds like a typical application 1157055332 M * Bertl daniel_hozac: interesting idea ... for a moment I thought that the xchg() might actually take two pointers, and I screwed up there, but no such luck :) 1157055372 M * Bertl daniel_hozac: and xchg should also work for shorts (at least in theory) 1157055377 M * daniel_hozac and bytes. 1157055382 M * Bertl yup 1157055461 M * rhodes Thanks for the help... 1157055464 Q * rhodes Quit: Xirc - MacOSX 1157055474 Q * meandtheshell Quit: bye bye ... 1157055489 Q * bonbons Quit: Leaving 1157055719 J * gerrit ~gerrit@bi01p1.co.us.ibm.com 1157056065 M * matti Hi Bertl. 1157056613 Q * cehteh Remote host closed the connection 1157056798 Q * mire_ Quit: Leaving 1157057262 J * cehteh ~ct@cehteh.homeunix.org 1157057501 Q * cehteh Remote host closed the connection 1157057572 J * jtine ~blah@83.110.124.122 1157057581 M * jtine right place to ask about vpn issues? 1157057619 M * mnemoc right place to ask about Linux-VServer issues 1157057731 M * jtine oh 1157057733 P * jtine 1157057766 M * mnemoc :( 1157058015 M * daniel_hozac lol 1157058036 M * daniel_hozac you'd think the channel name and/or topic would've provided that hint... 1157058093 Q * gerrit Ping timeout: 480 seconds 1157058110 M * mnemoc :) 1157058284 J * cehteh ~ct@cehteh.homeunix.org 1157058290 M * cehteh mhm 1157058441 A * cehteh playing with the network, sorry for my frequent joins/parts 1157058503 Q * cehteh Remote host closed the connection 1157058874 J * cehteh ~ct@cehteh.homeunix.org 1157058938 A * mnemoc should start a campain promoting the usage of screen 1157058955 J * gerrit ~gerrit@bi01p1.co.us.ibm.com 1157058975 A * cehteh uses a bouncer ... but that doesnt really help if you deal with the networking of the machine where the bouncer rund 1157059020 M * cehteh anyways ... i am mostly finished now :) 1157059056 M * cehteh where can i set some longer timeout for killing processes from the vserver when shutting down? 1157059095 M * cehteh i always get a process-trace that my squid gets hard-killed because its not gracefully shut down in time 1157059109 M * daniel_hozac /etc/vservers/.defaults/apps/vshelper/sync-timeout, IIRC. 1157059141 M * cehteh ah ok 1157059167 M * cehteh bad nameing i would say .. /kill-after would be better 1157059179 M * daniel_hozac you can also increase it for just that guest by replacing .defaults with the guest's name. 1157059213 M * cehteh yes sure .. thats what i intend to do 1157059213 M * ntrs__ Bertl, daniel_hozac, over five hours and still works fine with unmodified devel + 2rb 1157059244 N * ntrs__ ntrs 1157059262 M * Bertl ntrs: good, now we just have to figure _why_ this fixes it :) 1157059272 M * ntrs hehe, ok. 1157059279 M * daniel_hozac indeed... do we print current->vx_info somewhere in /proc? 1157059301 M * Bertl the pointer itself? 1157059310 M * daniel_hozac yeah. 1157059319 M * daniel_hozac that would make my patch useless :) 1157059324 M * ntrs Bertl, keep in mind that I can't reboot this server much more. 1157059325 M * Bertl sec 1157059342 M * Bertl ntrs: I guess you can leave it at this state for now 1157059348 M * ntrs ok 1157059364 M * Bertl ntrs: it is a harmless modification, and we do not need to try unless we know what's going on 1157059486 M * Bertl daniel_hozac: nope, we had that at some point, but it seems I removed it :/ 1157059498 M * daniel_hozac ntrs: if you could just compile a original devel kernel the same way you did the others, and post the objdump output i requested earlier, i guess that might help. 1157059537 M * Bertl (but keep the current one too :) 1157059540 M * daniel_hozac (shouldn't take too long, just cp -al, patch -Rp1 2rb, make) 1157059569 J * ntrs_ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1157059569 Q * ntrs Read error: Connection reset by peer 1157059593 M * ntrs_ daniel_hozac, with no 2ra or 2rb? 1157059598 M * daniel_hozac exactly. 1157059735 M * ntrs_ I have vmlinux and system.map for the unmodified devel 1157059743 N * ntrs_ ntrs 1157059745 M * daniel_hozac ok then. 1157059806 J * ntrs_ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1157059806 Q * ntrs Read error: Connection reset by peer 1157059813 N * ntrs_ ntrs 1157059827 M * ntrs what do you need? 1157059836 J * Aiken ~james@tooax8-078.dialup.optusnet.com.au 1157059837 M * daniel_hozac objdump -d vmlinux | awk '/<__do_softirq>/ { do_print=1; } do_print == 1 && /^$/ { do_print=0; } do_print==1 { print; }' 1157059907 M * michal_ and in human language? 1157059919 M * daniel_hozac actually, objdump -d vmlinux | awk '/<__do_softirq>/ { do_print=1; } do_print == 1 && /^$/ { exit(0); } do_print==1 { print; }' would be better. 1157059934 M * daniel_hozac in human: the asm of __do_softirq :) 1157059962 M * ntrs daniel_hozac, http://paste.linux-vserver.org/327 1157059988 M * ntrs this was with do_print=1 1157060002 M * ntrs I don't understand this do_print==1 thing. 1157060015 M * ntrs Please give me the complete command. 1157060048 M * daniel_hozac it doesn't matter, i guess it's pretty easy to see where it ends. 1157060060 M * ntrs Ok 1157060226 M * michal_ daniel_hozac: just put it into script and saved for later, i think i'll have some good use of this example ,thx :] 1157060256 M * daniel_hozac hmm, just to make sure, could you post the result of grep do_softirq System.map? 1157060295 M * daniel_hozac michal_: hmm, i just realized it's flawed, it should be ... awk '/<__do_softirq>:/... 1157060313 M * michal_ fixing... 1157061019 M * daniel_hozac ntrs: well, it has the two xchg's, so my theory was incorrect... 1157061655 Q * shedi Quit: Leaving 1157061655 Q * ntrs Read error: Connection reset by peer 1157061656 J * ntrs_ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1157061895 J * Loki|muh loki@satanix.de 1157062527 J * blobaugh ~blobaugh@dialup-226-108.lacey.reachoneinternet.net 1157062721 P * blobaugh 1157062934 A * Belu_zZz is away (iŽll be back later...) 1157064220 Q * yarihm Quit: Leaving 1157064409 J * ntrs ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1157064409 Q * ntrs_ Read error: Connection reset by peer 1157064414 Q * dna Quit: Verlassend 1157064873 J * Pazzo ~thomas@host130-250-static.72-81-b.business.telecomitalia.it 1157065289 Q * Pazzo Quit: Ex-Chat 1157065423 J * ntrs_ ~ntrs@68-188-51-87.dhcp.stls.mo.charter.com 1157065424 Q * ntrs Read error: Connection reset by peer 1157065684 J * Doener_ ~nadine@p54877363.dip.t-dialin.net 1157065694 M * Doener_ evening! 1157066033 M * Bertl evening Doener_ ! 1157066107 M * Bertl Doener_: mind having a look at the most recent issue? we are looking for creative ideas there :) 1157066137 M * Doener_ sure. 1157066164 M * Doener_ my brain is still kind of roasted from the egyptian sun though ;) 1157066173 M * Bertl thing is, ntrs observed 'stange' mysql locking up issues 1157066199 M * Bertl basically all tables got locked and mysql stopped serving queries 1157066220 M * Bertl now, the first funny thing is, it is there with devel, but not with stable 1157066239 M * Doener_ MyISAM tables I assume, right? 1157066249 M * Bertl the second one is, the following patch seems to fix it in devel 1157066349 M * Bertl http://vserver.13thfloor.at/Stuff/delta-rc31.2rb.diff 1157066378 M * Bertl it only affects multi threaded (probably NPTL or TLS) mysql 1157066428 M * Bertl and it might somehow be related to sockets (unconfirmed) 1157066517 M * Doener_ hm, what was it that the admin mode allows? I'm not at my box and fetching a kernel tree might take a bit long here 1157066638 M * Bertl the relevant code basically 'switches' the context from 'current' to the host context, and back after execution 1157066727 M * Bertl it does this so by xchg()ing the vci of current with NULL, and saving the old one in the struct 1157066738 M * Bertl (same for xid) 1157066839 M * Doener_ hm, got some gut feeling. need to browse the source a bit, guess the dl will be finished in about 10 minutes, someone stopped hogging the line 1157066994 M * Bertl :) 1157067395 M * derjohn what is 2rb ? 1157067838 M * Bertl 31.2 was the release version, r means revert, and b is the second part (a,b) 1157067870 M * derjohn *kopfkratz* 1157067877 M * derjohn ah 1157067881 M * derjohn ein delta! 1157067898 M * derjohn what did you revert ? 1157067930 M * Doener_ the second part of the changes that came with version 31.2! :) 1157067937 M * Bertl it's just for a bug hunt, see IRC logs, ntrs-mysql issues 1157067951 M * derjohn Bertl, could you reprocuce them? 1157067963 M * derjohn *have to been able to 1157067972 M * Bertl yes, really? 1157067977 M * derjohn :) 1157068007 M * Bertl so you can do testing for us? 1157068041 M * derjohn oh my typo - sorry, I mean: Have you been able to reproduce what ntrs_ reported? 1157068074 M * Bertl nope, but he can reproduce it (with a customer guest :) quite easily and constantly 1157068097 M * Bertl i.e. it hangs after a few minutes, always the same symptoms 1157068102 M * derjohn and it only occurs under very heahy load ? 1157068108 M * derjohn *heavy ? 1157068123 M * Bertl and the delta 'fixes' it (higher load, but important seems to be the multi threading) 1157068207 M * derjohn and the bug was introduced with rc31 ? 1157068238 M * derjohn or is it longer there and was simply not detected ? 1157068286 M * Bertl the code is there since a long time 1157068293 M * matti Hi Bertl. 1157068305 M * Bertl nobody reported any issues yet, and I have several mysals running quite fine 1157068313 M * Bertl *mysqls 1157068315 M * derjohn and we expect it to make mysql threads (mysql4 ?) hand? 1157068318 M * derjohn *hang 1157068323 M * derjohn or other stuff too ? 1157068344 M * derjohn I run a mysql on a production server with devel -rc29 1157068369 M * matti What issue with MySQL? 1157068413 M * derjohn matti, with devel rc31 it seems to hang in guests. might have to do with (kernel)threads somehow 1157068429 M * matti derjohn: You mean 2.0.2? 1157068433 M * Bertl well, ntrs can probably describe that much better, but it seems that at some point, several (all?) mysql tables end up locked, and no queries are served 1157068446 M * derjohn matti, devel = 2.1.1 currently 1157068460 M * Bertl unix sockets start timing out, clients 'hang' around 1157068465 M * matti derjohn: Oh, well. With 2.0.2 everything work like a charm. 1157068469 M * matti :) 1157068479 M * Bertl indeed:) 1157068568 M * derjohn so we dont have some kind of test data / "evil query" to try on differnt guests ? 1157068581 M * Bertl nope 1157068584 M * matti Bertl: I am very happy with latest 2.0.2. Is really stable. Even with grsecurity (few chroot restrictions also turned on) and pax (segmexec mostly, sanitize free memory, ALSR, ...). Just perfect! 1157068608 M * Bertl good to hear! 1157068634 M * matti k, sorry for disturbing both of you :) 1157068665 M * derjohn matti, you're always welcome ! 1157068671 M * matti derjohn: Thanks :)