1297988053 Q * Piet Ping timeout: 480 seconds 1297988077 Q * bonbons Quit: Leaving 1297988583 J * Piet ~Piet__@82VAAB0GN.tor-irc.dnsbl.oftc.net 1297989321 J * petzsch1 ~markus@dslb-092-078-224-117.pools.arcor-ip.net 1297989412 Q * petzsch Ping timeout: 480 seconds 1297989725 Q * bsingh Ping timeout: 480 seconds 1297990037 Q * imcsk8 Quit: Leaving 1297990264 J * bsingh ~balbir@122.172.29.26 1297991216 Q * petzsch1 Quit: Leaving. 1297991227 J * speedy ~speedy@rsm2.koutbo6.com 1297991330 Q * dowdle Remote host closed the connection 1297993412 Q * speedy Quit: Ex-Chat 1297995384 Q * hparker Quit: Quit 1297996133 M * Bertl off to bed now ... have a good one everyone! 1297996138 N * Bertl Bertl_zZ 1297996455 J * ktwilight_ ~keliew@91.176.133.120 1297996743 Q * ktwilight__ Ping timeout: 480 seconds 1298000003 Q * manana Ping timeout: 480 seconds 1298005497 Q * bsingh Read error: Operation timed out 1298006990 J * bsingh ~balbir@122.172.29.26 1298007178 Q * Piet Ping timeout: 480 seconds 1298007192 J * Piet ~Piet__@82VAAB0IQ.tor-irc.dnsbl.oftc.net 1298007518 Q * bsingh Read error: Operation timed out 1298007619 J * nkukard ~nkukard@41-133-112-153.dsl.mweb.co.za 1298008132 J * bsingh ~balbir@122.172.8.206 1298013005 J * ghislain ~AQUEOS@adsl2.aqueos.com 1298016884 J * thierryp ~thierry@zankai.inria.fr 1298018618 J * kirodar ~kirodar@p4FCDFD9E.dip.t-dialin.net 1298018688 P * kirodar 1298020106 J * BenG ~bengreen@cpc12-aztw24-2-0-cust146.aztw.cable.virginmedia.com 1298020259 J * manana ~mayday090@84.17.25.149 1298021685 J * bonbons ~bonbons@2001:960:7ab:0:2c0:9fff:fe2d:39d 1298021920 Q * FireEgl Ping timeout: 480 seconds 1298021958 J * petzsch ~markus@dslb-092-078-224-117.pools.arcor-ip.net 1298022050 Q * swen_away Quit: KVIrc Insomnia 4.0.1, revision: 4541, sources date: 20100627, built on: 2010-08-03 16:04:47 UTC http://www.kvirc.net/ 1298022636 Q * BenG Quit: I Leave 1298022720 J * BenG ~bengreen@cpc12-aztw24-2-0-cust146.aztw.cable.virginmedia.com 1298023414 J * swenTjuln ~kvirc@77.111.2.36 1298023730 Q * BenG Read error: Connection reset by peer 1298024022 J * kir ~kir@swsoft-msk-nat.sw.ru 1298024048 P * kir 1298024265 J * BenG ~bengreen@cpc12-aztw24-2-0-cust146.aztw.cable.virginmedia.com 1298025711 J * ghislain1 ~AQUEOS@adsl2.aqueos.com 1298025846 Q * BenG Quit: I Leave 1298026091 Q * ghislain Ping timeout: 480 seconds 1298026635 Q * bsingh Ping timeout: 480 seconds 1298029147 Q * Piet Quit: Piet 1298030846 J * Piet ~Piet__@82VAAB0PC.tor-irc.dnsbl.oftc.net 1298032143 J * FireEgl ~FireEgl@173-25-19-139.client.mchsi.com 1298033743 M * Mr_Smoke Yay 1298033753 M * Mr_Smoke First "almost" live VServer migration successful :) 1298033788 M * Mr_Smoke I had only forgotten to update the router's cache when I migrated the IP address 1298033932 M * Wonka how do you migrate them? 1298033971 M * Mr_Smoke live rsync, stop on host 1, cold rsync, start on host 2 1298033972 M * Mr_Smoke :) 1298034101 M * Mr_Smoke The only thing I forgot was arping , after host2 was up 1298034113 M * Mr_Smoke (or before, if I'd added the IP manually) 1298034138 M * Mr_Smoke Of course, it breaks stuff like IRC, but it's a matter of minutes 1298036957 N * Bertl_zZ Bertl 1298036962 M * Bertl morning folks! 1298037078 M * Mr_Smoke mornin 1298037151 Q * manana Remote host closed the connection 1298037383 N * BobR_oO BobR 1298037491 J * nimmersatt ~chatzilla@HSI-KBW-095-208-086-087.hsi5.kabel-badenwuerttemberg.de 1298037517 M * nimmersatt heya #vserver, hi Bertl 1298037535 M * Bertl hi nimmersatt 1298037861 J * manana ~mayday090@84.17.25.149 1298037882 M * Mr_Smoke Bertl: hi 1298037905 M * Mr_Smoke Is there anything that can cause a process to *NOT* listen to localhost ? 1298037988 M * Bertl yes, if it is _not_ bound to the IP used for localhost, then it will not listen there 1298038048 M * Bertl what's the actual problem (if there is one) you are facing? 1298038182 M * Mr_Smoke This daemon I have, a flight simulator server software 1298038192 M * Mr_Smoke It used to live on a 2.6.22.19 vserver kernel 1298038201 M * Mr_Smoke I could telnet localhost and it worked 1298038208 M * Mr_Smoke There was no lo/127.0.0.1, but it worked 1298038217 M * Mr_Smoke Now I've moved it to a very recent exp/rc kernel 1298038228 M * Mr_Smoke and the daemon doesn't bind to 0.0.0.0 1298038239 M * Mr_Smoke It does that everywhere else, on other boxes 1298038347 M * Bertl so the software, probably proprietary, decides not to bind to 0.0.0.0 on this specific instance, or what? 1298038359 M * Mr_Smoke Yes 1298038366 M * Mr_Smoke That's exactly what it looks like 1298038404 N * BobR BobR_afk 1298038405 M * Bertl well, if that is the case, then your best chance is to contact the support for this fine piece of software and ask them why it would do so ... :) 1298038434 M * Bertl but I'd first check with strace and compare the startup cases (if possible) 1298038542 M * nimmersatt netstat problem - is there a way to identify a not showed PID? inside guest 1298038548 M * Mr_Smoke Not easily done, but I'll check 1298038570 M * Bertl nimmersatt: why wouldn't is show inside the guest? 1298038591 M * Bertl *it 1298038601 M * nimmersatt netstat -tulpnve 1298038610 M * nimmersatt tcp 0 0 10.3.253.8:80 0.0.0.0:* LISTEN 0 754062570 - 1298038629 M * nimmersatt also i found no corresponding entry in /proc giving the inode 1298038634 M * nimmersatt find /proc/[0-9]*/fd -lname 'socket:\[754062570\]' 1298038740 M * nimmersatt it is a ajax push engine writte in C called APE - it crashed - i restarted the vserver but the socket was still open 1298038773 M * Bertl that is not unusual, sockets can outlive a process 1298038839 M * nimmersatt i still have entries in the conntrack table - after stopping the vserver and unbinding its IP 1298038851 M * nimmersatt that is strange IMHO too 1298038883 M * Bertl iptables are not affected by guest start/stop 1298038885 M * nimmersatt but maybe i hitted another Debian bug ;) 1298038900 M * Bertl ah, it's the known broken 2.6.26 kernel? 1298038945 M * nimmersatt but ip addr ls does not show the IP of the guest - conntrack should be aware - right? 1298038947 M * nimmersatt 2.6.32-5-vserver-amd64 1298038959 M * Bertl in general, sockets should time out after a while, typically a few minutes but in unfortunate cases up to several hours 1298039000 M * Bertl is the IP configured on the host? 1298039010 M * nimmersatt i tweaked the sysctl to 5s for the different wait and fin stages 1298039026 M * nimmersatt in interfaces/0/ip 1298039053 M * Bertl does it show up on 'ip a l' on the host? 1298039101 M * nimmersatt not if the guest is stopped 1298039132 M * Bertl then start the guest and try to connect to that IP/port 1298039148 M * Bertl (as client not as service) 1298039171 M * nimmersatt from the host or inside the guest? 1298039178 M * Bertl from the host 1298039189 M * Bertl and check dmesg for unusual traces 1298039387 M * nimmersatt hmmm - i can telnet to the port - i got no respose but a connection.... 1298039461 M * nimmersatt dmesg is silent 1298039489 M * Bertl that would suggest that something is actually listening there :) 1298039617 M * nimmersatt but nor ps inside guest or vps on the host is showing the process - and my attempt to use the inode from netstat was also not successful for finding the responsible PID 1298039686 M * Bertl maybe 'something' happened which managed to hide that process, I'd start checking for kernel exploit code 1298039698 M * nimmersatt 15:34:08 up 186 days, 9:36, 5 users, load average: 0.75, 1.90, 4.79 - maybe its time for a reboot ;) 1298039934 M * nimmersatt cat /proc/sys/net/netfilter/nf_conntrack_count 1298039936 M * nimmersatt 190023 1298039938 M * nimmersatt is also quite high 1298040049 M * nimmersatt the code of the ajax engine is just a few kb - no kernel tampering 1298040106 M * nimmersatt is strace maybe responsible - in the run when it crashed i used strace to debug.... 1298040145 M * Bertl there are three possible situations IMHO 1298040163 M * Bertl 1) a hidden process is still listening and working 'as intended' 1298040187 M * Bertl 2) your process crashed (usually results in a kernel trace in dmesg) 1298040206 M * Bertl 3) you hit an obscure and bizarre bug in the networking stack 1298040327 M * nimmersatt [16086010.401017] aped[21126]: segfault at 0 ip 00007f9bc4b5ad57 sp 00007fff4b8e49a8 error 4 in libc-2.11.2.so[7f9bc4ad7000+158000] 1298040367 M * Bertl segfaults are usually uncritical, you are looking for a kernel stack trace if at all 1298040430 Q * petzsch Quit: Leaving. 1298040442 M * nimmersatt just segfaults, no stack trace 1298040733 N * BobR_afk BobR 1298040751 N * BobR BobR_oO 1298040850 M * nimmersatt is there any other way to free up the socket, other then restarting the host? 1298040883 M * nimmersatt maybe a cruel and rude way? ;) 1298041279 M * Mr_Smoke 49 seconds downtime. This is great :-) Another successful VServer migration across physical hosts :) 1298041357 M * Bertl sounds long, do your services take that long to shutdown/startup? 1298041412 M * Bertl a guest start/stop here usually takes a few seconds, 10 at max, the rsync with a 'hot' cache should not take more than a few seconds as well 1298041441 M * Bertl so all in all, I'd expect something around 15 seconds or so 1298041669 M * Mr_Smoke Bertl: This is actually between the time syslog went down on one host and back up on the other 1298041678 M * Mr_Smoke In the meantime I did a last, cold-state rsync 1298041799 M * Mr_Smoke And I think I also fixed a couple of missing subdirs like /proc and /usr/portage 1298041812 M * Mr_Smoke And I did that manually :) 1298041818 M * Bertl ah, okay :) 1298041831 M * Mr_Smoke So I agree, it should be much faster if automated etc 1298042104 N * ensc Guest1825 1298042114 J * ensc ~irc-ensc@p5DF2F0F6.dip.t-dialin.net 1298042512 Q * Guest1825 Ping timeout: 480 seconds 1298042861 J * bsingh ~balbir@122.172.205.15 1298043816 Q * thierryp Remote host closed the connection 1298043830 J * thierryp ~thierry@zankai.inria.fr 1298044393 Q * bsingh Ping timeout: 480 seconds 1298044938 J * dowdle ~dowdle@scott.coe.montana.edu 1298046772 Q * Chlorek Quit: Changing server 1298049063 J * Chlorek chlorek@2001:1418:1d4::3ffe 1298049477 Q * thierryp Remote host closed the connection 1298049486 J * petzsch ~markus@dslb-092-078-224-117.pools.arcor-ip.net 1298051481 J * imcsk8 ~ichavero@148.229.1.11 1298051956 J * hparker ~hparker@2001:470:1f0f:32c:beae:c5ff:fe01:b647 1298053698 J * Walex ~Walex@o1.phyip3.dur.ac.uk 1298056445 M * Mr_Smoke Hm 1298056464 M * Mr_Smoke Whats the command to change cgroup limits on a live guest? 1298056498 M * Mr_Smoke just echo stuff into /dev//cgroup ? 1298056637 M * Mr_Smoke Yeah that works :) 1298056773 M * Mr_Smoke Gotta love cgroups ! 1298057530 Q * petzsch Quit: Leaving. 1298058201 Q * manana Read error: Connection reset by peer 1298058802 M * Mr_Smoke And 3rd migration 1298058807 M * Mr_Smoke Boy I love it. 1298058890 M * Bertl :) 1298059203 J * manana ~mayday090@84.17.25.149 1298061249 J * cuba33ci_ ~cuba33ci@111-240-170-46.dynamic.hinet.net 1298061303 Q * Chlorek Quit: Reconnecting 1298061303 J * Chlorek chlorek@2001:1418:1d4::3ffe 1298061601 Q * cuba33ci Ping timeout: 480 seconds 1298061602 N * cuba33ci_ cuba33ci 1298063515 M * Mr_Smoke Bertl: questino regarting memory 1298063532 M * Mr_Smoke You can answer by pointing me towards theurl where the answer is :) 1298063533 M * Mr_Smoke here goes 1298063541 M * Mr_Smoke In a guest, I use free -m 1298063562 M * Mr_Smoke the "used" memory is 700M, bug actually 650M are cache, so it's just 50M really 1298063583 M * Mr_Smoke In vserver-stat, the RSS is 700M, the VSZ is 1.1G 1298063591 M * Mr_Smoke How come ? 1298064294 Q * hparker Remote host closed the connection 1298064579 J * hparker ~hparker@2001:470:1f0f:32c:beae:c5ff:fe01:b647 1298064703 J * petzsch ~markus@dslb-092-078-224-117.pools.arcor-ip.net 1298064793 M * Bertl well, assuming that you use the cgroup accounting, it works like on the host 1298064823 M * Bertl i.e. caches are accounted in a similar way if they 'belong' to the guest 1298064951 Q * Walex Remote host closed the connection 1298065112 M * Mr_Smoke Bertl: what's the difference between RSS and VSZ accounting for then ? 1298065162 M * Bertl depends on where it comes from 1298065176 M * Bertl usually it is the sum over all commited memory ranges 1298065193 M * Bertl (note that only a fraction is actually assigned) 1298065261 M * Mr_Smoke Hm 'k 1298065269 M * Mr_Smoke Wish I understood all that a lil' better 1298065589 M * Bertl take java for example, when you start it, it'll allocate 2GB of memory 1298065605 M * Bertl that's immediately accounted as VM 1298065618 M * Mr_Smoke ain't that sweet :) 1298065635 M * Bertl when it needs actual memory, it gets allocate in pages (within that 2GB area) 1298065658 M * Bertl that will be shown as RSS 1298065706 M * Mr_Smoke ok 1298066026 Q * manana Read error: Connection reset by peer 1298069590 Q * hparker Remote host closed the connection 1298070760 Q * bonbons Quit: Leaving 1298072377 Q * imcsk8 Quit: Leaving