1215130715 Q * dowdle Remote host closed the connection 1215131132 J * doener ~doener@i577BB6D2.versanet.de 1215131237 Q * doener_ Ping timeout: 480 seconds 1215133588 J * daemon nobody@laptop2.daemoncore.org 1215133596 N * daemon Diluvium 1215133599 M * Diluvium Good day :) 1215133612 M * Diluvium im following the openvpn guide at http://linux-vserver.org/Frequently_Asked_Questions 1215133618 M * Diluvium and am having a few problems 1215133638 M * Diluvium namely this section: http://laptop2.daemoncore.org/~paul/Screenshot-176.png 1215133646 M * Diluvium I used /dev/MAKEDEV 1215133650 M * Diluvium instead of ./MAKEDEV 1215133658 M * Diluvium but the problem comes with the next part: 1215133669 M * Diluvium (the guest now has /dev/net/tun [note]) 1215133676 M * Diluvium # head /etc/vservers//interfaces/1/* 1215133687 M * Diluvium I of course substituted the right server for 1215133697 M * Diluvium a quick ls of that directory reveals: 1215133721 M * Diluvium vortex:/var/lib/vservers/vortex-26/dev# ls -la /etc/vservers/vortex-26/interfaces/ 1215133729 M * Diluvium . and .. as usual and only one other entry: 1215133732 M * Diluvium drwxr-xr-x 2 root root 4096 2008-06-02 02:36 0 1215133740 M * Diluvium What is going wrong? 1215133832 M * Diluvium the server version is: Linux vortex 2.6.22-3-vserver-686 #1 SMP Sun Feb 10 23:11:25 UTC 2008 i686 GNU/Linux 1215134487 M * Diluvium any help at all would be really great :o 1215134634 M * daniel_hozac i don't get it. what are you doing, and what's the problem? 1215134713 M * Diluvium daniel_hozac, I want one of my vserver to run openvpn host 1215134719 M * Diluvium daniel_hozac, of course that requires TUN access 1215134727 M * Diluvium so Ifollowed the guide on http://linux-vserver.org/Frequently_Asked_Questions 1215134734 M * Diluvium and it all seems good till this part: 1215134748 M * Diluvium # head /etc/vservers//interfaces/1/* 1215134758 M * Diluvium I quickly checked if it exists or not 1215134766 M * Diluvium and found in /etc/vservers/vortex-26/interfaces/ 1215134772 M * Diluvium . and .. as usual and only one other entry: 1215134775 M * Diluvium drwxr-xr-x 2 root root 4096 2008-06-02 02:36 0 1215134812 M * Diluvium of course '0' is the NIC on the guest 1215134831 M * Diluvium so im not sure where I have gone wrong 1215134860 M * daniel_hozac did you create the 1 directory with the necessary settings? 1215134899 M * daniel_hozac also note that most of that isn't necessary with recent util-vservers. 1215134903 M * Diluvium daniel_hozac, I have not been informed to do that by the howto 1215134920 M * Diluvium http://laptop2.daemoncore.org/~paul/Screenshot-176.png 1215134921 M * daniel_hozac 0.30.215 knows how to setup tun/tap interfaces. 1215134929 M * Diluvium daniel_hozac, ah ok 1215134933 M * Diluvium so how should I proceed 1215134944 M * daniel_hozac create the 1 directory, add the settings you need. 1215134950 M * Diluvium a simple mkdir 1 1215134963 M * daniel_hozac and subsequent echo's, yes. 1215135000 M * Diluvium daniel_hozac, Ah thank you very much :) 1215135006 M * Diluvium I shall have to update that on the wiki 1215135595 M * Diluvium daniel_hozac, hmm I have set it up and restarted the jail but there is no tu ndevice showing up in /dev or ifconfig still 1215135606 M * Diluvium whoops 1215135610 M * Diluvium my fault sorry ;) forgot something 1215135682 M * Diluvium daniel_hozac, ah no i didnt 1215135700 M * Diluvium still not showing up 1215135703 M * Diluvium I added the following 1215135708 M * Diluvium /etc/vservers/vortex-26/interfaces/1 1215135725 M * Diluvium ip (in file: 10.0.0.1) 1215135731 M * Diluvium nodev (infile tun17) 1215135743 M * Diluvium prefix (infile: 24) 1215135748 M * Diluvium and restarted the serer 1215135755 M * Diluvium tun17 is not in /dev or ifconfig 1215135848 M * daniel_hozac well, if you want the utils to set it up for you, you want to remove nodev, set dev to the interface, and touch tun. 1215135885 M * Diluvium vortex:/etc/vservers/vortex-26/interfaces/1# rm nodev; touch tun 1215135887 M * Diluvium as such 1215135904 M * daniel_hozac and setting dev. 1215135912 M * Diluvium vortex:/etc/vservers/vortex-26/interfaces/1# cat dev 1215135914 M * Diluvium tun17 1215135917 M * Diluvium sorry I didnt paste that bit ;) 1215135928 M * daniel_hozac and does that work? 1215135929 M * Diluvium ok 1215135932 M * Diluvium lets give it a wqhirl 1215135943 M * Diluvium just stopping starting vserver 1215135976 M * Diluvium daniel_hozac, \o/ i have tun17 thank you very much 1215136081 M * Diluvium daniel_hozac, A huge thank you my friend you just saved me a big headache 1215136120 M * daniel_hozac you're welcome 1215138626 J * z0d ~z0d@fw.wonderline.hu 1215138630 M * z0d hello 1215138677 M * daniel_hozac hi 1215139002 M * z0d what is vserver-utils? 1215141333 J * EsnP[gc^G ~hollow@proteus.croup.de 1215141334 Q * Hollow Read error: Connection reset by peer 1215141349 J * bzed_ ~bzed@devel.recluse.de 1215141392 N * EsnP[gc^G Hollow 1215141716 Q * bzed Ping timeout: 480 seconds 1215141716 N * bzed_ bzed 1215143046 Q * Diluvium Quit: Leaving 1215147377 M * Bertl_oO finally off to bed ... have a good one everyone! 1215147386 N * Bertl_oO Bertl_zZ 1215147692 J * meandtheshell1 ~sa@88-117-20-200.adsl.highway.telekom.at 1215147708 Q * nenolod Remote host closed the connection 1215147755 J * nenolod ~nenolod@ip70-189-74-62.ok.ok.cox.net 1215149095 Q * balbir Ping timeout: 480 seconds 1215149482 J * sharkjaw ~gab@149-67-194.231210.adsl.tele2.no 1215151949 J * balbir ~balbir@59.145.136.1 1215152014 Q * balbir 1215152029 J * balbir ~balbir@59.145.136.1 1215153130 Q * larsivi Quit: Konversation terminated! 1215156306 J * larsivi ~larsivi@85.221.53.194 1215156686 Q * z0d 1215156733 J * z0d ~z0d@fw.wonderline.hu 1215158210 Q * balbir Remote host closed the connection 1215158462 J * z0d- ~z0d@fw.wonderline.hu 1215158477 N * z0d Guest3509 1215158477 N * z0d- z0d 1215158766 Q * Guest3509 Ping timeout: 480 seconds 1215159197 J * joern42 ~jr@dyndsl-091-096-062-088.ewe-ip-backbone.de 1215159491 Q * joern421 Ping timeout: 480 seconds 1215159560 J * derjohn_foo ~aj@92.117.130.201 1215160637 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d 1215160945 N * DoberMann[ZZZzzz] DoberMann 1215161201 J * z0d- ~z0d@fw.wonderline.hu 1215161208 N * z0d Guest3513 1215161208 N * z0d- z0d 1215161481 J * esa bip@ip-87-238-2-45.static.adsl.cheapnet.it 1215161491 Q * esa` Ping timeout: 480 seconds 1215161507 Q * Guest3513 Ping timeout: 480 seconds 1215161935 J * Pazzo ~ugelt@reserved-225136.rol.raiffeisen.net 1215162194 M * pmjdebruijn I want to limit a vservers rss 1215162216 M * pmjdebruijn however when I do "sudo vlimit -c 1234 -a -d" 1215162223 M * pmjdebruijn I get a whole lot of N/A and inf 1215162236 M * pmjdebruijn do I need a to enable a specific kernel option? 1215162246 M * marv_ no thats normal output 1215162287 M * marv_ its a bit fuzzy in my head 1215162298 M * marv_ but i get N/A inf on limits that i dont have set 1215162303 M * marv_ the limits i have set are stated 1215162331 M * pmjdebruijn ok, we I remember in the past, I used to see the current usage instead of that 1215162346 M * marv_ thats in /proc/virutal 1215162348 M * marv_ no? 1215162401 M * marv_ i'm too lazy to login to my vserver host to check it out lol :P 1215162476 M * marv_ the 3 roes are min/soft/hard limits i beleive 1215162480 M * marv_ rows 1215162515 M * marv_ cat /proc/virtual//limit 1215162524 M * marv_ u have current usage there 1215162538 M * marv_ vserver-stat will also display some current usage 1215162604 M * pmjdebruijn oh crap, indeed 1215162608 M * pmjdebruijn doh! 1215162693 M * marv_ your welcome :P 1215162806 A * pmjdebruijn suffers from senility at a young age... 1215162845 M * marv_ we all have our issues. 1215164364 J * _Hunger Hunger.hu@Hunger.hu 1215164396 Q * Hunger Read error: Connection reset by peer 1215164625 J * yarihm ~yarihm@whitehead2.nine.ch 1215165281 Q * meandtheshell1 Ping timeout: 480 seconds 1215165648 J * esa` bip@ip-87-238-2-45.static.adsl.cheapnet.it 1215165656 Q * esa Ping timeout: 480 seconds 1215165867 Q * derjohn_foo Ping timeout: 480 seconds 1215166400 J * xdr ~xdr@164-173-96-87.cust.blixtvik.se 1215167169 J * dna ~dna@250-216-dsl.kielnet.net 1215168116 Q * marv_ Quit: Leaving 1215168615 Q * SpComb Ping timeout: 480 seconds 1215169372 N * Bertl_zZ Bertl 1215169376 M * Bertl morning folks! 1215169465 J * SpComb terom@zapotek.paivola.fi 1215169549 J * the-herby ~herby@p5496F709.dip.t-dialin.net 1215169549 M * z0d hi Bertl 1215170065 J * loddafnir ~mike@193.170.138.233 1215171582 Q * phedny Remote host closed the connection 1215171615 J * _kwowt ~quote@pomoc.ircnet.com 1215171619 M * _kwowt damn 1215171621 M * _kwowt i got a huge problem 1215171633 M * _kwowt my network just stops working 1215171646 M * _kwowt all the ips on the machine, i cant login/ssh 1215171646 M * joern42 is the interface still there? 1215171649 M * _kwowt w8 1215171657 M * _kwowt yes 1215171672 M * _kwowt i'm logged in, but i cant reconnect 1215171676 M * _kwowt if i logout, i cant get back 1215171696 M * _kwowt if i restart the machine, or sometimes restart some guest servers, the network starts working again 1215171784 M * joern42 hmm, i had some problems when restarting vservers, that some interfaces were gone.. was net.ipv4.conf.all.promote_secondaries=1 in /etc/sysctl.conf 1215171838 M * joern42 which is described here: http://oldwiki.linux-vserver.org/some_hints_from_john 1215171894 M * Bertl not really a Linux-VServer issue, but folks tend to encounter it more often with guest restarts 1215171927 Q * kir Quit: Leaving. 1215171971 M * _kwowt no, i got that set 1215171974 M * Bertl _kwowt: double check that the guests are using secondary ips, or that, as joern42 suggested, promotion is activated 1215171990 M * _kwowt promotion IS activated 1215172011 M * daniel_hozac sounds more like you're using connection tracking and running in to the connection limit. 1215172012 M * Bertl okay, stops working means, you cannot connect from inside the machine (to the outside) either? 1215172015 M * daniel_hozac or similar. 1215172060 M * _kwowt Bertl: i cant connect outside either 1215172070 M * _kwowt daniel_hozac: connection tracking? whats that 1215172098 M * _kwowt ssh: box3.si-shell.net: Temporary failure in name resolution 1215172103 M * _kwowt but i got nameservers added right 1215172118 M * daniel_hozac iptables. 1215172137 M * _kwowt i dont use iptables 1215172150 M * _kwowt iptables v1.3.5: can't initialize iptables table `filter': Table does not exist (do you need to insmod?) 1215172150 M * _kwowt Perhaps iptables or your kernel needs to be upgraded. 1215172152 M * Bertl _kwowt: did you try with 'tracepath' or similar in such a case? 1215172172 M * joern42 do you got the same effect if you use the ip address? 1215172206 M * _kwowt sh: connect to host 91.185.193.178 port 22: No route to host 1215172209 M * joern42 vs01:~# ssh box3.si-shell.net 1215172209 M * joern42 The authenticity of host 'box3.si-shell.net (91.185.193.178)' can't be established. 1215172209 M * joern42 RSA key fingerprint is 94:84:0c:24:c4:6f:a4:32:6d:4a:d1:d4:1b:8f:14:e4. 1215172209 M * joern42 Are you sure you want to continue connecting (yes/no)? 1215172218 M * _kwowt tracepath: 1215172218 M * _kwowt 1: box2 (91.185.193.146) 0.072ms pmtu 1500 1215172219 M * _kwowt 1: no reply 1215172219 M * _kwowt 2: send failed 1215172219 M * _kwowt Resume: pmtu 1500 1215172234 M * joern42 i think you got a network problem.. 1215172248 M * _kwowt ... like? 1215172261 M * _kwowt the broken machine is 91.185.193.146 1215172299 J * kir ~kir@swsoft-msk-nat.sw.ru 1215172315 M * joern42 ah ok.. cant ping it and cant ssh .. 1215172865 P * joern42 1215172872 J * joern42 ~jr@dyndsl-091-096-062-088.ewe-ip-backbone.de 1215173060 J * phedny ~mark@2a02:348:35:5a26::1 1215173479 M * z0d do you take the effort to split the guests into separate partitions for / /tmp /usr /var /var/log etc.? 1215173497 M * Bertl what for? 1215173581 M * z0d e.g. you can make /usr read-only, nodev etc. 1215173604 M * Bertl you can do that without separate partitions too (on Linux-VServer) 1215173621 M * Bertl with --bind mounts 1215173850 M * z0d so users won't see the 'real' /usr? 1215173866 M * _kwowt so.. 1215173867 M * _kwowt Bertl 1215173869 M * _kwowt any ideas? 1215174001 M * Bertl _kwowt: what does 'tracepath 64.233.167.147' give _on_ that machine? 1215174051 M * Bertl (please use paste.linux-vserver.org for everything longer than 3 lines) 1215174112 M * _kwowt 1: box2 (91.185.193.146) 0.071ms pmtu 1500 1215174112 M * _kwowt 1: send failed 1215174112 M * _kwowt Resume: pmtu 1500 1215174155 M * Bertl and the network is up? i.e. 'ip link ls' shows it as up and such? 1215174174 M * Bertl box2 is the machine itself, yes? 1215174176 M * _kwowt yep 1215174178 M * _kwowt yes 1215174179 M * _kwowt 1: eth0: mtu 1500 qdisc pfifo_fast qlen 1000 1215174179 M * _kwowt link/ether 00:1e:0b:4e:03:d4 brd ff:ff:ff:ff:ff:ff 1215174182 M * _kwowt its up :) 1215174204 M * Bertl what kernel is that? and what patches? 1215174217 M * _kwowt Linux box2 2.6.20-vs2.2.0-gentoo #1 SMP Wed Nov 28 23:06:07 GMT 2007 i686 Intel(R) Xeon(R) CPU 3060 @ 2.40GHz GenuineIntel GNU/Linux 1215174225 M * _kwowt but this started like 1 week ago 1215174227 M * _kwowt or less 1215174231 M * _kwowt before it worked fine.. 1215174273 M * z0d Bertl: so the general approach with vserver is to have bind mounts and vroot quota instead of physical partitions? 1215174291 M * Bertl _kwowt: what about a ping to 64.233.167.147 ? 1215174305 M * _kwowt PING 64.233.167.147 (64.233.167.147) 56(84) bytes of data. 1215174312 M * _kwowt nothing.. 1215174323 M * _kwowt ping: sendmsg: No buffer space available 1215174328 M * Bertl z0d: physical partitions are fine, but usually no need for more than one of them (except for /tmp which is a tmpfs by default) 1215174354 M * z0d Bertl: you mean no more than one / vserver? 1215174373 M * Bertl one partition per guest, if you like to keep them separated 1215174389 M * _kwowt ... 1215174536 M * _kwowt the machine is very slow 1215174545 M * Bertl memory problems? trashing? 1215174552 M * _kwowt trashing ? 1215174567 M * _kwowt Mem: 3038 2837 201 0 18 1901 1215174567 M * _kwowt -/+ buffers/cache: 916 2121 1215174567 M * _kwowt Swap: 980 2 977 1215174568 M * Bertl constant page in/out (swap activity) 1215174593 M * _kwowt seems like enough memory, doesnt it? 1215174604 M * Bertl depends ... 1215174619 M * _kwowt what am i to do?:S 1215174644 M * Bertl let's upload /proc/meminfo and /proc/zoneinfo 1215174668 M * Bertl also, do you get any messages in 'Dmesg'? 1215174675 M * Bertl *'dmesg' 1215174688 M * _kwowt TCP: Treason uncloaked! Peer 92.37.45.82:52212/45329 shrinks window 2323978359:2323981263. Repaired. 1215174694 M * _kwowt things like that 1215174695 M * _kwowt ALOT 1215174764 M * Bertl well, seems like you are under attack then :) 1215174776 M * _kwowt under attack? 1215174779 M * _kwowt like ddos? 1215174795 M * Bertl well, or like p2p traffic banging on your machine 1215174808 M * _kwowt http://paste.linux-vserver.org/12297 1215174819 M * _kwowt i run torrentflux (seedboxes) 1215174822 M * joern42 iptraf is a nice tool for monitoring your bandwidth.. 1215174827 M * _kwowt but i've been running it for liek 6 months 1215174831 M * _kwowt and i never had this problem 1215174838 M * _kwowt now every 1-2days this happens 1215174866 M * _kwowt http://paste.linux-vserver.org/12298 1215174883 M * Bertl well, I would suggest to upgrade the kernel, maybe it is really a DDoS, and newer kernels already have a fix 1215174939 M * Bertl also get rid of the highmem (in that process) 1215175003 M * _kwowt highmem? 1215175043 M * Bertl you have 3GB of memory, are obviously using a 32bit kernel, and about 2GB are configured as highmem 1215175083 M * Bertl it would be better to configure a 1/3 split and use all as 'normal' memory 1215175137 M * _kwowt so i need to change that in kernel config? 1215175177 M * _kwowt lol 1215175181 M * _kwowt now i stopped one vserver 1215175185 M * _kwowt and its all working again 1215175188 M * _kwowt just like that... 1215175199 M * _kwowt i've stopped 2 guests,nothing 1215175202 M * _kwowt then apache, nothing 1215175209 M * _kwowt now one more, and it works 1215175386 M * _kwowt :S 1215175388 M * _kwowt what tha hell 1215175553 M * _kwowt now its like nothing happened 1215175557 M * _kwowt all works quick again 1215175643 M * Bertl well, investigate _what_ that guest was doing 1215175700 J * lauanana ~lauanana@ANantes-257-1-64-75.w90-25.abo.wanadoo.fr 1215175711 M * _kwowt nothing running there 1215175713 M * _kwowt nothing in logs 1215175716 M * _kwowt just a few bouncers runned 1215175727 M * _kwowt i use that guest myself, for IRC 1215175746 M * _kwowt but yesterday when this happened, i turned off a different guest and it started working 1215175785 M * _kwowt so it must be something else, not some runnin process 1215175802 Q * lauanana 1215175916 M * Bertl well, try with a new kernel, could be some network card issues too 1215175959 M * _kwowt network card? 1215175982 M * Bertl the 'no buffer space' is a sign of contention 1215176013 M * Bertl i.e. the kernel network stack tries to send a packet, but there simply is no space left in the buffer to actually get it out 1215176078 M * Bertl that might be caused by memory pressure on dma zones (didn't look like it from your zoneinfo) or extremely high throughput (p2p traffic/DDoS) or by 'slow' physical transfer (bad driver, bad nic, bad irq) 1215176126 M * Bertl but it definitely doesn't look like being Linux-VServer related :) 1215176233 M * _kwowt it could be p2p traffic 1215176238 M * _kwowt but why did it start now? 1215176246 M * _kwowt i've been running on the same speeds for 5months 1215176309 M * Bertl no idea, maybe you have been slightly below the maximum and now you are slightly above ... 1215176340 M * _kwowt would extra memory (RAM) help? 1215176344 M * _kwowt or that doesnt matter? 1215176367 M * Bertl doesn't look like it would in your case, but a new nic driver or better nic could help 1215176373 M * joern42 i would say no, perhaps a better network card.. 1215176399 M * _kwowt so...kernel update wouldnt either? 1215176411 Q * dna Read error: Connection reset by peer 1215176413 M * joern42 maybe 1215176418 M * Bertl it might (probably includes the new nic driver :) 1215176430 M * _kwowt well, the guy whos takin care of my server said its a HP nic 1215176431 M * _kwowt 1gbit 1215176441 M * joern42 try ethtool -i eth0 1215176537 M * _kwowt adriver: tg3 1215176537 M * _kwowt version: 3.72 1215176537 M * _kwowt firmware-version: 5714-v3.24 1215176537 M * _kwowt bus-info: 0000:03:04.0 1215176571 M * joern42 im using 2.6.22.19 , i have version 3.77 1215176595 M * _kwowt which means, kernel update would change the version? 1215176602 M * joern42 maybe.. :) 1215176637 M * _kwowt damn 1215176713 M * _kwowt gotta update the kernel then 1215176720 M * _kwowt but i dont think i still know how to :) 1215176724 M * _kwowt havent done that ina while.. 1215176777 Q * kir Remote host closed the connection 1215176783 M * joern42 just do a " zcat /proc/config.gz > /usr/src/linux/.config" and then make oldconfig 1215176801 M * joern42 (if your kernel was compiled with /proc/config.gz support...) 1215176922 Q * bronson Quit: Ex-Chat 1215177151 M * _kwowt thats it? 1215177274 Q * sharkjaw Remote host closed the connection 1215177487 M * _kwowt joern42? 1215177502 M * _kwowt thats all i need to do to update kernel?:d 1215177721 M * joern42 ups 1215177762 M * joern42 hmm, at first you copy your current config, then you get the kernel, and then you read the README, then compile.. :) 1215177800 M * _kwowt thought so:P 1215177815 M * _kwowt so i can just use the old config 1215177819 M * _kwowt for the new kernel 1215177824 M * joern42 yes should work 1215177825 M * _kwowt using zcat and make oldconfig 1215177827 M * _kwowt great :) 1215177884 M * joern42 perhaps you want to take a look at the nic settings.. 1215178198 J * meandtheshell ~sa@88-117-22-103.adsl.highway.telekom.at 1215178764 J * dna ~dna@153-231-dsl.kielnet.net 1215179667 Q * larsivi Quit: Konversation terminated! 1215179886 J * docelic ~docelic@78.134.195.216 1215180434 Q * Aiken Remote host closed the connection 1215181243 J * dna_ ~dna@33-232-dsl.kielnet.net 1215181348 J * pisco ~pisco@tor.noreply.org 1215181640 Q * dna Ping timeout: 480 seconds 1215182314 Q * z0d Remote host closed the connection 1215182922 J * dna__ ~dna@67-201-dsl.kielnet.net 1215183249 P * joern42 1215183290 M * yarihm Bertl, around? 1215183294 M * Bertl yep 1215183318 M * yarihm we got stuck last time in the discussion about work W being done in time T 1215183321 M * yarihm :) 1215183325 Q * dna_ Ping timeout: 480 seconds 1215183346 M * Bertl did we?, last thing I remeber was that you needed to think about my arguments :) 1215183439 M * yarihm I more felt that we were stuck in the debate :) but my memories might play on me. However, you suggested trading CPU time for physical memory, right? 1215183480 M * Bertl kind of 1215183533 M * yarihm ok. Well, as I probably said before already, I do trust your judgement, so if you say this is how we distribute ressources fairly among guests in case of ressource starvation, that is what we are going to do 1215183565 M * Bertl okay, the main question is, do we have a setup to test for such cases? 1215183594 Q * bonbons Quit: Leaving 1215183619 M * yarihm well, we do as in 'we have servers to test this on' but not as in 'we do have a synthetic benchmark that provides for the clean conditions needed to test this' from what I can tell 1215183625 M * Bertl i.e. do we have something where we can a) show that the fair distribution is not given, and b) that the issue is caused by memory pressure/swapping? 1215183660 M * Bertl doesn't need to be a synthetic benchmark, it should suffice if we have some kind of indication/test 1215183792 M * yarihm well, I guess I can produce something of that kind. I just create a server with two context's on it, in one I run a mysql-DB that allocates a lot of ram in one and keep a DB-Benchmark running on it, on the other one i run a normal mysql db and check how much of its memory is swapped out ... 1215183796 M * yarihm ok, that's kind of unfair 1215183817 M * yarihm hmm ... how to reproduce this properly 1215183853 M * Bertl I think we need to find something where we can calculate the 'work' done 1215183883 M * Bertl something like compiling the kernel or so 1215183885 M * yarihm well yes, but that'd be something like queries per second or pages served per second 1215183888 M * yarihm ok, that's fine, too 1215183910 M * Bertl so, let's take the following setup for a start 1215183933 M * Bertl two contexts, one is (somehow) hogging some memory 1215183971 M * Bertl first test: 1215183989 M * Bertl context A (no memory hog) alone, kernel compile 1215184008 M * Bertl second: context B (with memory hog) alone, kernel compile 1215184026 M * Bertl the difference should yield the amount of work consumed by the memory hog 1215184045 M * Bertl third test: both contexts running side by side 1215184065 M * Bertl if there is some kind of unfairness, we should be able to see it, no? 1215184091 M * yarihm hmm ... what would you expect to see that indicates the unfairness? 1215184104 M * Bertl a drastically longer kernel compile? 1215184117 M * yarihm for context A you mean? 1215184130 M * Bertl for the one which is treated unfair :) 1215184153 M * Bertl *unfairly 1215184210 M * Bertl feel free to suggest some other test ... this was just a suggestion 1215184263 M * yarihm what about this: we have one context repeatedly running 1 kernel compile with a and the other one 3. By the concept of fair ressource-allocation (in our case the combination of memory, cpu and disk ressources) the one with 3 compile-runs should tike 3 times as long as the one with 1 compile run 1215184282 M * yarihm s/with a// 1215184299 M * yarihm well, fair ressource-allocation over all I mean, that is work being done 1215184327 M * yarihm and with 'taking 3 times as long' i mean '3 times as long in average per compile run' 1215184365 M * Bertl I'm fine with that, but you need to evaluate the 'additional' overhead you get by doing a single run vs. 3 compiles in parallel (without any other context running) too 1215184388 M * yarihm ok, that would need to be taken into account as well 1215184394 M * Bertl i.e. we can strive to do better than a real machine, but we cannot expect it per se :) 1215184427 M * yarihm ok 1215184446 M * Bertl sounds like a plan then 1215184469 M * yarihm good, so we have to get a machine and then: 1215184486 M * yarihm 1. run extract/compile/remove time for one single run 1215184500 M * yarihm 2. get extract/compile/remove time for 3 runs in parallel 1215184546 M * yarihm 3. get extract/compile/remove time for the above setup (1 context with 3 runs in parallel, 1 context with a single run) 1215184583 M * Bertl yep, with the scheduler set to hard/equal rates 1215184592 M * yarihm now is there something special i should set (e.g. memlimits) to have a useful setup 1215184607 M * yarihm ok, scheduler set to hard/equal 1215184619 M * yarihm what exactly do you mean by this? :) 1215184619 M * Bertl you probably also want to keep the context with 1 compile repeating the test over and over 1215184638 M * yarihm yes, of course, not that ctx 1 stops and then the other 3 run alone 1215184656 M * yarihm the 3 runs in the other ctx I mean 1215184671 M * Bertl well, I guess you know how to set the hard cpu scheduler 1215184687 M * Bertl in this test case a simple 5/10 setup for both should be fine 1215184705 M * Bertl (you can test that with cpuhog in advance (the 50/50 distribution) 1215184813 M * Bertl memory limits are not really an issue here, but you might want to run this kind of test with and without some memory pressure, otherwise you won't detect any paging related effects 1215184846 M * Bertl (and the memory/paging related inbalance is what we are looking for) 1215184879 M * Bertl *imbalance (my english sucks today :) 1215185060 J * kir ~kir@swsoft-msk-nat.sw.ru 1215185062 Q * kir Remote host closed the connection 1215185067 J * kir ~kir@swsoft-msk-nat.sw.ru 1215185146 M * yarihm ok, i got a dual xeon with 2 gig ram for testing. are the compile-runs memory-pressing enough for 2GB or should i boot the machine with mem=512 to simulate a more 'realistic' MEM/VServer ratio? 1215185979 N * DoberMann DoberMann[PullA] 1215186500 M * Bertl yarihm: double check with vmstat (look for paging) 1215186517 M * Bertl run the tests with and without swapping, to get a feeling for the difference 1215186521 M * Bertl +s 1215186631 J * dowdle ~dowdle@67-42-169-29.blng.qwest.net 1215187932 J * z0d ~z0d@apn-89-223-129-66.vodafone.hu 1215187948 M * z0d hello 1215188077 M * Bertl nap attack ... bbl 1215188083 M * yarihm cu 1215188090 N * Bertl Bertl_zZ 1215188693 J * larsivi ~larsivi@169.80-202-217.nextgentel.com 1215188963 J * fatgoose ~samuel@82.80.modemcable.oricom.ca 1215190279 Q * Pazzo Quit: Ex-Chat 1215191611 Q * pisco Ping timeout: 480 seconds 1215191626 J * lRem lrem@ip-89.171.49.146.crowley.pl 1215191637 M * lRem hi 1215191658 M * lRem i'm having a problem with vservers on debian stable 1215191686 M * lRem fragzone:~# vserver host1 start 1215191686 M * lRem vsysctl: open("."): Permission denied 1215191717 M * lRem it happens since I did chxid -c host1 -R /var/lib/vservers/host1/ 1215191749 M * lRem changing xid back to 0 makes things work all right 1215191753 M * lRem any ideas? 1215192993 Q * kikoo__ Quit: Lost terminal 1215194256 Q * meandtheshell Ping timeout: 480 seconds 1215194852 N * DoberMann[PullA] DoberMann 1215196368 J * lilalinux ~plasma@dslb-084-058-226-172.pools.arcor-ip.net 1215196637 J * meandtheshell ~sa@88-117-15-72.adsl.highway.telekom.at 1215196801 P * meandtheshell 1215196912 J * meandtheshell1 ~sa@88-117-15-72.adsl.highway.telekom.at 1215197219 Q * yarihm Quit: Leaving 1215197425 Q * z0d Remote host closed the connection 1215199583 J * marv_ ~marv@modemcable139.74-82-70.mc.videotron.ca 1215199780 Q * fatgoose Remote host closed the connection 1215199813 J * fatgoose ~samuel@82.80.modemcable.oricom.ca 1215200092 Q * kir Remote host closed the connection 1215200541 M * lRem found that: missing /etc/vservers/host1/context file 1215201956 J * Aiken ~james@ppp118-208-119-17.lns4.bne4.internode.on.net 1215202367 J * joern421 ~jr@dyndsl-080-228-183-091.ewe-ip-backbone.de 1215202449 J * fatgoose_ ~samuel@82.80.modemcable.oricom.ca 1215202450 Q * fatgoose Read error: Connection reset by peer 1215202552 Q * meandtheshell1 Quit: Leaving. 1215202833 J * ntrs ~ntrs@77.29.76.232 1215203340 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d 1215204423 J * derjohn_mob ~aj@p5B23D574.dip.t-dialin.net 1215205118 Q * loddafnir Remote host closed the connection 1215207049 Q * lilalinux Remote host closed the connection 1215207500 N * DoberMann DoberMann[ZZZzzz] 1215207700 Q * duckx Remote host closed the connection 1215207920 Q * ntrs Ping timeout: 480 seconds 1215208060 Q * bonbons Quit: Leaving 1215208097 Q * derjohn_mob Ping timeout: 480 seconds 1215208348 J * britneypire ~britneypi@ANantes-257-1-64-75.w90-25.abo.wanadoo.fr 1215208359 Q * docelic Quit: http://www.spinlocksolutions.com/ 1215208472 J * docelic ~docelic@78.134.195.216 1215208479 Q * britneypire 1215208946 J * dna ~dna@36-241-dsl.kielnet.net 1215209328 Q * dna__ Ping timeout: 480 seconds 1215209424 P * lRem 1215209605 J * dniel ~ary@host67.190-138-103.telecom.net.ar 1215209839 M * dniel Hi daniel_hozac! 1215210083 M * dniel I've an idea. I want to install three virtual server in the same host. I'll install one and the others I'll mount with AuFS the filesystem of this. Can you understand me? 1215210132 Q * Aiken Quit: Leaving 1215210478 Q * dniel Quit: Leaving 1215210907 Q * docelic Quit: http://www.spinlocksolutions.com/ 1215211623 Q * the-herby Quit: Ex-Chat 1215212400 J * the-fafa ~fafa@p5496EDA0.dip.t-dialin.net 1215212974 J * geb ~geb@41.207.197.36 1215213047 M * geb hi 1215213461 J * hparker ~hparker@linux.homershut.net