1107043517 Q * we2by Remote host closed the connection 1107044332 J * we2by ~jinxi@145.116.6.75 1107045413 J * prae ~prae@sherpadown.net 1107045423 Q * prae Quit: 1107048595 Q * sladen Ping timeout: 480 seconds 1107048848 J * sladen paul@starsky.19inch.net 1107057231 N * Bertl_oO Bertl 1107057246 M * Bertl evening folks! 1107057317 M * Bertl jmp: if you have some time tomorrow, we can schedule a special ngnet session ... 1107057424 M * Bertl okay, I'm off to bed for today ... (was a long night ;) 1107057440 N * Bertl Bertl_zZ 1107057935 Q * Snow-Man Ping timeout: 480 seconds 1107058209 J * Snow-Man ~sfrost@snowman.net 1107060290 J * wishes ~wishes@203-118-140-90.pppoe.ihug.co.nz 1107060429 Q * wishes_ Ping timeout: 480 seconds 1107070263 Q * flock Ping timeout: 480 seconds 1107070820 J * flock ~restless@l192-117-111-12.broadband.actcom.net.il 1107079942 J * Diadoxx jhkjhkjk@60.246.253.42 1107079944 M * Diadoxx A Brand new Space RPG game, click here for info: http://diadoxx.ingame.org 1107079946 M * Diadoxx A Brand new Space RPG game, click here for info: http://diadoxx.ingame.org . 1107079952 P * Diadoxx 1107080247 Q * Seraph Remote host closed the connection 1107082787 J * Seraph kk@projects.verfaction.de 1107082910 M * aba ok, one question: I want to be able to mount /proc safely in a vserver. Is that possible? 1107083715 M * daniel_hozac aba: the tools should mount /proc automatically. 1107083745 M * aba daniel_hozac: not for me, I want to be able to mount proc inside a chroot inside a vserver. 1107083762 M * aba (or with other words, I want to be able to use pbuilder inside a vserver) 1107085760 J * prae ~prae@sherpadown.net 1107086879 J * matti matti@linux.gentoo.pl 1107092248 N * Bertl_zZ Bertl 1107092259 M * Bertl morning folks! 1107092264 M * jmp Hello 1107092272 M * aba Hi Bertl 1107092302 M * Bertl hey jmp! aba! 1107092305 M * jmp if you have time I am ready to do some test with you about ngnet 1107092341 M * Bertl great! yeah, guess we can do some testing ... 1107092356 M * Bertl aba: there is a SECURE_MOUNT capability in 2.6(1.9.x 1107092363 M * jmp was my yesterday E-mail making sensn?? 1107092403 M * Bertl well, partially ... but it was late when I read it ... so let me reread it again ... 1107092440 M * jmp I proved the ARP point by forcing remote to with an ARP value, the remote was able 1107092443 M * Bertl mserver means master-server aka as host, right? 1107092445 M * jmp to ping vserver. 1107092453 M * jmp (yes) 1107092473 M * jmp the baseic level the "real physical one" 1107092476 M * Bertl okay, let's stick to host for now ... 1107092482 Q * we2by Remote host closed the connection 1107092496 M * jmp you mean "mserver must be called "host" :-}} 1107092549 M * Bertl I reserve the 'real' for the machine which is running my QEMU (atm) so it's easier to communicate the setup 1107092838 M * jmp interesting point if on the vserver I force the route (route add -net....) 1107092860 M * jmp then on the host telnet 127.0.0.1 smtp is not repsonding any, while 1107092872 M * jmp vserver same test still respond... 1107092886 M * jmp you have 2 sendmail running both on the host and vserver 1107092958 M * jmp bertl stil here? 1107093139 M * Bertl jup, just my internet connection is still bad ... 1107093180 M * Bertl so as far as I understood, it's the arp going wrong for you? 1107093207 M * jmp I am pretty sure I missed (messed?) something with ARP, but can't put my finger on it 1107093237 M * jmp here is what I set (MANUALY) as route on the vserver 1107093240 M * Bertl okay, give me a few minites to get my test setup up and running ... then we'll try to redo your setup ... 1107093247 M * jmp estination Gateway Genmask Flags Metric Ref Use Iface 1107093247 M * jmp 192.219.254.0 * 255.255.255.0 U 0 0 0 eth0 1107093247 M * jmp 169.254.0.0 * 255.255.0.0 U 0 0 0 eth0 1107093247 M * jmp 127.0.0.0 * 255.0.0.0 U 0 0 0 lo 1107093248 M * jmp default 192.219.254.33 0.0.0.0 UG 0 0 0 eth0 1107093285 M * jmp what is the most strant is the lost of 127.0.0.1 on host when route 192.219.254 is set on vserver 1107093291 M * jmp (stange) 1107093322 M * jmp (169.254.0.0) was set by service network start 1107093329 M * jmp on the vserver 1107093701 J * nish ~nish@220.224.28.111 1107093715 M * nish evening all :) 1107093800 M * jmp hello 1107093858 M * Bertl hey nish! 1107093889 M * Bertl jmp: could you paste the output of 'cat /proc/vnet' please? 1107093996 M * jmp 2 sec 1107094028 M * jmp cat /proc/vnet 1107094028 M * jmp #vnet #xid real vnet dev 1107094029 M * jmp 10 49153 lo lo 0009 9 1107094029 M * jmp 11 49153 eth0 eth0 0003 8 1107094042 M * jmp same thing on both side (vserver and main) 1107094074 M * Bertl yep, just the host is relevant here ... 1107094096 M * jmp cat /proc/vnet 1107094096 M * jmp #vnet #xid real vnet dev 1107094097 M * jmp 10 49153 lo lo 0009 9 1107094097 M * Bertl you are using dynamic ids? 1107094097 M * jmp 11 49153 eth0 eth0 0003 8 1107094102 M * jmp niet 1107094107 M * jmp (I mean no) 1107094121 M * jmp caution you mean dynamic IP? 1107094129 M * jmp what are "ids" ? 1107094133 M * Bertl but 49153 is a strong indication for dynamic xids ;) 1107094146 M * jmp Ok yes dynamic context 1107094175 M * Bertl I'd suggest (to simplify further testing) to use an xid of 10 (for this vserver) 1107094201 M * Bertl further, to reduce the possibility of confusion, to use en0 inside the vserver (instead of eth0) 1107094242 M * Bertl you can automate the interface setup with a small modification to vshhelper and a script ... 1107094272 M * jmp 2 sec 1107094273 M * Bertl http://vserver.13thfloor.at/Experimental/NGNET/delta-vshelper-delegate.diff 1107094326 M * Bertl http://vserver.13thfloor.at/Experimental/NGNET/startup.sh <-- script to be copied to /etc/vservers/.defaults/apps/vshelper-delegate/startup 1107094338 M * jmp to setup vserver I am using the vserver command (I would say , as usual) 1107094345 M * Bertl then touch touch /etc/vservers/.defaults/apps/vshelper/warning-disabled 1107094352 M * jmp what is the command I should use to set the vserver? 1107094361 M * Bertl vserver start 1107094391 M * Bertl this will call the startup.sh script (when modified as I just pasted) 1107094410 M * Bertl which in turn will create the interfaces and set an 'allowed' ip 1107094424 M * Bertl (you have to adjust it to your needs/setup) 1107094466 M * Bertl if your setup (as pasted above) uses two ips 1107094471 M * jmp ok, let me work here, (give me one hour or two), obviously we are not on the same track here, I am using Jack 1107094482 M * Bertl one in 192.219.254.0, and another in 169.254.0.0 1107094507 M * jmp "/etc/vservers" config setup... 1107094511 M * jmp Bertl 1107094521 M * Bertl you mean the legacy tools? 1107094537 M * jmp is my undertaing right we should be able to run"service network start" on the vserver 1107094547 M * Bertl yes, that's correct 1107094551 M * jmp (Yes The legacy tools) 1107094573 M * jmp ok 169.254 was set up via service network start 1107094596 M * Bertl yes, but you have to 'allow' those ips via the iptables setup (vnet table) 1107094598 M * jmp ok, let me align myself on your tools 1107094632 M * jmp Yes, I understood this and have done it manually (such I was able to see effect) 1107094655 M * Bertl okay, did you do it for both addresses? 1107094679 M * jmp from my stand point beside the dynamic context assigment , legacy setup should make any difference 1107094721 M * jmp I have done it manually for 127.0.0.1 and the public ip 1107094723 M * Bertl the context creation is a little different, but it might work with the legacy tools too, never tried though ... 1107094731 M * jmp 169.254 is just noise here for now 1107094753 M * jmp (it was my understanding too) 1107094756 M * Bertl okay, if you don't expect it work (the 169.254*) then that'sfine 1107094770 M * jmp ok... 1107094772 M * Bertl you should make sure that the legacy tools do not use any chbind stuff ... 1107094785 M * jmp 2 sec 1107094803 M * Bertl well, basically you could disable the chbind in the kernel (to make sure that it doesnt) 1107094871 M * jmp I do confirm legacy vserver us chbind... 1107094912 M * Bertl well, let me upload a new version of the ngnet patches which just ignores chbind calls ... 1107094920 M * jmp chbind is used to setup a non routable IP as aliase on eth1 1107094948 M * jmp no, do not upload a new version just for me 1107094966 M * Bertl they just don't make any sense right now .. so no problem with that (and it will sync your version with mine) 1107094974 M * jmp if you are telling my setup should work, but discrapancy is legacy tools 1107094985 M * jmp I'll align myself on the "new" tools" 1107094999 M * Bertl don't know now .. we'll see, stick to the legacy tools for now 1107095010 M * jmp ok... 1107095023 M * jmp so may be there is data meaningfull here 1107095041 M * jmp legacy tools set up a non routable IP as Aliase on a second interface 1107095061 M * jmp on vserver I concentrate on lo and eth0 1107095107 M * jmp the fact legacy set aliases on eth1 is purely historycall 1107095112 M * Bertl yes, a few things here ... you should deconfigure any ip stuff from your vserver setup i.e. 1107095156 M * Bertl you should set IPROOT="" 1107095158 M * jmp ? manually or, you want me to change legacy to NOT set up aliases IP??? 1107095170 M * jmp Ok 1107095171 M * Bertl and IPROOTDEV= 1107095173 M * jmp 2 sec 1107095654 M * Bertl I don't think it's really relevant, but here is a patch to disable the ipv4root calls from the kernel side (http://vserver.13thfloor.at/Experimental/NGNET/delta-disable-ipv4root.diff) 1107095828 M * jmp at /proc/vnet 1107095828 M * jmp #vnet #xid real vnet dev 1107095828 M * jmp 10 49155 lo lo 0009 10 1107095829 M * jmp 11 49155 eth0 eth0 0003 5 1107095848 M * jmp NO ip aliases set by vserver.... 1107095870 M * Bertl good, now let's get a fixed XID with S_CONTEXT=20 1107095877 M * Bertl ahem S_CONTEXT=10 I meant 1107095884 M * jmp after network start on vserver, /sbin/route 1107095886 M * jmp estination Gateway Genmask Flags Metric Ref Use Iface 1107095886 M * jmp 169.254.0.0 * 255.255.0.0 U 0 0 0 eth0 1107095945 M * Bertl and could you change either the host or the vserver interface name (to make sure we do not confuse them) 1107095955 M * jmp ok 2 sec (from the top of memeory I tried to convince vserver to do it but was not succesfful (lets try again) 1107095965 M * jmp ok 1107095972 M * jmp (ok, for en0) 1107095984 M * Bertl what distro has your vserver? 1107095989 M * jmp FC3 1107096002 M * jmp (SMP) 1107096012 M * Bertl check /etc/network-scripts/ifcfg-eth0 and change the DEVICE part to en0 1107096024 M * jmp 2 sec 1107096032 M * jmp need to address vserver first 1107096047 M * Bertl (well, yes , inside the vserver I mean) 1107096113 M * jmp I set S_CONTEXT=10 but "vserver start" still asign a dynamic one... 1107096114 M * jmp 2 sec 1107096147 M * Bertl that's a bug in the tools then ... 1107096161 M * Bertl (I remember that ancient tools had this ...) 1107096172 M * jmp 2 sec (checking this right now0 1107096340 M * Bertl http://www.13thfloor.at/vserver/s_release/v1.22/ 1107096349 M * Bertl if you use Jacks tools, make sure to apply the fix01 1107096355 M * Bertl http://www.13thfloor.at/vserver/s_release/v1.22/patch-vserver-0.29-fix01.diff 1107096396 M * jmp was not aware about this (is the patch related??) 1107096417 M * Bertl well, if your tools ignore the S_CONTEXT, then yes ... 1107096433 M * jmp let me grab the diff 1107096434 M * Bertl (as I said, it's an ancient bug) 1107096467 M * jmp (working with vserver for 3 weeks now (all bugs are recent to me :-}})) 1107096504 M * Bertl hmm, why did you go for the deprecated tools then? 1107096574 M * jmp 1) Jack is a very close friend of mine (familly) 1107096600 M * jmp (such I look on its site firest) 1107096632 M * jmp 2) I like vserver.conf file (to have only one file with all meaning information) 1107096653 M * jmp anyway... let me fix vserver 1107096711 M * Bertl ad 1) great! send greetings from us if you see him ... 1107096738 M * Bertl ad 2) util vserver (stable) uses the same single file config 1107096777 M * Bertl in general: 80% of the 2.6/1.9.x features will not be available without the alpha util-vserver tools 1107096832 M * Bertl (I have no problem if jack wants to support them with new tools ;) 1107096988 M * jmp Bert.. 1107097022 M * jmp is there somewhere the full vserver with the fixe rather the diff?? 1107097059 M * Bertl well, jacks source (for that fix) is on the same page ... 1107097071 M * Bertl (the mdk rpm contains the fix) 1107097095 M * Bertl (but it's ancient too) 1107097685 J * nox ~nox@c135224.adsl.hansenet.de 1107097761 M * Bertl welcome nox! 1107097901 M * jmp Ok, fixed /usr/lib/vservers/printconf.sh 1107097909 M * jmp now context is 10 1107097915 M * jmp cat /proc/vnet 1107097915 M * jmp #vnet #xid real vnet dev 1107097915 M * jmp 10 10 lo lo 0008 1 1107097916 M * jmp 11 10 eth0 en0 0002 1 1107097950 M * jmp stil same problem with ARP 1107098004 M * Bertl okay, excellent ... let me try to reproduce this here ... 1107098051 M * jmp Bert should on vserver service network start 1107098056 M * jmp not reporting trouble??? 1107098093 M * Bertl in your mail you listed a few things where it tired to modify the /proc IIRC 1107098096 M * jmp here is reporting trouble working on sysctl.conf 1107098125 M * Bertl this doesn't make sense for now, so just ignore the errors there 1107098131 M * jmp OK (agreed) 1107098160 M * jmp please restat your last question I do not understand 1107098198 M * Bertl hmm, did I state a question? no guess it's fine ... 1107098213 M * jmp ((in your mail you listed a few things where it tired to modify the /proc IIRC))? 1107098316 M * Bertl net.ipv4.ip_forward <- I meant those (in this case sysctl) 1107098364 M * jmp Well, I suspect /etc/sysctl.conf is configured such network start try to work on /proc 1107098370 M * jmp let me check this 1107098380 M * Bertl as I said, just ignore it for now ... 1107098389 M * Bertl should not be able to do anything bad there ... 1107098462 M * jmp Bert, since we setup "en0" (as you suggested), network start is setting the network route 1107098465 M * nox thx Bertl (: (10min late) 1107098472 M * jmp 92.219.254.0 * 255.255.255.0 U 0 0 0 en0 1107098488 M * jmp by itself (so there is a litle bit of progress) 1107098499 M * Bertl s/92/192/ ? 1107098507 M * Bertl or really 92? 1107098508 M * jmp (yes, cut and past) 1107098516 M * Bertl k, np 1107098530 M * Bertl but the arp issue still remains, right? 1107098554 M * jmp the only system responding to vserver on the network is the one I foced the ARP 1107098584 M * jmp (yes, from my stand point ARP is still not reacting as it should) 1107098595 M * Bertl okay, do you ahve a server where you can upload stuff like console output and such? 1107098617 M * jmp ?? 1107098630 M * jmp do you want me to make data avaialble with FTP? 1107098636 M * Bertl (if not, just use something like pastebin.com) 1107098645 M * Bertl http actually ... 1107098671 M * Bertl it's much easier to have the data (I will require in a few moments) on my web broser ... 1107098675 M * Bertl +w 1107098676 M * jmp 2 sec looking what is pastebin.com... 1107098765 Q * nish Ping timeout: 480 seconds 1107098771 M * jmp ok, no experience on this tools, but pretty sure we can setup something, yes I have here a WEB server 1107098791 M * Bertl okay, it should be somewhat comfortable for you to put stuff there ... 1107098808 M * jmp (time to time I am using vncviewer, which seems the kind of founction you want) 1107098867 M * Bertl no, it's just avoiding to flood the channel with 100 lines of strace and dmesg output ... and I still can have it available when I need it ... 1107098896 M * jmp I can give you an ssh access on my host if you like... 1107098939 M * Bertl well, if that is what you prefer, no problem here ... (but it will have to be root on the host) 1107099009 M * jmp root is not really a problem, but I would like to understand what was the/my mistake, so lets see how 1107099021 M * jmp I can provide you with strace information 1107099050 M * jmp (what do I need to do on my side?) 1107099088 M * Bertl first we need a dump of the arp caches inside and outside, and a tcpdump of a simple ping -c 1 from inside the vserver (tcpdump done on the host, of course) 1107099157 M * Bertl ip neigh/route/addr ls (on the host and vserver) 1107099203 M * jmp where do you want me to put the info 1107099219 M * Bertl somewhere where I can reach it via http ;) 1107099244 M * Bertl (and let's keep some kind of history, i.e. use a new paste/file on new data) 1107099328 M * jmp is your E-mail 1107099349 M * Bertl yes 1107099361 M * jmp 2 sec 1107099493 M * jmp sent you an E-mail 1107099589 M * Bertl you got email too ;) 1107099744 M * jmp please try 1107099806 M * jmp 2 sec 1107099813 M * jmp seen you lock out 1107099817 M * Bertl k 1107099851 M * jmp my mistake set the wrong ip 1107099860 M * jmp try again 1107100013 M * Bertl hmm, still no luck ... 1107100046 M * jmp sorry seen the log, try again 1107100056 M * jmp 2 sec 1107100124 M * jmp Try 1107100164 M * Bertl 120x32 1107100234 M * jmp vserver = dor 1107101488 J * sannes ~ace@bjs1-dhcp275.studby.uio.no 1107101532 M * Bertl welcome sannes! 1107101634 M * sannes hi everyone :) 1107105993 Q * tchan Ping timeout: 480 seconds 1107106913 M * jmp so bertl, would you like to restat the issue? 1107106921 Q * click Read error: Connection reset by peer 1107106945 M * Bertl jmp: yes or a short overview what the question was ... 1107106994 M * Bertl simply correct me if I'm wrong ... 1107107040 M * jmp go ahead 1107107069 M * Bertl with ngnet (8.12) the atm module, more precisely net/atm/clip.c failed to compile because clip_constructor() uses inet_addr_type() which has now 2 instead of one argument 1107107117 M * Bertl this raised the question if it isn't a design flaw (of ngnet) if 'other' layers than the pure ip/networking layer need modifications ... 1107107117 J * micah ~micah@66.6.182.99 1107107122 M * Bertl welcome micah! 1107107135 M * micah wow you are quick Bertl :) 1107107139 M * jmp and my "gust feeling" we are at that stage so low in the driver we shouldn't be concerned by verserv context 1107107175 M * jmp so the question is why inet_addr_type must be aware of vserver context.... 1107107188 M * jmp what are the answer returned by inet_addr_type 1107107189 M * jmp ? 1107107207 Q * hesus Ping timeout: 480 seconds 1107107216 M * micah Bertl: I sucessfully compiled 2.6.11-rc2 with vs-1.9.4rc4, and I was just trying to compile util-vserver-0.30.196, but I get a configure error that does not make sense: checking whether to enable dietlibc... no (too old; 0.24+ required, diet version 0.27-7 found) -- isn't .27 greater than .24? 1107107222 M * Bertl jmp: now my answer so far is: looking at inet_addr_type() we see that this is a low level ip(v4?) specific function, which uses the fib hash to decide what type a given (specified) ip address is ... 1107107276 M * jmp what are the answer returned by inet_addr_type? 1107107279 M * Bertl as the fib hash is now virtualized (i.e. each context has basically a separate one), it is necessary to specify the context for inet_addr_type() 1107107304 M * daniel_hozac micah: that problem is fixed in CVS, IIRC. 1107107313 J * hesus darksoul@pingu.ii.uj.edu.pl 1107107326 M * micah daniel_hozac: ah, maybe I should be getting the stable util-vserver 1.30, instead? 1107107373 M * Bertl jmp: inet_addr_type does a few checks which are not context related (like ZERONET, BADCLASS, and MULTICAST) 1107107431 M * Bertl and some which are based on the fib_local table 1107107439 M * daniel_hozac micah: that would likely have the same problem. 1107107470 M * jmp Bertl:, could give me an example such 10.0.0.1 on vserver1 and vserver2 inet_addr_type sould be different 1107107491 M * micah daniel_hozac: apparantly it does not 1107107536 M * micah daniel_hozac: I just tried it 1107107544 M * daniel_hozac http://list.linux-vserver.org/archive/vserver/msg08635.html 1107107577 M * Bertl simple, if you do ifconfig en0 10.0.0.1 (on vserver1) and ifconfig en0 10.0.0.2 (on vserver2) then vserver1 will know the 10.0.0.1 as 'local' but the the vserver2 will not know it at all 1107107577 J * tchan ~tchan@c-24-13-81-164.client.comcast.net 1107107599 M * micah daniel_hozac: or is it much better to have the newer version of the utils? 1107107627 M * daniel_hozac micah: if you're using 2.6/1.9, you are pretty much required to have them. 1107107666 M * Bertl jmp: so vserver1 will return RT_LOCAL, but vserver2 will return RTN_UNICAST 1107107691 M * micah daniel_hozac: the 0.30 stable isn't good enough?I need to have 0.30.196? 1107107710 M * daniel_hozac micah: 0.30 doesn't support all the new features, so yes. 1107107719 M * jmp Ok... 1107107729 M * micah daniel_hozac: hmm, how do I get the CVS version? 1107107751 M * Bertl jmp: or precisely, 'would be supposed to return ...' 1107107757 M * daniel_hozac micah: did you read Enrico's email (and the follow-ups)? 1107107774 M * micah daniel_hozac: ah, no... I will do so 1107107868 M * micah hmm it does not talk about how to get it via CVS 1107107882 M * Bertl savannah cvs that is 1107107903 M * jmp Bertl: it is good enough for me 1107107941 M * jmp so everytime inet_addr_type is not right in the Kernle code, we should align it no problem? 1107108010 M * Bertl basically yes, it needs some kind of investigation, if the 'context' is present (or can be deduced from the routing/flow/device data) and pass that to the low level calls 1107108020 M * jmp ok... 1107108053 M * Bertl if that is done properly, it should be a proper fix to allow the module/code to work in _and_ outside vservers 1107108072 M * Bertl if you 'just' want to make it work on the host, passing '0' would be sufficient 1107108114 M * Bertl (but that could, and probably would, lead to strange effects if used inside) 1107108129 M * jmp dev->nfxid is zero by default when running on the host itself? 1107108133 M * jmp right? 1107108150 M * Bertl when the device comes from the host, that is ... 1107108159 M * jmp ok... 1107108190 M * jmp about the ARP, I seen some "magic" done on the test bench... 1107108218 M * jmp but I didn't take note of the detail... 1107108230 M * jmp could you refresh my mind? 1107108246 Q * micah Quit: leaving 1107108251 M * Bertl I added the no-arp-for-vserver issue to the todo list, for now it basically boils down to, you have to anser the arp requests on the host to make the vserver ips reachable ... 1107108265 M * Bertl *answer even 1107108331 M * Bertl so for example if vserver1 uses 10.0.0.1 and vserver2 uses 10.0.0.2 you could do an ip addr add 10.0.0.1 dev eth0 on the host (ans same for 10.0.0.2) to make it visible 1107108370 M * jmp THis quit acceptable from my stand point 1107108377 M * Bertl okay, so let's attack the next issue, namely the strange 127.x.x.x one then 1107108533 M * jmp how come seems to be working now? 1107108547 M * Bertl give it some time, caches will time out ... 1107108555 M * Bertl or try with a different ip ... 1107108598 M * Bertl hmm 1107108606 M * Bertl maybe some kernel config option then? 1107108608 M * jmp As you are saying 1107108635 M * Bertl can we try with a reboot/vserver restart? 1107108653 M * jmp yes 1107108661 M * jmp let me proceed 1107108882 M * jmp so the problem is here again... 1107108898 M * jmp something you enter previously did the "trick" 1107109336 M * jmp Bertl: I found something 1107109742 J * DuckMaster ~Duck@dyn-83-157-198-14.ppp.tiscali.fr 1107109966 M * Bertl jmp: yep? 1107110079 M * Bertl welcome DuckMaster! 1107110138 Q * DuckKing Ping timeout: 480 seconds 1107110169 M * sannes :) 1107110175 M * jmp seems when I "/sbin/route add -net 127.0.0.0/8 dev lo" on vserver , host is not responding 1107110187 M * jmp not everytime unfortunatly 1107110249 M * Bertl hmm, okay, that sounds like another report which said that 'host' lo stops working (i.e. is taken down' when the vservers lo goes up or so ... 1107110449 M * jmp Bertl, how come on the vserver a ping return invalide argument for a specific IP (the other are working) 1107110473 M * Bertl hmm, let's see ... 1107110573 Q * tchan Ping timeout: 480 seconds 1107110676 M * jmp working now?!! 1107110685 M * jmp what have you done?! 1107110743 M * Bertl hmm, looks to me like related to the arp issue ... 1107110792 M * jmp I would like to reboot the whole system and see if I can get it working as weel it is working now.... 1107110812 M * jmp do you still suggest to boot with the ipv4root patch? 1107110831 M * Bertl yeah, it won't hurt and avoid possible issues with ipv4root 1107110862 M * jmp ok, so you give me the clear to reboot the server 1107110874 M * Bertl sure, whenever you're ready ... 1107110880 M * jmp (if you want to save someting, take your time... 1107111848 M * jmp Bertl, I rebooted, still working :-}} , still shaky 1107111882 M * jmp neet to ip addr add x.y.x.t dev eth0 scope global 1107111906 M * jmp to satisfied ARP, and then lost the lo interface on route table on host server 1107111951 M * jmp but from my stand point, it is willing to work as I need it! Bravo!! 1107112420 M * Bertl hmm, thanks! 1107112449 M * Bertl jmp: when is the lo route lost on the host? 1107112509 M * jmp according to me when the command p addr add x.y.x.t dev eth0 scope global is 1107112511 M * jmp issued the 1107112525 M * jmp lo definition is lost from /sbin/route display 1107112546 M * jmp (is my explainantion good enough, I can test again...) 1107112800 M * Bertl hmm, maybe you could do a before and after for me? 1107112930 M * jmp 2 sec 1107112954 J * axu_ ~Axu@81.223.125.177 1107112969 M * axu_ good evening folks :) 1107113047 M * Bertl evening axu_! 1107113067 M * axu_ hi bertl :) 1107113195 M * jmp I can't reproduce it (but it was after a reboot), I am rather positive 1107113196 M * jmp 27.0.0.0 * 255.0.0.0 U 0 0 0 lo 1107113205 M * jmp was missing from /sbin/route 1107113223 M * jmp just after I have done ip addr add x.y.x.t dev eth0 scope global 1107113233 M * jmp (I do not like thigs like that 1107113234 M * jmp ) 1107113257 J * nish ~nish@220.224.6.132 1107113275 M * nish Hi :) 1107113294 M * axu_ hi nish 1107113341 M * axu_ mal X neu starten 1107113348 M * axu_ sorry :) 1107113353 M * axu_ brb 1107113364 Q * axu_ Remote host closed the connection 1107113757 M * jmp bert: just an hint, the new iptable (the one working with vnet), seems not really working with init.d/iptables script 1107113773 M * jmp problem seem mainly at iptables-restore /etc/sysconfig/iptables 1107113783 M * jmp which is sending iptables-restore: target `LOG' v1.2.9 (I'm v1.2.11). 1107113791 M * jmp does this means something? 1107113851 M * Bertl hmm, sure that that target is compiled/enabled/loaded? 1107113884 M * Bertl and why are you using v1.2.9 iptables save/restore when the patched one is 1.2.11? 1107113886 M * jmp hmmmm.... NOT 100% sure about this 1107113893 M * jmp :-}}} 1107113925 M * jmp I took the iptable from EXPERIMENTAL 1107114028 M * Bertl the patch I guess, or? 1107114189 M * jmp iptables 05-Dec-2004 20:20 50K 1107114234 M * Bertl hmm, didn't even remember I uploaded the binary ;) 1107114256 M * Bertl well, it would be advised you recompile your own iptables (and friends) 1107114272 M * Bertl after adding the appropriate patch 1107114294 M * jmp let double check all this 1107114357 M * Bertl (so if your distro uses 1.2.9 then get the source package/rpm and add the 1.2.9 patch to it, change the version, recompile and install it) 1107114514 M * jmp distro say iptables-1.2.11-3.1.FC3 1107114540 M * jmp let me check... 1107114551 Q * nish Ping timeout: 480 seconds 1107114554 M * Bertl okay, so get the source rpm for that, add the http://vserver.13thfloor.at/Experimental/NGNET/iptables-1.2.11-vnet.patch 1107114556 M * jmp I am not 100% sure used kernel is LOG alloed 1107114581 J * nish ~nish@220.224.31.74 1107114586 M * Bertl wb nish! 1107115852 J * we2by ~jinxi@145.116.6.75 1107115906 M * Bertl welcome we2by! 1107115954 M * we2by ty Bertl 1107116184 M * jmp bertl: FYI: net/ipv4/netfilter/ipt_addrtype.c (return !!(mask & (1 << inet_addr_type(addr))); 1107116214 M * Bertl ah, another candidate ;) 1107116246 M * Bertl jmp: actually I'm investigating a tcp issue which micht be related to one of the 127.0.0.x effects you saw ... 1107116356 M * Bertl in the netfilter case, I'd go with the nfxid from the skb itself 1107116490 M * Bertl (added a patch to my source, will be in 8.13 1107116584 J * MairaV ~maira@115-231-114-200.fibertel.com.ar 1107116587 M * MairaV Hi 1107116597 M * MairaV Hello - Hola 1107116603 M * Bertl welcome MairaV! 1107116606 M * MairaV ; 1107116608 M * MairaV ;) 1107116611 M * MairaV Thanks. 1107116629 M * MairaV It is the first time with vserver. 1107116643 M * MairaV I will try to install it in a FC2 Test Server 1107116663 M * MairaV Sorry for my bad english, my natal language is spanish 1107116690 M * Bertl well, most folks are not native english speakers either ... 1107116691 M * MairaV You have any easy how-to for install it? 1107116731 M * MairaV =) 1107116754 M * Bertl get a vanilla kernel, your favorite patch, apply it, configure it, compile and isntall it, then reboot, get the tools, compile and isntall them, that's it ... 1107116787 M * MairaV FC2 have kernel 2.6, is it possible run the vserver with the 2.6 kernel? 1107116801 M * Bertl yes, but not with the FC2 2.6 kernel 1107116816 M * daniel_hozac well, with the right patch.. ;) 1107116821 M * MairaV Yeah 1107116841 M * Bertl daniel_hozac: not even with the right patch, you need at least to modify the kernel *G* 1107116843 M * MairaV I will download the source rpm and apply the patch, really? 1107116879 P * jmp 1107116902 M * daniel_hozac Bertl: not too much though :) 1107116912 M * MairaV ?? 1107116940 M * Bertl MairaV: I'd suggest using a vanilla kernel, jsut to make sure 1107116956 M * MairaV what is vanilla kernel? 1107116972 M * Bertl one of the declared aims of Redhat/FC is to support mainline (aka. vanilla) kernels 1107116982 M * MairaV any link? 1107116997 M * Bertl www.kernel.org (has the kernels) 1107117040 M * Bertl <- here are the latest linux-vserver patches 1107117045 M * Bertl hrm ... 1107117054 M * Bertl http://vserver.13thfloor.at/Experimental/RC-1.9.4/ 1107117081 M * MairaV see it 1107117081 M * daniel_hozac http://rpm.hozac.com/fedora/dhozac/3/i386/vserver/ for RPMs ;) 1107117087 M * MairaV =) 1107117090 M * MairaV Ohh good 1107117096 M * Bertl daniel_hozac: guess you win ;) 1107117097 M * MairaV All done :) 1107117135 Q * nish Ping timeout: 480 seconds 1107117142 J * nish ~nish@220.224.6.45 1107117154 M * Bertl wb new nish! 1107117200 M * MairaV This kernerls, have the patchs? 1107117220 M * daniel_hozac or i screwed up when building them :) 1107117221 Q * nish Read error: Connection reset by peer 1107117254 M * MairaV is it possible download it from yum or apt respository? 1107117274 J * mbucc ~konversat@host-216-153-147-194.spr.choiceone.net 1107117544 M * MairaV Yeah =) http://rpm.hozac.com/fedora/yum.conf 1107117639 M * daniel_hozac just add the http://rpm.hozac.com/fedora/dhozac/2/i386/vserver/ entry. 1107117730 Q * serving Ping timeout: 480 seconds 1107117762 J * CristianM ~CristianM@115-231-114-200.fibertel.com.ar 1107117772 Q * MairaV Quit: 1107117781 M * CristianM Hi 1107117795 M * CristianM Sorry, i changed the Nick for CristianM... 1107117834 M * daniel_hozac you know you can do that without reconnecting, right? ;) 1107117876 M * CristianM no 1107117882 M * CristianM The old nick are MairaV 1107117896 M * CristianM I changed the IRC client 1107117903 M * CristianM =) 1107117905 M * daniel_hozac ok. 1107117975 M * CristianM I will download http://rpm.hozac.com/fedora/dhozac/2/i386/vserver/kernel-2.6.9-1.667.vs1.9.3.2.i686.rpm 1107117995 M * daniel_hozac that'd be an old kernel. http://rpm.hozac.com/fedora/dhozac/2/i386/vserver/kernel-2.6.10-1.741.vs1.9.4.0.rc3.1.i686.rpm would be better. 1107118013 M * CristianM ok. 1107118082 M * Bertl daniel_hozac: still catching up, eh? 1.9.4-rc4 is out! 1107118110 M * daniel_hozac what changed between rc3 and rc4? 1107118127 M * daniel_hozac i thought it was just relevant for 2.6.11-rc*. 1107118191 M * Bertl there is one relevant patch in that ... 1107118210 M * CristianM Which rpms i need download? 1107118219 M * Bertl @@ -667,7 +667,7 @@ static int do_lookup(struct nameidata *n 1107118219 M * Bertl inode = dentry->d_inode; 1107118219 M * Bertl if (!inode) 1107118219 M * Bertl goto done; 1107118219 M * Bertl - if (!vx_check(inode->i_xid, VX_WATCH|VX_HOSTID|VX_IDENT)) 1107118221 M * Bertl + if (!vx_check(inode->i_xid, VX_WATCH|VX_ADMIN|VX_HOSTID|VX_IDENT)) 1107118224 M * Bertl goto hidden; 1107118226 M * Bertl if (inode->i_sb->s_magic == PROC_SUPER_MAGIC) { 1107118229 M * Bertl struct proc_dir_entry *de = PDE(inode); 1107118240 M * daniel_hozac CristianM: you'd probably want util-vserver-* as well. 1107118256 M * CristianM Ok. Thanks 1107118273 M * CristianM Exist any control panel for vserver? 1107118288 M * daniel_hozac Bertl: ah, right. sorry, forgot about that one. 1107118295 M * we2by CristianM, nope 1107118305 M * we2by as far as I am aware 1107118316 M * Bertl hmm, wasn't there somebody working on something ;) 1107118318 M * we2by I don't see why u need a cp 1107118333 M * daniel_hozac Bertl: that's fix01 in RC-1.9.4, right? 1107118335 M * CristianM create, delete or modify vps 1107118344 J * tchan ~tchan@c-24-13-81-164.client.comcast.net 1107118345 Q * tchan Quit: 1107118356 J * tchan ~tchan@c-24-13-81-164.client.comcast.net 1107118358 M * Bertl daniel_hozac: http://vserver.13thfloor.at/Experimental/RC-1.9.4/delta-xid-fix01.diff 1107118369 M * daniel_hozac Bertl: right, thanks. 1107118706 M * Bertl you're welcome, btw, this might be intersting too: 1107118716 M * Bertl http://vserver.13thfloor.at/Experimental/RC-1.9.4/delta-assert-fix01.diff 1107118748 M * Bertl (it's in 2.6.11-rc2, but should work for 2.6.10 or older too) 1107118780 M * daniel_hozac but that's only relevant when spinlock debugging is enabled, right? 1107118801 M * Bertl yep, which probably isn't true for the FC kernels, right? 1107118828 M * daniel_hozac crap, it is. :| 1107119110 M * daniel_hozac thanks, i'll have to apply it. 1107119123 M * Bertl no problem ... 1107119641 M * CristianM I am installed all the rpms 1107119651 M * CristianM and now? which is the next step? 1107119679 M * daniel_hozac vserver - build --help 1107119685 M * CristianM yeah 1107119694 M * CristianM reboot the server and try 1107119891 Q * tchan Ping timeout: 480 seconds 1107120346 Q * sannes Ping timeout: 480 seconds 1107120357 M * CristianM daniel: any doubt, the cq-tools, are included in the rpms? 1107120395 M * daniel_hozac is that for per context disk limits? 1107120404 M * daniel_hozac (as in, not quota?) 1107120469 M * Bertl nope, vdlimit are the disk limit tools ... 1107120534 M * daniel_hozac yeah... so what is cq-tools for? 1107120576 M * Bertl per context quota tools for 2.4 ... 1107120605 M * daniel_hozac that's what i thought. 1107120650 Q * mbucc Ping timeout: 480 seconds 1107120911 M * CristianM and is it possible to install via rpm? 1107120926 M * daniel_hozac as Bertl said, cq-tools is for 2.4. 1107120981 M * Bertl to clarify: there is no shared quota support for 2.6 yet ... 1107121038 M * CristianM ok 1107121680 M * Bertl okay, back later ... 1107121688 N * Bertl Bertl_oO 1107121839 Q * CristianM Quit: • IRcap [7.51] • http://www.ircap.net • 1107124808 Q * we2by Remote host closed the connection 1107125982 T * * http://linux-vserver.org/ | latest stable 1.2.10, devel 1.9.{3,4-rc4}, ng8.12 1107125982 T * Bertl - 1107126134 J * we2by ~jinxi@145.116.6.75 1107126575 Q * SiD3WiNDR Ping timeout: 480 seconds 1107126939 Q * we2by Remote host closed the connection 1107126973 J * we2by ~jinxi@145.116.6.75 1107129338 J * tchan ~tchan@c-24-13-81-164.client.comcast.net 1107129349 J * serving ~serving@213.186.187.9