1102982561 N * Bertl_zZ Bertl_ILL 1102984484 M * rs got a good rest? 1102984494 M * Bertl_ILL yeah, thanks ... 1102984565 M * Bertl_ILL have to do something now ... got bore of 'just' sleeping ;) 1102984615 M * Loki|muh hehe 1102984620 M * Loki|muh eat bananas ;) 1102984629 M * Bertl_ILL ook, ook! 1102984638 M * Loki|muh and salzletten and cola 1102984656 M * Bertl_ILL yeah, I'm 100% on cola ... 1102984662 M * rs salzletten ? 1102984670 M * rs sounds german :) 1102984713 M * Bertl_ILL it's kind of pretzel stick ... with salt 1102984735 M * rs pretzel ? 1102984816 M * Loki|muh lol 1102984822 M * Bertl_ILL rs: http://www.m-w.com/cgi-bin/dictionary?va=pretzel 1102984852 M * Loki|muh pretzel, like beergarden :) 1102984863 M * rs oh it's brezels 1102984894 M * Loki|muh ah ;) 1102984906 M * Loki|muh where do you live? 1102984910 M * rs paris :) 1102984950 M * Loki|muh ah 1102985391 N * nox nox{{away}} 1102985812 M * Bertl_ILL rs: so you talked about testing ngnet, IIRC ... what are your plans in this regard? 1102985838 M * rs nights :) 1102985849 M * rs do you want to start some tests ? 1102985884 M * Bertl_ILL well, ngnet changes a lot of things ... and how they are done ... 1102985903 M * Bertl_ILL one thing is the network setup (devices) which is now done from _inside_ the vserver 1102985936 M * Bertl_ILL this requires a slightly modified approach to settting it up at all ... 1102985940 M * rs hmm will needs some patchs for running distributions 1102985960 M * rs but I have some ideas... 1102985966 M * Bertl_ILL to emulate the current behaviour is simple ... 1102985981 M * Bertl_ILL we can use the pre/psot scripts for that 1102985984 M * rs but how to limit the vserver to one or more IP ? 1102985991 M * rs yeah was my idea :) 1102986005 M * rs really original :o) 1102986016 M * Bertl_ILL not at all ... but you should have a look at the ngnet stuff ... so you get a feeling for that ... 1102986044 M * rs you can't limit people to given ips ? 1102986073 M * Bertl_ILL you do not do that, you just 'allow' some IPs to be 'processed' by the ip stack 1102986091 M * rs for each context ? 1102986103 M * Bertl_ILL so if the customer decides to use a different one, then he can, but he will not get conenctivity 1102986113 M * Loki|muh hehe, thats nice 1102986127 M * rs nice yeah 1102986141 M * rs what happens if the vserver A setup the ip of the vserver B on the same node ? 1102986155 M * rs the same ip in its context that is 1102986158 M * Bertl_ILL that's no problem ... 1102986172 M * rs amazing 1102986178 M * Loki|muh yeah 1102986180 M * Bertl_ILL isn't it? ;) 1102986183 M * rs how does it work ? 1102986194 M * Bertl_ILL by pure magic ;) 1102986223 M * rs sure, hmmm but... give me more :) 1102986226 M * Bertl_ILL you also got 'working' interfaces inside ... 1102986244 M * rs working interfaces ? 1102986255 M * Bertl_ILL and usually you want 'lo' inside too ;) 1102986272 M * rs you mean I can get statistics of the iface and I will have statistics of only my context ? 1102986279 M * Bertl_ILL yep 1102986288 M * Bertl_ILL also ifup/down is possible ... 1102986295 M * Loki|muh great 1102986304 M * Bertl_ILL not to speak of creating new devices (like tun/tap at least in the future) 1102986306 M * rs what happens in the host context when you do that ? 1102986313 M * Loki|muh so no problem with using a ntp-server inside a vserver 1102986318 M * Bertl_ILL rs: nothing 1102986348 M * rs so, it's completly virtual 1102986355 M * Bertl_ILL yep 1102986371 M * Bertl_ILL but the network stack is still shared 1102986381 M * Bertl_ILL (required for good performance) 1102986394 M * rs ok you don't duplicate the stack as freevps does? :) 1102986402 M * Loki|muh has this method disadvantages? 1102986415 M * Bertl_ILL I don't know what freevps does ... but alexey said he uses the same method ... 1102986416 M * rs I guess it more complexe to code 1102986578 M * rs http://www.freevps.com/docs/faq.html#Inside%20VPS0 1102986667 M * Bertl_ILL well, the first part is similar then, the 'assign' part is different, this is done via a separate iptable ... 1102986762 M * rs I have an unrealated question: do you think that it could be possible to implement some kind of burst allocation for memory, I elaborate: 1102986823 M * rs if the vserver reach its rss limit, it can get more (configured value) if available on the memory, but all those allocated pages will be flagged to go to the swap more quickly if needed ? 1102986846 M * rs dunno if it clear :) 1102986901 M * Bertl_ILL yes, I know what you mean, but I think this requires a 'better' solution. why? 1102986913 M * Bertl_ILL simple, ebcause if you do so .. the following happens: 1102986946 M * Bertl_ILL context A,B, and C have 64MB space ... plus 64M burst 1102986957 M * Bertl_ILL the machine has a total of 256MB memory ... 1102986979 M * Bertl_ILL all contexts will burst, when needed to 128MB each ... 1102986986 M * rs yeah 1102986999 M * Bertl_ILL putting about 48MB into swap after that (in average) 1102987021 M * Bertl_ILL and the system will immediately start to trash ... 1102987039 M * rs trash, you mean OOM killer ? 1102987052 M * Bertl_ILL because every time a different context is run, the memory is swapped in, and the other context is swapped out 1102987069 M * Loki|muh Bertl_ILL: trashing = seitenflattern, oder? 1102987070 M * Bertl_ILL trashing = heavy swapping activity 1102987076 M * rs ok 1102987101 M * Bertl_ILL i.e. swapping is so havy that computation is not really possible ... 1102987120 M * rs hmm ok so it's not really efficient 1102987154 M * rs but we have to find a solution to emulate swap inside vserver without big performance hit 1102987174 M * rs I guess you have the 'better' solution in mind ? :) 1102987180 M * rs the/a 1102987195 M * Loki|muh fewer vservers on a host 1102987199 M * Loki|muh more ram in the host 1102987205 M * Bertl_ILL well, as usual there are already plans ;) 1102987227 M * Bertl_ILL that is what is 'planned' as soft limit for RSS ... (or VM) 1102987249 M * Bertl_ILL but to make it work, it requires the token bucket for swap-in pages 1102987260 A * albeiro thinks that Bertl have plans about everything up and not limited to world domination ;p 1102987271 M * Loki|muh lol 1102987274 M * rs wow we will have buckets all over the place :) 1102987291 M * Bertl_ILL yeah, another bucket for monsieur! ;) 1102987321 M * rs :) 1102987351 M * Bertl_ILL albeiro: hmm .. you know my plans about world domination? 1102987413 M * albeiro ups... 1102987418 M * albeiro i know too much :/ 1102987433 A * Bertl_ILL makes a note about albeiro ... 1102987434 M * albeiro subject to be ... eliminated 1102987456 A * albeiro feels threatned 1102987759 M * Bertl_ILL rs: so to get back to ngnet ... 1102987793 M * Bertl_ILL the changed setup required to have some 'script' to add the network devices for each context 1102987811 M * Bertl_ILL which should happen right after the start but from the host 1102987821 M * rs ok 1102987833 M * rs do you have those script or I have to do them ? 1102987835 M * Bertl_ILL and as the current util-vserver lacks such a mechanism atm (will require the C reqrite and sync) 1102987860 M * Bertl_ILL I decided to make a Q&D workaround by adding a callback via the vshelper 1102987913 M * Bertl_ILL so there is now a vshelper call, right after a context is created, but _before_ the syscall returns (and it is synchroneous) 1102987928 M * rs ok 1102987955 M * Bertl_ILL and a similar before the context is destroyed 1102987968 M * rs do we need an adopter for util-vserver or enrico plans to get back to it after his thesis ? 1102988010 M * Bertl_ILL I hope he gets back to it after that .. but if not we have to find someone ... 1102988067 M * rs ok I'm trying to figure out which patch to apply and in which order 1102988086 M * Bertl_ILL sec 1102988095 M * Bertl_ILL http://vserver.13thfloor.at/Experimental/patch-2.6.10-rc3-vs1.9.3.10.diff 1102988095 M * rs with .11 I just needs delta-11-ng7.2 ? 1102988104 M * rs ok so no .11 ? 1102988114 M * Bertl_ILL .10 is the best atm for ngnet ... 1102988121 M * rs ok 1102988128 M * Bertl_ILL http://vserver.13thfloor.at/Experimental/NGNET/diff-2.6.10-rc3-vs1.9.3.10-ng7.diff 1102988133 M * Bertl_ILL http://vserver.13thfloor.at/Experimental/NGNET/delta-2.6.10-rc3-vs1.9.3.10-ng7-ng7.2.diff 1102988142 M * Bertl_ILL you also need to patch iptables 1102988148 M * Bertl_ILL http://vserver.13thfloor.at/Experimental/NGNET/iptables-1.2.9-vnet.diff 1102988157 M * Bertl_ILL and to compile the vnet tool 1102988160 M * Bertl_ILL http://vserver.13thfloor.at/Experimental/NGNET/vnet-0.02.tar.bz2 1102988233 M * rs iptables patch is for host or vserver ? 1102988240 M * Bertl_ILL for the host 1102988250 M * rs ok 1102988441 M * Bertl_ILL ah, and start with the http://vserver.13thfloor.at/Experimental/NGNET/vnet3_setup.sh script then (without using any other vservers) 1102988539 M * rs this will break my node :) 1102988564 M * Bertl_ILL well, it _will_ break it anyway, it's highly experimental stuff ;) 1102988576 M * rs if you change the ip of the iface mounting the fsroot on a diskless node, you can imagine what it will happen :) 1102988606 M * Bertl_ILL of course you can adapt the part for the host ;) 1102988613 M * rs :) 1102988644 M * Bertl_ILL yeah, just play your games when I'm at 50% performance ;) 1102988661 M * rs :P 1102989509 M * rs the iptables patch seems to not apply on 1.2.11 1102989527 M * Bertl_ILL yeah, but it can be easily applied by hand ... 1102989538 M * Bertl_ILL (somebody updated it IIRC, to 1.2.11) 1102990496 A * albeiro goes to sleep. cu and have good whatever ;p 1102990500 M * rs yeah it's easy, got it to work with patch-o-matic too if someone interested 1102990508 M * Bertl_ILL albeiro: good night! 1102990510 M * rs albeiro: cya 1102990527 M * Bertl_ILL rs: can't hurt if you make it available somewhere ... 1102990549 M * Bertl_ILL (the patch is not final yet, but it's pretty close, I guess) 1102990601 M * rs http://rs.rhapsodyk.net/vserver/iptables-1.9.11-vnet.patch 1102990703 M * Bertl_ILL okay, added it ... 1102991020 M * rs net/core/dev.c: In function `deliver_skb': 1102991020 M * rs net/core/dev.c:1587: error: structure has no member named `nfxid' 1102991020 M * rs net/core/dev.c: In function `netif_receive_skb': 1102991020 M * rs net/core/dev.c:1698: error: structure has no member named `nfxid' 1102991020 M * rs net/core/dev.c: In function `unregister_netdevice': 1102991022 M * rs net/core/dev.c:3094: error: `__vx_dev_tail' undeclared (first use in this function) 1102991025 M * rs net/core/dev.c:3094: error: (Each undeclared identifier is reported only once 1102991028 M * rs net/core/dev.c:3094: error: for each function it appears in.) 1102991031 M * rs did I apply the patch wrongly ? 1102991055 M * rs gonna redo the whole process to be sure 1102991061 M * Bertl_ILL might be .. but I forgot something ... 1102991074 M * rs oh ? 1102991075 M * Bertl_ILL you should start with a config similar to that one: 1102991081 M * Bertl_ILL http://vserver.13thfloor.at/Experimental/NGNET/kernel-ng5.config 1102991089 M * rs what are important points ? 1102991098 M * Bertl_ILL the network stuff ... 1102991104 M * rs because my kernel config is somewhat special 1102991106 M * rs ok 1102991111 M * Bertl_ILL everything else isn't important ... 1102991140 M * Bertl_ILL just make sure that all the net stuff matches as close as possible and the *VSERVER* settings are identical 1102991236 M * Bertl_ILL just as checklist, 2.6.9, 2.6.10-rc3, 2.6.10-rc3-vs1.9.3.10, then ng7, then ng7-ng7.2 1102991444 M * rs I need multiple tables support, do you think it will hurt ? 1102991455 M * Bertl_ILL yep, it will ... 1102991464 M * rs damn 1102991475 M * rs maybe I don't need it with ngnet ? 1102991506 M * Bertl_ILL as I said, I would not think about starting a vserver atm ... 1102991513 M * rs ok 1102991563 M * rs I need CONFIG_IP_PHP_*, will it hurt ? 1102991577 M * Bertl_ILL what's that? 1102991584 M * rs CONFIG_IP_PNP_DHCP=y 1102991584 M * rs CONFIG_IP_PNP_BOOTP=y 1102991586 M * rs CONFIG_IP_PNP_RARP=y 1102991599 M * Bertl_ILL ah, PNP, no should be fine ... I hope ;) 1102991600 M * rs it's for diskless boot 1102991861 M * rs 802.1q should hurt ? 1102991888 M * Bertl_ILL hmm, vlans won't work as expected, but might work on the host side ... 1102991922 M * rs yeah I hope vservers wont be able to override vlan setings :) 1102992058 M * Loki|muh rs: quite a special configuration ;) 1102992220 M * rs Loki|muh: yeah, my nodes are diskless actually, and use a different vlan for all backend traffic 1102992254 M * Loki|muh all mounting over nfs? 1102992263 M * rs yeah for moment 1102992272 M * Loki|muh sounds interesting 1102992281 M * rs should change in the future, NFS is a hell 1102992310 M * Loki|muh change to what? 1102992313 M * rs we plan to use something like GFS over iSCSI or another transport 1102992340 M * rs or maybe lustre, we have to make some tests 1102992379 M * Loki|muh damn, wish I knew where I have stored this ****** mysql-file 2 months ago... 1102992401 M * Bertl_ILL slocate? 1102992494 M * Loki|muh i don't knew on which machine and with which filename 1102992507 M * Loki|muh on this night I was VERY stressed :/ 1102992611 M * Loki|muh yeah 1102992616 M * Loki|muh got it :) 1102992637 M * Loki|muh slocate bz2 :) 1102992676 A * Loki|muh hugs Bertl_ILL 1102992704 M * Loki|muh in the place i would have searched maybe in 2 hours or so *g* 1102992705 M * rs Bertl_ILL: same error with new config and correct patches 1102992850 M * Bertl_ILL sec 1102993117 M * Bertl_ILL please upload your .config somewhere 1102993175 M * rs http://rs.rhapsodyk.net/vserver/config-2.6.10-rc3-vs1.9.3.10-ng7.2.txt 1102993219 M * Bertl_ILL had to make sure ;) 1102993220 M * Bertl_ILL # CONFIG_VSERVER_NGNET is not set 1102993231 M * rs mouhahaha :) 1102993233 M * rs sorry 1102993267 M * Bertl_ILL you also have to enable the VNET stuff ... 1102993324 Q * albeiro Ping timeout: 480 seconds 1102993372 J * albeiro albeiro@linux.gentoo.pl 1102993377 M * rs it was but I guess it got dropped by the make config because the VSERVER_NGNET was not set 1102993405 M * Bertl_ILL yeah, Kconfig dependancies are tricky ;) 1102994559 M * rs hum 1102994697 M * Bertl_ILL means? 1102994735 M * rs lets try again, my node seems to not be reachable via the network 1102996051 M * rs ok I can get my node to boot with this kernel 1102996065 M * Bertl_ILL excellent! 1102996070 M * rs and while I don't have serial console on this machine, it will be hard to know why :) 1102996076 M * rs can't sorry :o) 1102996090 M * rs missed the negation one more time ;) 1102996093 M * Bertl_ILL ah ;) well, probably vlans ... 1102996108 M * rs unlikly 1102996109 M * Bertl_ILL what about firing up qemu and check with that? 1102996118 M * rs yeah why not 1102996139 M * rs btw, the node doesn't need vlan support to boot 1102996149 M * Bertl_ILL but maybe to logon? 1102996158 M * rs no 1102996178 M * rs vlan tagging is only needed for the second interface (the one with vserver ips) 1102996191 M * rs maybe I could remove vlan support for this kernel 1102996366 M * Bertl_ILL verify that you have all drivers you need to boot, otherwise try to get as close to the config as possible ... 1102996440 M * matti Bertl_ILL: You're ill? :-( 1102996460 M * Bertl_ILL yep, stomach flu ... 1102996469 M * matti :-(((((((( 1102996475 M * Bertl_ILL but it's getting better ... 1102996480 M * matti ;-D 1102996508 M * matti :] 1102999545 M * Bertl_ILL okay, I guess I'm off to bed now ... 1102999581 M * Bertl_ILL have a nice whatever everyone ... 1102999599 N * Bertl_ILL Bertl_zZ 1102999687 M * Loki|muh hmmm I should go to bed, too 1103000618 J * tnichols ~tnichols@202.191.111.12 1103000637 M * tnichols howdy, has anyone successfully got vservers working on amd64? 1103000827 M * tnichols I have it all compiled and installed, except am getting segfaults when trying to start a vserver 1103000880 M * tnichols secure-mount[10415]: segfault at 00000001bffff9e8 rip 0000000000401726 rsp 0000007fbffff720 error 4 1103000880 M * tnichols vuname[10430]: segfault at 00000004bffff930 rip 0000000000401f46 rsp 0000007fbfffe610 error 4 1103000910 M * tnichols Linux internal 2.6.9-vs1.9.3 #1 SMP Tue Dec 14 15:08:27 CST 2004 x86_64 AMD Opteron(tm) Processor 246 AuthenticAMD GNU/Linux 1103001182 M * Loki|muh is working gently for me 1103001190 M * Loki|muh but normal amd64 1103001194 M * Loki|muh not opteron 1103001226 Q * tnichols Quit: BitchX: No windows left! 1103001235 J * tnichols ~tnichols@202.191.111.12 1103001306 M * Loki|muh nice ip ;) 1103001341 M * tnichols ta :-) 1103001744 M * mugwump tnichols: yes, I have 1103001764 M * mugwump however, not recently 1103001855 M * tnichols I'm using 2.6.9-vs1.9.3 and the 0.30.196 tools 1103001868 M * mugwump tnichols: try the latest patch: (http://vserver.13thfloor.at/Experimental/patch-2.6.10-rc3-vs1.9.3.11.diff) 1103001904 M * mugwump if it still happens on that combination, then try to locate the segfault address in the System.map 1103001913 M * tnichols ok 1103001924 M * mugwump I've got to go - but with that information it should be possible to see which line of code or function is causing the segfault 1103001935 M * tnichols thanks for the tip 1103001959 M * mugwump if the System.map isn't useful, look into the ksymoops program 1103002449 M * infowolfe mugwump: it's a known problem with gentoo's hardened toolchain 1103002464 M * infowolfe util-vserver likes to die in mysterious ways 1103002494 M * tnichols as I am finding out ... :-) 1103002823 M * infowolfe tnichols: i'm in love with your machine... lol 1103002830 M * tnichols hehehe 1103002832 M * infowolfe scsi? 1103002835 M * tnichols nah sata 1103002841 M * infowolfe still sexy as hell 1103002843 M * tnichols unfortunately software raid too :-( 1103002847 M * infowolfe doesn't matter... 1103002850 M * infowolfe it's still BLAZING fast 1103002856 M * tnichols yeah :-) 1103002863 M * tnichols i wish it was mine for home 1103002887 M * infowolfe lol, work machine? 1103002895 M * tnichols yeah 1103002899 M * infowolfe i'm still jealous 1103002912 M * infowolfe i've got a celery 1.7 right now... *sigh* 1103002921 M * tnichols i'd shove a nice powerful pci-x gfx card in and that would be superb for gaming 1103002926 M * infowolfe lol 1103002931 M * infowolfe i'd run gentoo on it 1103002939 M * infowolfe use it as a terminal server 1103002961 M * infowolfe what raid level are you using? 1103002966 M * tnichols I'm doing some paperwork... back shorlty 1103002967 M * tnichols (raid5) 1103002981 M * infowolfe right on, i'll go poking in a moment 1103003165 M * infowolfe hrm... 250GB drives... software raid5... that's pretty damn quick, faster than my old LSI MegaRAID 150-4 was :( 1103003215 A * infowolfe is definately jealous 1103003221 M * tnichols in a way it is a little over-kill for what we will ultimately be using on it ... :-) 1103003330 M * tnichols looking at the unpack output it looks like it is including the PIE patches etc 1103003631 M * infowolfe i was making coffee, so i didn't see it, but i'll start it again 1103003708 M * infowolfe nope, the pie/ssp patches aren't applied now... 1103003735 M * infowolfe hrm, that's odd 1103003736 M * infowolfe brb 1103003816 M * tnichols wtf 1103003865 M * tnichols what about * Updating gcc to use SSP from libc ... 1103003894 M * tnichols then the various bugfixes mentioning pie and ssp below or am I thinking wrong? :-) 1103003930 M * infowolfe i'm trying to get ahold of the amd64 herd right now to figure out what the deal is 1103003951 M * tnichols ok :-) 1103004205 M * infowolfe hardened isn't being very helpful :-p 1103005302 M * tnichols looks the same ... the hardened folks still not being helpful? 1103005337 M * infowolfe i'm talking to amd64 right now 1103005363 M * tnichols ok.. thanks for all the help with this btw :-) 1103005433 M * infowolfe no problem 1103005443 M * infowolfe i was just trying to fix the 2.6.7 pty bug anyway 1103007115 Q * tchan Quit: leaving 1103007158 J * tchan ~tchan@c-24-13-81-164.client.comcast.net 1103008427 J * t ~tnichols@CPE-139-168-208-47.sa.bigpond.net.au 1103008552 J * ntrs ntrs@SP2-24.207.228.55.charter-stl.com 1103008864 Q * ntrs_ Ping timeout: 480 seconds 1103011532 J * dsanta_ ~santa@c68.190.156.105.roc.mn.charter.com 1103011600 Q * dsanta Ping timeout: 480 seconds 1103012338 J * BWare ~bware@212.26.196.41 1103015202 M * infowolfe .196 is hideously broken on x86_64 1103015246 M * Loki|muh ack 1103015246 Q * sannes Read error: Connection reset by peer 1103016621 J * mhepp ~mhepp@r72s22p13.home.nbox.cz 1103016629 M * mhepp hello 1103016657 M * infowolfe hi mhepp 1103016664 M * infowolfe Loki|muh: you do x86_64, right? 1103016772 M * infowolfe Error: Incorrect register `%rax' used with `l' suffix 1103020570 M * infowolfe Loki|muh: adding -m32 to CFLAGS before compiling util-vserver does tht rick 1103020571 M * infowolfe trick* 1103021192 N * nox{{away}} nox 1103022272 M * infowolfe ensc: ping? 1103022430 J * sannes ~ace@home.skarby.no 1103022707 N * nox nox|afk 1103023154 N * nox|afk nox 1103023442 N * nox nox|afk 1103023501 P * nox|afk donĀ“t wanna disturb while ircplaying 1103023698 M * infowolfe anybody other than me alive? 1103023826 M * Loki|muh infowolfe: doesnt work for me 1103023839 M * Loki|muh as I had another problem 1103023879 M * infowolfe Loki|muh: what other problem is that? 1103023911 M * infowolfe btw, -m32 only works when you've got a non amd64 /usr/include/bits/select.h 1103023930 M * infowolfe which, i might mention is really effin ugly because the x86 version has inline assembly in it 1103023957 M * infowolfe so at this point, since i'm not exactly *sure* what enrico was doing with keep-ctx-alive.c i can't fix.. 1103024053 M * Loki|muh infowolfe: http://sial.org/pbot/6023 this is the error 1103024241 M * infowolfe Loki|muh: that's because dietlibc doesn't have a nice syscall under x86_64 1103024281 M * Loki|muh yeah, but its not so bad, 195 compiles, so is use this 1103024284 M * Loki|muh version 1103024312 M * infowolfe i'm going to just make a gentoo package on a p4 machine i think 1103024341 J * nox ~nox@c150103.adsl.hansenet.de 1103024481 M * infowolfe Loki|muh: you use gentoo, right? 1103024521 M * Loki|muh no, we are using ubuntu for the hosts 1103024617 M * infowolfe sorry then? 1103024620 M * infowolfe gtg though 1103024628 M * Loki|muh gtg? 1103024736 M * t gtg = got to go 1103024742 M * Loki|muh ah 1103025192 M * infowolfe back 1103025213 M * infowolfe and mildly perplexed 1103025280 M * Loki|muh within 10minutes? 1103025292 M * infowolfe filipino "girl" that sounds and looks like a man that claimed to be of japanese and spanish descent... enough for me to say "eeks" 1103025313 M * infowolfe i wonder why so many people are so dishonest... 1103025327 M * Loki|muh uah 1103025349 A * infowolfe isn't sure if he should be confused or disturbed 1103025586 M * infowolfe Loki|muh: isn't ubuntu a desktop distro? 1103025739 M * Loki|muh yes, mainly, but it is even very usable for servers 1103025747 M * Loki|muh and has a good amd64 support 1103025786 M * infowolfe hrm... interesting 1103025929 M * infowolfe plone looks cool, but i'm not enthusiastic about running zope :-p 1103025930 Q * grecea Read error: Connection reset by peer 1103027443 M * infowolfe Loki|muh: .195 doesn't work either... i'm wondering if anybody's been able to build util-vserver for x86_64 at all :-\ 1103027554 M * Loki|muh .195 is working for me 1103027668 J * grecea ~grecea@h-195-22-237-74.mdl.net 1103027705 Q * t Ping timeout: 480 seconds 1103029296 J * Duckx ~duckx@195.75.27.158 1103030385 Q * mhepp Quit: mhepp caught signal: Autobus error 1103030508 J * t_ ~tnichols@CPE-139-168-208-168.sa.bigpond.net.au 1103033410 Q * sannes Read error: Connection reset by peer 1103035705 J * DuckKing ~Duck@dyn-83-152-205-175.ppp.tiscali.fr 1103036067 Q * Duckx Ping timeout: 480 seconds 1103036130 Q * DuckMaster Ping timeout: 480 seconds 1103036415 J * Duckx ~duckx@195.75.27.158 1103037003 N * Bertl_zZ Bertl_ILL 1103037011 M * Bertl_ILL greetings folks! 1103037036 M * Bertl_ILL t_: still around? 1103037037 M * TheSeer hmm.. if you're sick you should stay in bed ;) 1103037054 M * Bertl_ILL thanks, I'm basically in bed ;) 1103037090 A * TheSeer sends Bertl_ILL a virtual 'get well soon' card 1103037108 M * Bertl_ILL tx again ... hopefully I will ... 1103037910 M * Bertl_ILL okay, off again ... back later ... 1103037918 N * Bertl_ILL Bertl_zZ 1103040376 J * sannes ~ace@home.skarby.no 1103041110 Q * sannes Read error: Connection reset by peer 1103041217 M * Loki|muh do you thank that /dev/stdin would be harmless inside a vserver? 1103041221 M * Loki|muh any suggestions? 1103045597 N * Bertl_zZ Bertl 1103045637 M * Bertl what is /dev/stdin supposed to be? 1103046085 Q * Duckx Quit: Leaving 1103046224 M * Bertl t_: now around? 1103046372 M * Loki|muh Bertl: its a link to /proc/self/fd/0 1103046406 M * Loki|muh Bertl: and this is for me: lrwx------ 1 loki adm 64 2004-12-14 18:46 /proc/self/fd/0 -> /dev/pts/5 1103046433 M * Bertl well, of course you can add symlinks to anywhere you like ... they are fine ... 1103046785 M * Loki|muh ah okay works 1103046790 M * Loki|muh got something wrong 1103048078 J * sannes ~ace@home.skarby.no 1103048090 M * Bertl wb sannes! 1103048230 J * mhepp ~mhepp@r72s22p13.home.nbox.cz 1103048241 M * Bertl welcome mhepp! 1103048256 M * mhepp Ahoj Bertl!!! 1103050829 J * smekkawi ~smekkawi@ip37net160.skylogicnet.it 1103050837 M * Bertl welcome smekkawi! 1103051005 M * smekkawi hi 1103051197 M * smekkawi I am planning to test the vserver and was wondering if i can get some assistance 1103051242 M * Bertl what kind of assistance do you expect for 'planning' a test? 1103051343 M * smekkawi planning to run 2 apache webservers 1 postfix server on a one server in preparation to consolidate 1103051413 M * Bertl well, apache and postfix works fine inside vservers ... 1103051497 M * smekkawi so I am currently reading the installation docs on the site , also trying to specify the minimum hardware requirements needed 1103051576 M * Bertl for about 3 servers (2x apache and 1x postfix) probably some PII/400 with 512MB memory will do fine ... 1103051650 M * smekkawi wow this is encouraging, the box I am will use for testing is P4, 1GB memory 1103051675 M * smekkawi I am doing this test a part of a migration plan from win2000 to linux 1103051697 M * Bertl the absolute limit for a dual Xeon box, with 2GB ram is about 100 vservers 1103051728 M * albeiro Bertl: what's inside ? 1103051778 M * Bertl inside where? 1103051786 M * albeiro 100 vservers 1103051793 M * albeiro you're speaking about 1103051809 M * Bertl typical setup apache, postfix, sshd, syslog 1103051817 M * albeiro k 1103051824 M * albeiro not bad ;p 1103051850 M * Bertl well, really depends on the load, usage and configuration 1103051871 M * albeiro sure 1103051885 M * Bertl just leaving the cron daily active at 4am for all vservers will kill the machine ... 1103052023 M * smekkawi Bertl : the I'm using redhat enterprise 3.0 ES , is there any recommended distribution or any recommended settings related to distrib used 1103052101 M * Bertl don't know about RHE 3.0 ES, but you probably need a vanilla kernel for linux-vserver to work properly ... so if that distro works fine with a 2.4.28 or 2.6.10, then it should be fine too ... 1103052250 M * smekkawi so far no problem, already migrated some domino servers, proxies and file servers its working fine so far 1103052883 M * smekkawi RHE 3.0 uses 2.4.21 kernel so I should update to 2.4.28 and patch it before installation 1103052935 M * Bertl yes, but you should try with a vanilla 2.4.28 first, just to make sure everything works as expected ... 1103052949 M * smekkawi ok 1103053343 P * nox bye 1103055116 Q * albeiro Ping timeout: 480 seconds 1103055123 J * albeiro albeiro@linux.gentoo.pl 1103056540 Q * sannes Read error: Connection reset by peer 1103056913 P * smekkawi Leaving 1103057234 M * infowolfe Bertl: when was the last time you saw ensc? 1103057265 M * daniel_hozac he has been active on the list recently. 1103057265 M * Bertl at the LinuxTag in germany ;) 1103058239 Q * mhepp Remote host closed the connection 1103061871 Q * virtuoso Read error: Connection reset by peer 1103061883 J * virtuoso ~s0t0na@spb.sot.com 1103061888 Q * no_maam Read error: Connection reset by peer 1103062846 J * noxlux ~nox@c150103.adsl.hansenet.de 1103062875 P * noxlux bye 1103062890 J * nox ~vps@c150103.adsl.hansenet.de 1103063513 J * sannes ~ace@home.skarby.no 1103063624 M * Bertl wb nox, sannes! 1103063647 M * nox thanks Bertl 1103063796 J * rpetre ~rpetre@82.76.1.167 1103063817 M * Bertl welcome rpetre! 1103063834 M * rpetre thanks 1103063864 M * rpetre what is the proper way to ask help for a newbie here? ;) 1103063877 M * Bertl hmm .. /msg ? ;) 1103063903 M * Bertl (i.e. no rules, just ask politely ;) 1103063919 M * rpetre i'm having a weird situation with an "immortal" process inside a vserver, maybe somebody saw that? 1103063939 M * Bertl stuck in 'D' mode, or a zombie? 1103063960 M * rpetre nope, running, but the /proc/$pid/exe points nowhere 1103063999 M * rpetre which pretty much makes "init" unhappy when killing the vserver 1103064009 M * Bertl hmm, points nowhere, or just points somewhere, but there is nothing? 1103064013 M * rpetre or whatever subtitutes init in vservers 1103064017 M * rpetre wait 1103064077 M * rpetre lrwxrwxrwx 1 root root 0 Dec 15 00:40 exe 1103064081 M * rpetre pointing nowhere 1103064106 M * rpetre also root and cwd look the same 1103064137 M * Bertl hmm, could you upload an ls -la /proc/$pid somewhere (i.e. pastebin.com) 1103064158 M * rpetre wait 1103064167 M * rpetre there are also two children 1103064184 M * rpetre ok 1103064334 M * rpetre http://www.pastebin.com/129739 1103064359 M * rpetre that was all inside the vserver 1103064396 M * Bertl yep 1103064416 M * rpetre i also tried killing it, no success 1103064419 M * Bertl looks like a parent process died, before the children could be reaped 1103064440 M * Bertl butthe state is funny ... does it still consume cpu? 1103064445 M * rpetre yup 1103064455 M * rpetre or at least seems to 1103064458 M * Bertl what if you send a -STOP ? 1103064516 M * rpetre no effect 1103064559 M * Bertl that is at least interesting .. okay, what kernel version and what patches, and how did you get there? 1103064588 M * rpetre it's a 2.4.26 1103064648 M * rpetre it uses the kernel from http://oss.sns.ro/Hilux.html (it also lists the patches) 1103064761 M * Bertl hum, well, no idea what 'ctx vserver' refers to ... 1103064840 M * Bertl there used to be patches called ctx, about more than a year ago ... I don't think they used those ... 1103064843 M * rpetre it's the one from linux-vserver afaik 1103064866 M * rpetre i guess it's a political correctness thing 1103064891 M * rpetre it comes from a guy that uses "gnu/linux" on a regular basis :) 1103064926 M * Bertl well, hmm, please explain the political correctness part in this specific case then ... 1103064951 M * rpetre dunno, don't jump at me, i'm an innocent victim :) 1103064965 M * Bertl don't worry, I won't jump at you ;) 1103065000 M * Bertl but I don't see how you deduce that from this page .. so you obviously know more than I (about that distro?) 1103065024 M * rpetre i had a little part in it 1103065064 M * Bertl okay, well, now linux-vserver might have a little part in it too ;) 1103065067 M * rpetre i know personally every guy involved 1103065073 M * rpetre yes, indeed 1103065081 M * Bertl excellent, then please let me know the following details: 1103065100 M * Bertl what does 'ctx vserver' refer to? and what aptches where used ... 1103065111 M * Bertl (well, I know they had to be 'adapted') 1103065141 M * rpetre i'll need to dig a little on that 1103065143 M * rpetre brb 1103065154 M * Bertl k, tx! 1103065183 M * mugwump bertl, you mention on the list "the nat table is not relevant for linux-vserver as there is nothing to 'forward'." 1103065194 M * mugwump and how are you btw, recovering? 1103065205 M * Bertl yes, and fine, thanks! 1103065242 M * mugwump however, it has been my experience that I have had to set up SNAT to allow vservers to connect to the outside world. ie, the source address is the 192.168.* address otherwise 1103065274 M * Bertl yep, but in the mangle table, not the nat table ... 1103065296 M * Bertl the reason is simple, nat is designed to 'map' ports 1103065313 M * Bertl (which isn't required if you allocate the ports on the same machine) 1103065374 M * mugwump odd, I add rules to the nat table, not the mangle table 1103065426 M * Bertl well, probably doesn't matter as long as you do not put them into the FORWARD chain .. 1103065463 M * Bertl but let me check the packet flow ... 1103065507 M * mugwump no, they do into either PREROUTING or POSTROUTING 1103065510 M * mugwump s/do/go/ 1103065590 M * Bertl yep, you are right, every incoming packet passes the nat table (PREROUTING and POSTROUTING on output) 1103065635 M * Bertl so please correct my bogus statement on the ml ... 1103065638 M * mugwump eg $IPTABLES -t nat -A POSTROUTING -o eth0 -s 192.168.255.0/24 -j SNAT --to-sourc 1103065653 M * mugwump bah, my serial terminal didn't wrap 1103065680 M * mugwump ok will do. just thought that source addresses might magically work without any firewalling whatsoever 1103065737 M * mugwump no, I've just checked and the source address gets set to the vserver address 1103065744 M * mugwump without any firewalling that is 1103065829 M * mugwump whoa - http://lists.labs.libre-entreprise.org/mailman/listinfo/debian-vserver-devel (there's a -user list as well) 1103065855 M * mugwump did they announce that? 1103065915 M * Bertl not that I know of ... 1103066190 M * nox if it helps the debian-vserver-devel (: 1103066230 M * Bertl yeah, sure, if it helps, let's add another 3 *G* 1103066247 M * nox *lol* 1103066270 M * rpetre Bertl: i got the kernel info you requested 1103066273 M * nox prob seems to be big enough 4 own list 1103066303 M * Bertl rpetre: great! let's hear ... 1103066320 M * rpetre a vanilla 2.4.26 was used 1103066343 M * rpetre with linux-vserver 1.28 from http://www.13thfloor.at/vserver/s_release/v1.28/ 1103066356 M * rpetre every patch applied clean 1103066372 M * rpetre no hacking was required 1103066380 M * Bertl excellent, so the ctx vserver is just for 'nostalgy' 1103066384 M * rpetre yes 1103066391 M * rpetre the guy admitted it :) 1103066419 M * rpetre if the current webpage offensive i can arrange to be changed 1103066449 M * Bertl well if he likes it confusing, no problem with that ... 1103066484 M * Bertl 1.29 didn't change anything regarding processes, so you should be up to date with 1.28 1103066493 M * rpetre ok then 1103066524 M * Bertl is this something which happened once, or can you reproduce this effect? 1103066529 M * rpetre except i haven't got that "McCloud" problem on any vserver setup yet 1103066531 M * mugwump Both of those lists have a single message from August. Simon Baptiste is the founder, and Benjamin Poussin is also involved 1103066565 M * rpetre i can reboot the machine and restart the offending process, presumambly it went crazy by itself 1103066575 M * Bertl mugwump: well, one message since August isn't that bad, is it? 1103066577 M * mugwump August 2003, mind 1103066588 A * Bertl looks busy ... 1103066648 M * rpetre dunno, maybe it's a problem with ices2 too, i haven't used that in vserver until now 1103066688 M * Bertl basically it's strange if you get 'unkillable' but running ('R') processes 1103066698 M * rpetre yes 1103066726 M * rpetre that's why i'm here and not on #ices2 :) 1103066733 M * mugwump whereas unkillable but blocking on a hung kernel thread ('D') happens all the time 1103066738 M * rpetre i'll reboot the machine again now 1103066754 M * rpetre and see if it happens again 1103066758 M * rpetre any other idea? 1103066767 M * Bertl not ATM 1103066803 M * rpetre running it under strace might provide some insights? 1103066843 M * Bertl hmm, I doubt that you can attach, because strace requires SIGSTOP to work ... 1103066890 M * rpetre i'll run strace after boot 1103066906 M * rpetre and start it with strace 1103066911 M * Bertl first check if it happens again ... 1103066931 M * Bertl if you _can_ reproduce the issue, we have half the way ... 1103066967 M * rpetre ok 1103068010 M * rpetre Bertl: it happened again 1103068058 M * Bertl good! 1103068105 M * Bertl okay, try ps -af and upload the output as before ... 1103068154 M * rpetre http://www.pastebin.com/129763 1103068194 M * rpetre what do the '[]' mean? 1103068207 M * Bertl it usually denotes a kernel thread 1103068229 M * Bertl and it shows the init as parent ... 1103068234 M * rpetre that's bad in my case, i presume 1103068250 J * locksy ~locksy@mrtg.sisgroup.com.au 1103068254 M * Bertl it's at least strange ... 1103068274 M * Bertl how is the vserver started, with fakeinit? 1103068281 M * Bertl welcome locksy! 1103068355 Q * jsambrook Ping timeout: 480 seconds 1103068364 M * Bertl rpetre: could you upload the vserver config too? maybe with IPs removed if that is sensitive ... 1103068416 M * rpetre http://www.pastebin.com/129764 1103068478 M * Bertl hmm, NET_RAW and RAW_IO? 1103068709 M * rpetre should i change that and try again? 1103068759 N * locksy locksy_AFK 1103068761 M * Bertl it's a security issue, so I ask myself why it is used?