1109462587 M * Doener guess i shouldn't try to work on anything serious tonight, will tackle networking once more when i'm sober again 1109462592 M * Doener night folks, Bertl! 1109462605 N * Doener Doener_zZz 1109462608 M * Bertl night! 1109462716 Q * DaPhreak Server closed connection 1109462761 J * DaPhreak ~DaPhreak@pc-outside.uni-greifswald.de 1109463027 J * lilo ~lilo@lilo.usercloak.oftc.net 1109463040 M * Bertl wb lilo! 1109463216 Q * eyck Ping timeout: 480 seconds 1109463422 M * sladen hello Bertl (and greetings from FOSDEM) 1109463454 M * Bertl hey paul, everything fine? 1109463503 M * DaCa sladen: do you have any idea at which time the keysigning is tomorrow, their site is down _again_ 1109463623 M * sladen DaCa: not sure, but somewone here is just printing out and cutting up slips :) 1109463643 M * sladen DaCa: I have no idea how or when it's happening. I might just end up ''signing on the fringe'' 1109463685 M * prae sladen: you are at fosdem ? 1109463803 M * sladen prae: not at this minute (I'm sleeping ;-) but yes, today and tomorrow---are you there? 1109463893 M * prae no, but friends, yes :) 1109463921 M * DaCa sladen: did you come by brompton? :) 1109464027 M * prae they are going directly at fosdem this friday 1109464049 M * sladen DaCa: how did you know? Yup, I have the Brompton---it was locked-up (folder) in the room next to reception 1109464066 M * sladen DaCa: though I saw 2-3 others (one belonging to my housemate from London) 1109464411 M * DaCa sladen: I know your name from bromptontalk, I am a bromptoneer too, you might remember seeing a brompton at debconf3, that was mine :) 1109464621 J * eyck eyck@81.219.64.71 1109464629 M * Bertl wb eyck! 1109464765 M * sladen DaCa: ahhhhh! 1109464788 M * sladen DaCa: yup, before I got mine, although I'd be wanting one for a couple of years then 1109465801 Q * prae Quit: Pwet 1109466052 J * prae ~prae@sherpadown.net 1109466151 Q * eyck Ping timeout: 480 seconds 1109466216 M * prae root@carbone:~# /usr/local/util-vserver//sbin/vserver-info - VERIFYCAP 1109466216 M * prae capset(): Operation not permitted 1109466216 M * prae root@carbone:~# zgrep CAPAB /proc/config.gz 1109466216 M * prae # CONFIG_SECURITY_CAPABILITIES is not set 1109466218 M * prae ?!!! 1109466285 M * Bertl hmm? 1109466351 M * prae hmm... mistake 1109466377 M * Bertl good that we have that check, thanks Enrico! 1109466395 M * prae I must be tired .. 1109466499 M * prae it is a good idea to enable Proc Security ? 1109466515 M * Bertl yes, usually it is ;) 1109466526 M * prae :p 1109466537 M * daniel_hozac why isn't enrico here any more? 1109466568 M * Bertl I have no idea, ... didn't reply to my last email either ... 1109466596 M * Bertl (maybe still? busy with his thesis?) 1109466605 M * prae *'be back* 1109466608 Q * prae Quit: Pwet 1109466640 M * daniel_hozac he's still on #fedora-devel on Freenode. 1109466668 M * Bertl hmm? could you contact him there? 1109466697 M * daniel_hozac well, he's not on right now. i've seen him join every now and then though. 1109466727 M * Bertl joining or saying something? 1109466741 M * daniel_hozac just joining. 1109466768 M * Bertl well, he _was_ joining and leaving some time ago here too ... (didn't say anything though) 1109466797 M * daniel_hozac ah, i must've missed that. 1109467561 J * eyck eyck@81.219.64.71 1109469385 M * jd86 can someone point me to where rlimits and or ulimits are described so i can limit mem and cpu usage of a vserver? 1109469410 M * Bertl ulimits are described in 'man bash' 1109469427 M * Bertl rlimits in 'man setrlimit' 1109469457 M * Bertl but most likely for 'cpu' limiting you want to look at the prio or hard token bucket scheduler ... 1109469640 M * jd86 oh 1109469641 M * jd86 my 1109469649 M * jd86 can you help me with that or tell me where to read? 1109469661 M * Bertl the tb scheduler stuff? 1109469663 M * jd86 i want to limit how much of the cpu they can attack so one of them cant hog it all and how much total memory they can use 1109469695 M * Bertl http://linux-vserver.org/Linux-VServer-Paper-06 1109469705 M * Bertl (06.3 explains the method) 1109469736 M * DaCa Bertl: is the tb scheduler only available on the 2.6 series? 1109469761 M * jd86 ok now how do i use the method? 1109469816 M * Bertl DaCa: basically, there was an early version in 1.3.9 ck ... 1109469827 Q * eyck Ping timeout: 480 seconds 1109469838 M * Bertl jd86: you have to enable the hard scheduler (kernel config) 1109469850 M * jd86 Bertl: ok, and then? 1109469851 M * Bertl then you set the sched_hard flag for the vserver context 1109469884 M * Bertl then you configure the tb parameters (a 25% setup should be the default) 1109469906 M * jd86 oh my 1109469911 M * jd86 we need to talk like specifics here 1109469917 M * jd86 i dont mind reading a paper or something on how to do it 1109469918 M * jd86 but 1109469927 M * jd86 whats this vserver context you are talking about? 1109469938 M * jd86 and where do i configure the tb parameters 1109469942 M * Bertl hmm, you sure you _read_ the paper? 1109469953 M * jd86 the long paper thing? i started to but it was all quite inspecific 1109469961 M * jd86 or the thing u sent me? 1109469963 M * jd86 i read that the whole way 1109469971 M * jd86 the sec 06.3 1109469988 M * Bertl http://linux-vserver.org/Linux-VServer-Paper (read the whole thing) 1109470074 M * jd86 alright 1109470079 M * jd86 i had read alot of it 1109470092 M * jd86 but found what i read to be very informative but not telling me how to do much 1109470114 M * Bertl so then you should know what that 'vserver context' is ;) 1109470156 M * Bertl do you have a vserver up and running? 1109470190 M * jd86 yea 1109470191 M * jd86 3 of them 1109470223 M * Bertl okay, so do the first steps (enabling HARD scheduler in kernel and setting the sched_hard flag for them) 1109470253 M * Bertl then get a nice a comfy cpuhog(tm) and let's do some hands-on testing ;) 1109470262 M * jd86 Bertl: so we are talking recompiling kernel? 1109470279 M * Bertl don't know, maybe you've already enabled it? 1109470291 M * jd86 where would it be found in the kernel config 1109470310 M * jd86 are we refering to Enable Hard CPU limits? 1109470318 M * Bertl yep, precisely 1109470327 M * jd86 it is not enabled i'll recompile the kernel with it on 1109470331 M * jd86 anything else that i shoudl have enabled? 1109470340 M * jd86 should i limit the idle task? 1109470349 M * Bertl that's usually not required 1109470353 M * jd86 alright 1109470361 M * jd86 its compiling 1109470442 M * jd86 oh, and what should i have for quotas? 1109470462 M * Bertl quotas? 1109470484 M * jd86 so that each vserver can only use so much disk space 1109470505 M * Bertl a disk limits ;) 1109470524 M * Bertl you need to enable xid tagging (default is on 24/24 taggin) 1109470533 M * jd86 another thing in the kernel? 1109470540 M * Bertl yes 1109470548 M * jd86 where would that be residing 1109470552 M * Bertl (but as I said, it's probably already on) 1109470561 M * jd86 i'd like to check 1109470604 M * jd86 its on 24/24 for the persisitant inode contexttagging is that right? 1109470614 M * Bertl yep 1109470654 M * jd86 let me reboot that box with new kernel now 1109470674 M * Bertl k 1109470755 M * jd86 ok, its rebooted 1109470828 M * jd86 what shal i do now? 1109470847 M * Bertl second step, add the sched_hard flag to one of your vservers 1109470879 M * jd86 ok when you say to one of my vservers, you mean the vserver.conf file? 1109470948 M * Bertl hmm, you are using the legacy config format? 1109470984 M * jd86 well one of my vservers has a veserver.conf and others a group of folders and files 1109470987 Q * matta Server closed connection 1109470991 J * matta ~matta@69.93.28.254 1109471003 M * Bertl okay, then take one with folders and files ;) 1109471014 M * jd86 one i followed one how to to setup and the other i followed another. (still utterly confused about alot but thats alright) 1109471015 M * jd86 alright 1109471021 M * Bertl add the sched_hard to the flags 1109471039 M * jd86 done. 1109471058 M * Bertl restart that vserver and add a nasty process inside ... 1109471100 M * jd86 whats a nasty process ? 1109471114 M * jd86 gimme an example 1109471116 M * Bertl well, a process which uses up all your cpu(s) 1109471124 M * jd86 yea, dunno one actually 1109471126 M * jd86 hrm 1109471149 M * Bertl http://vserver.13thfloor.at/Experimental/TOOLS/cpuhog.c 1109471150 M * jd86 got any good examples? 1109471162 M * Bertl (this one is trained ;) 1109471237 M * jd86 wow, of coarse this doesnt have gcc on it (:() 1109471281 M * Bertl if you prefer a precompiled binary I can upload my backdoor^Wcompiled version ;) 1109471289 M * jd86 naw 1109471291 M * jd86 i compiled it 1109471300 M * jd86 with host sys (what do you call the main os running? host?) 1109471312 M * Bertl yes, host and guest 1109471325 M * jd86 its running 1109471341 M * Bertl and the 'guest' (vserver or vps) has no OS of it's own 1109471351 M * zimdog Hello All. I have installed the base redhat9 vserver form the website and need the kernel-sources to compile an application. How is this done in a vserver? 1109471366 M * jd86 25% of the cpu is being used 1109471373 M * jd86 now how would one change such a 25% 1109471377 M * Bertl zimdog: well, just install it ... 1109471407 M * Bertl jd86: look at /proc/virtual//sched 1109471440 M * jd86 look at what part of it? 1109471470 M * jd86 its very pretty looking ;-) 1109471501 M * Bertl you read the paper, ehre you see example values for the theory ... 1109471522 M * jd86 where do you set it tho? like in what config file, i guess i'll head over to the paper 1109471522 M * Bertl changing those (with vsched) will change the behaviour ... 1109471532 M * jd86 ok what happens when the vserver resets 1109471540 M * jd86 or the host reboots 1109471545 M * Bertl you can put the 'start' values into the config of course 1109471611 M * jd86 could you hint to as where in this paper the examples are? 1109471630 M * jd86 ah! maybe the example section 1109471682 M * jd86 To be completed. 1109471715 M * zimdog Do I need a vserver-src rpm to match the kernel 2.4.29-vs1.2.10 on the vserver? Is a patched source code available 1109471717 M * jd86 thats what it says 1109471799 M * Bertl zimdog: no, just take the kernel sources from the distro 1109471815 M * Bertl the vserver will not be able to do much with the vserver kernel anyways ... 1109471837 M * Bertl (i.e. all kernel source requirements are not really valid) 1109471858 M * zimdog I see. Thanks 1109471913 M * jd86 Bertl: i dont see where you are talking about 1109472039 M * Bertl you did look at /proc/virtual//sched right? 1109472070 M * jd86 yes 1109472071 M * jd86 i did 1109472083 M * Bertl that is an example for achieving 25% right? 1109472083 Q * Zoiah Ping timeout: 480 seconds 1109472088 M * jd86 yes. 1109472123 M * Bertl so combine that with the 'theoretical' knowledge from the paper and you can change it to whatever you want, right? 1109472164 M * jd86 but i A. dont know how to set it, B. dont know what any of this stuff and waaa :( 1109472204 M * Bertl so what do you expect from me? 1109472214 M * Bertl another example maybe? 1109472227 M * jd86 i'd like an example. not another 1109472230 M * jd86 lol 1109472234 M * jd86 where in the paper is the example? 1109472252 M * Bertl its in you proc ;) 1109472264 M * jd86 ok can i have another example? 1109472279 M * jd86 and how do i _change_ those values? 1109472304 M * jd86 (i'm thinking this is as much fun as interpeting shakespere) 1109472321 M * Bertl FillRate: 45 1109472321 M * Bertl Interval: 100 1109472321 M * Bertl TokensMin: 3 1109472321 M * Bertl TokensMax: 500 1109472334 M * Bertl ah, yes I love shakespere! ,) 1109472345 M * jd86 no wonder you excell at vservers ;-) 1109472363 M * jd86 should be the motto for vservers You get shakespere? you will understand vservers! 1109472378 M * jd86 at least shakespere has parallel texts. 1109472387 M * Bertl *shakespeare btw ;) 1109472390 M * jd86 yea 1109472402 M * jd86 i am horrible at spelling, english and now cpu limits. 1109472410 M * jd86 can you give me a hint as to how to set these limits? 1109472423 M * Bertl vsched 1109472451 M * jd86 dev-box 49153 # man vsched 1109472451 M * jd86 No manual entry for vsched ; me cries. 1109472463 M * Bertl try 'vsched --help' 1109472480 M * jd86 ahh 5 lines of goodness. 1109472536 M * jd86 dev-box 49153 # vsched --xid 49153 --fill-rate 45 --interval 100 --tokens-min 3 --tokens-max 500 1109472536 M * jd86 Segmentation fault 1109472547 M * Bertl looks like older tools 1109472561 M * jd86 vsched 0.30.196 1109472567 M * Bertl (should be fixed in 0.30.204) 1109472574 M * jd86 ugh 1109472579 M * jd86 how do i get 0.30.204 1109472583 M * Bertl but you can workaround that 1109472592 M * jd86 should i just get new tools? 1109472595 M * Bertl by just adding '-- true' 1109472606 M * jd86 aha it didnt complain 1109472629 M * jd86 now it uses 95% 1109472640 M * jd86 (at least i know how to set it so i can play with it even knowing i wont know what i'm changing) 1109472723 M * Bertl that's the theory part I've pointed to multiple times ... 1109472728 M * jd86 do i have to restart the vserver? 1109472748 M * Bertl I can explain it to you if you do not understand ait or have specific questions ... 1109472758 M * Bertl no vsched works immediately 1109472835 M * jd86 i read it twice, i get it more now but still dont know exactly how it works, i'll have to play with it some more i guess. 1109472860 M * jd86 seems like whatever i do now, the process uses 96% 1109472889 M * Bertl check the values with the proc ... 1109472964 M * jd86 ok so the fill rate seems to directly effect the cpu usage 1109472979 M * Bertl as does the interval ;) 1109472995 M * jd86 so its like a big guessing game 1109473013 M * Bertl no, it's a pretty simple model ... 1109473033 M * jd86 maybe i'm stupid then :( 1109473057 M * Bertl okay, get you favorite bucket and a few tokens, I'll try to explain ... 1109473074 M * Bertl ah, you'll need a stop watch too ;) 1109473114 M * jd86 call me stupid if i'm wrong but, is it kind of like giving a kid tokens to play a video game and the interval is how often you give the kid tokens to use, and the rate is how many per time and the more the kid uses them the less stuff he can do lesser cpu, and if the interval is lessened then the kid gets them faster and he can use more cpu.. or am i totally wrong 1109473142 M * Bertl no, you're exactly right! 1109473172 M * jd86 and the kid can accomulate up to the max tokens and have only that minimum token meaning he can always use some percentage 1109473173 M * jd86 right? 1109473218 M * jd86 or no? 1109473222 M * Bertl the max is correct, the min is what he needs to get, once he spent all the tokens to play another round ;) 1109473231 M * jd86 ooooh 1109473238 M * jd86 ah so thats how romeo and juliet ended ;-) (heh) 1109473245 M * jd86 aah thanks alot for putting up with me 1109473258 M * Bertl happily ever after ... *G* 1109473260 M * jd86 heh 1109473268 M * Bertl you're welcome, btw! 1109473281 M * jd86 now, on to memory limiting! 1109473288 M * jd86 only if you're up to it 1109473299 M * Bertl ah, I'm already enjoying it ... 1109473308 M * jd86 oh and how do you save those settings in config files? 1109473335 M * Bertl the flower page has the details ... 1109473344 M * jd86 oh my 1109473346 M * jd86 back to the flower page. 1109473347 M * jd86 alright 1109473358 M * Bertl IIRC it's called scheduler ... 1109473362 M * jd86 alright 1109473370 M * jd86 wait 1109473372 M * jd86 flower page for the memory 1109473374 M * jd86 or the saving settings 1109473384 M * Bertl savin the settings ... 1109473388 M * jd86 alright 1109473394 M * jd86 i'll look at the flowers later for that info 1109473399 M * jd86 i want to get this memory thing sorted out 1109473405 M * Bertl k 1109473428 M * jd86 so how do i limit max memory for a vserver? 1109473434 M * Bertl look at /proc/virtual//limits 1109473456 M * jd86 i have a "limit" but no limits. 1109473462 M * Bertl limit sorry 1109473465 M * jd86 no prob 1109473471 M * jd86 just wanted to verify we are talking about the same thing 1109473483 M * jd86 alright i see it 1109473492 M * jd86 what are the column headings ? 1109473504 M * Bertl okay, VM, VML and RSS are memory related and the columns are 1109473529 M * Bertl first: limit, second current, third max observed, 4th limit, 5th hits 1109473540 M * jd86 sounds greek to me. 1109473553 M * jd86 so whats VM VML and RSS stand for? anything intersting? 1109473584 M * Bertl VM is the 'virtual memory' or address space 1109473595 M * Bertl VML are the number of pages locked into memory 1109473613 M * Bertl and RSS is the resident set size (or the amount of pages swapped in) 1109473618 M * jd86 me thinks this might be worse then the cpu limiting. 1109473631 M * jd86 let me take a wild guess and say i need to work with RSS eh? 1109473668 M * Bertl in a first approximation you can see the VM as ram+swap and the RSS as ram but that's not entirely correct 1109473715 M * Bertl as linux vserver is based on sharing resources, it's not easy to separate them in accounting 1109473728 M * jd86 RSS: 1943 3679 -1 0 whats the 0 and the -1 and what are these columns all about you told me the headings but i kinda dont get it 1109473758 M * Bertl okay, consider a process which requires some resources 1109473772 M * jd86 mysqld? 1109473784 M * Bertl let's say it allocates fir 1MB then another 2MB after that it releases 1.5MB okay? 1109473790 M * Bertl *first 1109473800 M * jd86 ok 1109473823 M * Bertl the first (number) colum will show 1.5M (in pages) 1109473841 M * Bertl the second will show 3MB (again inpages) 1109473856 M * jd86 ok 1109473859 M * Bertl the third will show your limit (currently unlimited) 1109473871 M * jd86 and forth? 1109473874 M * Bertl and the last will show how often that limit was hit 1109473884 M * jd86 hrm intersting. 1109473901 M * jd86 how do you convert pages > mb? 1109473916 M * Bertl the pagesize is 4k on x86 1109473947 M * jd86 so 1943 is ~7.5 mb? 1109473983 M * Bertl more 7.6 but yes 1109473994 M * jd86 well 7.58 1109474000 M * Bertl 7.5898437500 1109474017 M * jd86 ah xcalc rounded to 7.5898438 1109474027 M * jd86 moving on... 1109474159 M * jd86 now is this rlimits or ulimits? 1109474184 M * Bertl rlimits 1109474238 M * jd86 so if i wanted my fedora vserver to have a 25mb limit it'd be 102400 and i'd put that in /etc/vservers/fedora-template/rlimits/rss and restart vserver? 1109474279 M * Bertl 102400 is about 400MB 1109474284 M * jd86 gah 1109474287 M * jd86 where did i mess up 1109474296 M * daniel_hozac Bertl: what are vdlimit flags used for? 1109474322 M * jd86 /usr/lib/util-vserver/vserver.functions: line 646: popd: /proc/virtual/49153: No such file or directory i did a vserver fedora-template restart and i got funky errors. 1109474330 M * Bertl daniel_hozac: unused atm 1109474360 M * jd86 gah its borked. 1109474368 M * Bertl jd86: you should move away from dynamic contexts 1109474376 M * jd86 dynamic contexts? explain? 1109474393 M * Bertl if you do not specify a context id, it will be dynamically assigned 1109474395 J * monrad ~monrad@213083190130.sonofon.dk 1109474397 M * jd86 ooh 1109474400 M * jd86 i should specify one 1109474402 M * Bertl 49153 is the second dynamic context 1109474421 M * monrad evening / morning 1109474423 M * Bertl static context values range from 2- 49151 1109474427 M * Bertl morning monrad! 1109474430 M * jd86 where would i set it? 1109474437 M * Bertl *flower-page 1109474481 M * jd86 it says context 1109474487 M * jd86 so i just make a file called context and put like 4 in it? 1109474493 M * Bertl yep 1109474496 M * jd86 alright 1109474521 M * jd86 so i wonde rhow i went wrong with my calculation 1109474555 M * jd86 cuz i thought i'd do 25 x 1024 (to get kb) x 4 (to get page) 1109474571 M * Bertl 25MB = 25*1024*1024 bytes = 25*128*4096 = 25*128 pages 1109474582 M * Bertl oops *256 1109474594 M * Bertl 25*256 1109474601 M * jd86 you quite confused me 1109474608 M * jd86 you mean all i do is mb x 256? 1109474618 M * jd86 so 6400 1109474636 M * Bertl yep 1109474642 M * jd86 you got a memhog.c too? 1109474657 M * Bertl hmm, not yet ... 1109474664 M * jd86 how can one hog some memory then? 1109474686 M * Bertl but there was some perl script creating an array IIRC, any perl freaks around? 1109474715 M * jd86 any quick like bash or something to hog mem? 1109474720 M * jd86 up to 25 mem perfrably 1109474723 M * jd86 well more then 25 1109474727 M * daniel_hozac Bertl: dlimits are per file system, right? 1109474734 M * Bertl yep 1109474758 M * Bertl jd86: iirc, sort is a good hogger 1109474797 M * jd86 http://www.linuxquestions.org/questions/archive/9/2002/03/1/14816 1109474800 M * jd86 that look like a good thingy? 1109474802 M * jd86 i dont know c 1109474840 M * Bertl hmm, yes looks fine ... 1109474844 M * jd86 first one? 1109474878 M * Bertl yeah, should work for your purpose 1109474908 M * jd86 well whats the include supposed to say? 1109474914 M * jd86 it dont have anything exdt to it 1109474915 M * jd86 next 1109474940 M * Bertl for the include part, it should work without 1109474958 M * jd86 memhog.c:2:9: #include expects "FILENAME" or 1109474958 M * jd86 memhog.c: In function `main': 1109474958 M * jd86 memhog.c:12: error: `NULL' undeclared (first use in this function) 1109474958 M * jd86 memhog.c:12: error: (Each undeclared identifier is reported only once 1109474958 M * jd86 memhog.c:12: error: for each function it appears in.) 1109474990 M * Bertl change the #include to #include 1109475010 M * jd86 ahh 1109475011 M * jd86 good 1109475025 M * jd86 llocated 17370 1109475025 M * jd86 Allocated 17371 1109475025 M * jd86 Allocated 17372 1109475025 M * jd86 Allocated 17373 1109475025 M * jd86 Allocated 17374 1109475025 M * jd86 Allocated 17375 1109475027 M * jd86 Killed 1109475078 M * jd86 how do i check how much memory the vserver is using! 1109475096 M * Bertl /proc/virtual//limit 1109475123 M * jd86 i tinky its worky 1109475131 M * jd86 RSS: 1944 6400 6400 14 1109475157 M * Bertl good to know ;) 1109475207 M * jd86 ok 1109475209 M * jd86 and now 1109475211 M * jd86 one last thing 1109475213 M * jd86 proc. 1109475263 M * jd86 whats the deal with proc, should it be mounted in some way? or what. because i see it is but there is like nothing in proc but i guess process (folders) and mounts and self what if some apps need other things? 1109475353 M * Bertl well, that's called proc security, and should be automatically configured by the vprocunhide script 1109475373 M * Bertl http://linux-vserver.org/Proc-Security 1109475378 M * jd86 well what if the user wants to run 'free' is that not allowed? 1109475425 M * Bertl don#t know what 'free' in your case does, but it should be possible ... 1109475426 M * jd86 oh 1109475436 M * jd86 so i can unhide stuff 1109475438 M * jd86 alright 1109475547 M * jd86 on to the last final thing. 1109475553 M * jd86 disk usage limits/quota stuff. 1109475579 M * Bertl ah, and I thought the final this is the project donation ;) 1109475587 M * Bertl *thing 1109475593 M * jd86 lol 1109475637 M * jd86 trust me if i had money to give you, i'd buy pizza ;-) (i'm just kidding i'll try to leave a few dollars after my next pay check :-) 1109475660 M * jd86 so, about this quota stuff, how do you do it? 1109475663 M * Bertl you can also donate some time, and write a nice beginners howto? 1109475674 M * jd86 yea i was definatly thinking that 1109475701 M * Bertl you need to mount the partition for your vserver(s) with tagxid (option) 1109475718 M * Bertl of course, static xids are a requirement for that 1109475718 M * jd86 i am setting up a vserver again for someone and trying to make it work nicey nicey and then i'll write one after he asks his bajillion additional questions. 1109475731 M * jd86 well i currently have all of this stuff, on one hard drige 1109475735 M * jd86 s/drige/drive 1109475753 M * jd86 so does that mean i should add tagxid to fstab for / and reboot? 1109475760 M * Bertl including the rootfs? (if so, that's unfortunate ;) 1109475777 M * jd86 everything but boot is on one partition 1109475780 M * jd86 oh so i cant do it? 1109475781 M * Bertl no you do not want tagxid on the rootfs ... 1109475786 M * jd86 oh 1109475790 M * jd86 :'( 1109475805 M * jd86 i'll have to remember to use seperate partition when setting up that other server 1109475825 M * Bertl you probably can shrink the existing one 1109475848 M * jd86 sounds _complicated_ 1109475850 M * jd86 lol 1109476107 M * jd86 i'm off to bed 1109476110 M * jd86 thank you very much for your hlep 1109476113 M * jd86 s/hlep/help 1109476116 M * Bertl k, have a good sleep! 1109476126 M * jd86 i'm taking my tokens and going to sleep 1109476138 M * Bertl lol 1109476535 M * daniel_hozac Bertl: if i just want to update the space_total of a dlimit, i set the others to -1, right? 1109476555 M * Bertl hmm, IIRC I had another magic for that -2 1109476567 M * daniel_hozac ah, ok. 1109476936 M * daniel_hozac thanks for your help. 1109476944 M * Bertl my pleasure! 1109479579 J * nox- ~nox@213.39.150.135 1109479908 Q * nox Ping timeout: 480 seconds 1109479939 N * nox- nox 1109482572 Q * IrishKitty Server closed connection 1109482590 J * IrishKitty freemadi@yoda.bowerham.net 1109483199 M * Bertl night folks! back tomorrow! 1109483216 N * Bertl Bertl_zZ 1109485490 J * Zoiah Zoiah@matryoshka.zoiah.net 1109488349 Q * jd86 Server closed connection 1109488365 J * jd86 ~jim@ip68-9-97-23.ri.ri.cox.net 1109490565 Q * zimdog Quit: Leaving 1109490835 J * kk ~kk@220.224.44.12 1109490853 Q * kk Quit: 1109490874 Q * alexx Server closed connection 1109490917 J * alexx ~alexx@82.225.136.176 1109493864 T * * http://linux-vserver.org/ | latest stable 1.2.10, devel 1.9.4.8, ng9.0 -- He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the wiki, and we'll forget about the minute ;) 1109493864 T * Bertl - 1109494185 Q * TheSeer Server closed connection 1109494207 J * TheSeer ~theseer@border.office.salesemotion.net 1109497096 J * eyck eyck@81.219.64.71 1109497926 Q * click Server closed connection 1109497937 J * click click@dsl-84-161.aal.tiscali.no 1109498094 Q * hvd Server closed connection 1109498798 Q * Loki|muh Quit: bbl 1109500058 T * * http://linux-vserver.org/ | latest stable 1.2.10, devel 1.9.4.8, ng9.0 -- He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the wiki, and we'll forget about the minute ;) 1109500058 T * Bertl - 1109500576 Q * Zoiah Remote host closed the connection 1109500679 J * Zoiah Zoiah@matryoshka.zoiah.net 1109502076 J * pulsar ~pulsar@213.184.238.34 1109502569 J * prae ~prae@sherpadown.net 1109502580 M * prae morrnnninnng 1109502581 M * prae :) 1109502718 M * DaPhreak moooorning prae ;) 1109502789 M * prae DaPhreak: =) 1109503061 M * pulsar I'v got a strange error: `mount` in _root_ vserver outputs /dev/hdv1 on / type ufs (defaults) (1.9.4.7 dev) 1109504357 Q * pulsar Remote host closed the connection 1109504415 Q * eyck Quit: leaving 1109504448 J * eyck eyck@81.219.64.71 1109505102 J * Loki|muh loki@satanix.de 1109505778 Q * Loki|muh Read error: Connection reset by peer 1109506606 J * Loki|muh loki@satanix.de 1109508855 M * SiD3WiNDR anyone at fosdem? :) 1109508952 J * rs ~rs@imhotep.rhapsodyk.net 1109508960 M * rs hi there 1109509899 M * prae morning rs 1109510860 J * kk ~kk@220.224.22.105 1109511039 M * kk how does a vserver start 1109511497 M * prae vserver start 1109511572 J * pulsar ~pulsar@213.184.238.34 1109512627 Q * DaPhreak Quit: Hahahaha. Green is best 1109514032 Q * Loki|muh Read error: Operation timed out 1109514136 Q * kk Ping timeout: 480 seconds 1109515135 J * kk ~kk@220.224.22.236 1109515184 M * daniel_hozac Bertl_zZ: why is CDLIM_KEEP defined as a ULL when the structure members are all uint32_ts? 1109515455 J * Loki|muh loki@satanix.de 1109515760 J * rs_ ~rs@imhotep.rhapsodyk.net 1109515831 J * DaPhreak ~DaPhreak@pc-outside.uni-greifswald.de 1109515848 Q * rs Ping timeout: 480 seconds 1109516438 M * pulsar can I use latests utils-vserver with 2.4 kernel ? 1109516790 M * daniel_hozac Bertl_zZ: ah, i see. nevermind. 1109517425 N * Doener_zZz Doener 1109517446 M * Doener morning! 1109518360 M * DaPhreak mornin Doener ! 1109518878 N * Bertl_zZ Bertl 1109518882 M * Bertl morning folks! 1109518907 M * DaPhreak morning Bertl :) 1109519008 M * Bertl daniel_hozac: hmm, good question, does it result in unwanted sideeffects? if not I'll call it future compatibility (i.e. 64bit counters) 1109519066 M * daniel_hozac Bertl: it just requires an explicit cast. 1109519592 M * pulsar Can I use on 2.4 kernel utils-vserver compiled against 2.6 headers ? 1109519606 Q * rs_ Ping timeout: 480 seconds 1109519643 M * Bertl pulsar: probably, but I would be careful as the tools might not use the right interfaces ... 1109519676 M * Bertl (i.e. it will 'just' use the legacy one I guess, but I might be completely wrong, Enrico did a good job on the tools) 1109519876 Q * kk Ping timeout: 480 seconds 1109519973 M * Bertl pulsar: ah, I misread your question, you want to use 2.6/new tools on 2.4, right? 1109520009 M * Bertl that should work fine as long as 'required' libraries are present 1109520116 Q * BobR_oO Server closed connection 1109520144 J * BobR_oO ~georg@212.16.62.51 1109521328 M * Doener Bertl: now i know we i didn't see that concept before, the last time that guy was here, was a day before i showed up the first time ;) 1109521367 M * Bertl so it's your fault? *G* 1109521496 M * daniel_hozac does anyone know how enrico decides what goes into what API? 1109521575 M * Bertl well, probably Enrico in the first place ... 1109521603 M * Bertl of course we discuss and decide on the kernel side api ... 1109521641 M * Doener seems so 8-) 1109521689 M * daniel_hozac i mean the v11, v13, fscompat etc. APIs that util-vserver keeps referring to... do they exist on the kernel side as well? 1109521715 M * Bertl yes, they are the result of kernel changes and abckwards compatibility 1109521743 A * Bertl .o( hmm a nice typo, I have to remember that one ) 1109521765 M * daniel_hozac ah. so what API do dlimits belong to? 1109521784 M * daniel_hozac (and how would i find that out?) 1109521800 M * Bertl AFAIK they did not make it into the tools yet? 1109521826 M * daniel_hozac nope, but i've been trying to add them ;) 1109521900 M * Bertl hmmm, how successfully? I mean I have no problem to start a discussion on the interface and to adjust some things ... 1109521957 M * daniel_hozac http://daniel.hozac.com/vserver/util-vserver-dlimit.patch 1109521959 M * Bertl (or maybe s/successfully/seriously/ ...) 1109521984 M * daniel_hozac for now, i put them in the v13 API as that's where i guessed it'd be. 1109521995 M * Bertl hmm, looks good ... 1109522016 A * Bertl is reading up on his (Bertl's) code 1109522091 M * prae hmm, when I start vserver with "vserver start", what is the first "program/script" that has been launched in vserver ? 1109522107 M * prae /sbin/init ? 1109522115 M * Bertl depends on the config and init style 1109522128 M * Bertl with sysv init style it's the rc script ;) 1109522162 M * prae /etc/rc.S/rc or /etc/init.d/rc ? 1109522172 M * Bertl daniel_hozac: probably a v19 interface would be the best 1109522184 M * Bertl (as the syscalls are not in vs1.3.x ... 1109522214 M * Bertl okay, off for dinner, but I have time for that after it ... 1109522226 N * Bertl Bertl_oO 1109522662 J * Tauras ~gregory@195.158.80.126 1109522712 M * daniel_hozac hmm, an entire new API? i think i'll leave that to Enrico... 1109522731 M * Tauras hello 1109522783 M * Tauras It's my first time trying install vserver and it cannot startup 1109522826 M * Tauras vs1.9.4 1109522841 M * Tauras Error: /proc must be mounted 1109522856 M * Tauras Failed to parse ps-output 1109522902 M * Tauras what should I do? 1109522913 M * Doener http://linux-vserver.org/Proc-Security 1109522926 M * Tauras I've read it 1109522945 J * kk ~kk@220.224.9.239 1109522962 M * Doener and you ran vprochunhide? 1109522974 M * Doener s/hunhide/unhide/ 1109523094 M * Tauras I have no vprochunhide, but I tried in host machine setattr --~hide /proc/uptime 1109523131 M * Doener which tools do you use? 1109523195 M * Tauras util-vserver 0.30.203 1109523211 M * Doener then you should have vprocunhide.... 1109523227 M * Tauras host is debian sarge and vserver is sarge too 1109523264 M * Doener if you didn't specify a prefix while configuring, an initscript should be at /usr/local/etc/init.d/vprocunhide and vprocunhide itself at /usr/local/lib/util-vserver/vprocunhide 1109523513 M * Tauras thank you, I found it, trying 1109523553 M * Tauras chcontext: vc_new_s_context(): Operation not permitted 1109523683 M * kk how does vserver start 1109523700 M * Tauras # vserver sarge1 start 1109523700 M * Tauras WARNING: can not find configuration, assuming legacy method 1109523700 M * Tauras Starting the virtual server sarge1 1109523701 M * Tauras Server sarge1 is running 1109523740 M * kk no i mean at which runlevel does it start 1109523809 M * Tauras that is all I can see when it starts 1109523816 M * kk thanx 1109523821 Q * kk Quit: Leaving 1109523902 M * Tauras what configuration vserver is missing here? 1109523971 M * Tauras I have configuration file /etc/vservers/sarge1.conf 1109523998 M * Tauras and these are some options there: 1109524000 M * Tauras S_FLAGS="lock nproc fakeinit" 1109524000 M * Tauras ULIMIT="-H -n 1024" 1109524001 M * Tauras S_CAPS="CAP_NET_RAW" 1109524277 J * rs ~rs@82.229.176.40 1109524370 M * daniel_hozac Bertl_oO: hmm, it seems there is already a v13b API requiring version 0x00010021 or higher... wouldn't that work for dlimits? 1109524417 N * Bertl_oO Bertl 1109524439 M * Bertl whoa, lot of talk since I left ;) 1109524483 M * Bertl daniel_hozac: well, contact enrico for that then, I guess he will gladly give you some advice ... 1109524528 M * jd86 how do you sqish a reiserfs partition? 1109524545 M * Bertl squish? fdisk, dd? 1109524563 M * jd86 like, theres tons of free space i want to rob some of that and make another partition 1109524573 M * Bertl Tauras: you should adapt that (probably debian default?) config 1109524589 M * Bertl a you mean shrink in size? 1109524608 M * jd86 yes Bertl 1109524623 M * Bertl jd86: AFAIK reiserfs doesn't support shrinking just expansion :/ 1109524627 M * jd86 aww :( 1109524632 A * jd86 re does it, no big 1109524689 M * Tauras Bertl: I think it is default 1109524717 M * Tauras Bertl: what should I write into it then? 1109524725 M * Bertl yes, but it's a bad config for a default ;) 1109524750 M * Bertl you should change the following at least: 1109524757 M * Bertl S_FLAGS="nproc fakeinit" 1109524768 M * Bertl ULIMIT="-HS -n 1024" 1109524775 M * Bertl S_CAPS="" 1109524781 M * Tauras ok 1109524789 M * Bertl and complain to the maintainer of those defaults! 1109524797 M * Tauras it was generated by vserver-debiantools 1109524804 M * jd86 Bertl: for the vservers partition should i use ext2, ext3, reiserfs or what do you suggest? 1109524811 M * Bertl in addition it would be advised to use static context 1109524853 M * jd86 Bertl: does it make a difference? 1109524854 M * Bertl Tauras: so add S_CONTEXT= 1109524877 M * Bertl jd86: it's probably best tested with ext2/3 but reiser and xfs should work too 1109524886 M * jd86 well i'd like to go with something tested. 1109524902 M * Bertl Tauras: so the vserver-debiantools maintainer is the one to complain to ;) 1109524943 M * Bertl Tauras: also I would have a look at the non-legacy config, as many new features (of 2.6/1.9) are not available with the 'old' config 1109524950 M * jd86 Bertl: i'm using gentoo should i manually get all of the things down so i have latest versions or just emerge them and use the (possibly) oudated version (anotherwords get the dev ver of util-vserver or use the ver that is in portage?) 1109524952 M * Tauras I changed and during restart I had this: 1109524954 M * Tauras New security context is 49153 1109524966 M * Tauras same error when I try enter 1109524977 M * Bertl probably nothing was started 1109524987 M * Tauras chcontext: vc_new_s_context(): Operation not permitted 1109524988 M * Bertl try with vserver-stat 1109525017 M * Tauras CTX PROC VSZ RSS userTIME sysTIME UPTIME NAME 1109525017 M * Tauras 0 45 118.5M 11K 2m40s69 2m28s75 1d47h09 root server 1109525017 M * Tauras 49153 1 1.4M 130 0m02s91 0m00s17 33m09s10 1109525044 M * Bertl what kernel version do you use? kernel and patches? 1109525073 M * Tauras debian-source-2.6.10 1109525079 M * Bertl maybe you want to give that one a try: 1109525081 M * Bertl http://vserver.13thfloor.at/Stuff/SCRIPT/testme.sh 1109525115 M * Bertl let me know about the output 1109525436 M * Tauras all succeeded 1109525455 M * Tauras can it be fakeinit problem? 1109525459 M * Bertl good, that's at least a start ... 1109525471 M * Bertl can you paste the first 3 lines? 1109525555 M * Tauras Linux-VServer Test [V0.09] (C) 2003-2005 H.Poetzl 1109525555 M * Tauras chcontext is working. 1109525555 M * Tauras chbind is working. 1109525556 M * Tauras chcontext 0.30.203 -- allocates/enters a security context 1109525556 M * Tauras This program is part of util-vserver 0.30.203 1109525575 M * Tauras what lines did you mean? 1109525601 M * Bertl hum, you get a chcontext version message? 1109525671 Q * pusling Server closed connection 1109525676 M * Tauras I ran with -v 1109525690 J * pusling ~pusling@195.215.29.124 1109525698 M * Bertl ah, okay, then please the version info without -v ;) 1109525734 M * Tauras # sh testme.sh 1109525735 M * Tauras Linux-VServer Test [V0.09] (C) 2003-2005 H.Poetzl 1109525735 M * Tauras chcontext is working. 1109525736 M * Tauras chbind is working. 1109525748 M * Tauras Linux 2.6.10-vs1.9.4-rm2 i686/0.30.203/0.30.203 [Ea] 1109525749 M * Tauras --- 1109525749 M * Bertl and the next line please ;) 1109525749 M * Tauras [001]# succeeded. 1109525762 M * Bertl okay, that's fine! 1109525792 M * Bertl did you add a static context id for your vserver? 1109525804 M * Tauras ? 1109525811 M * Tauras what is this 1109525823 M * Bertl S_CONTEXT= 1109525857 M * Tauras Its my firat day... 1109525866 M * Tauras no I have no such line in config 1109525866 M * Bertl if not, stop your vserver and change it now 1109525909 M * Bertl then start it with 'vserver --debug start' and upload the output somewhere (e.g. pastebin.com) 1109525922 M * Tauras btw on wiki it should be emphasized on configuration :-) 1109525935 M * Bertl feel free to emphasize on that ;) 1109526048 M * Tauras hmm, vserver sarge1 running shows it is still running 1109526089 M * Bertl did you stop it? 1109526093 M * Tauras yes 1109526108 M * Bertl and that wen without errors? 1109526177 M * Tauras kill init, pid 6969 1109526183 M * Tauras /usr/lib/util-vserver/legacy/vserverkillall: line 59: kill: (6969) - No such process 1109526214 M * Tauras can I clen up manually? 1109526217 M * Bertl what does vps list? 1109526245 M * Bertl (vps is like ps but for all cotnexts) 1109526256 M * Tauras 3 processes bash, vps, ps 1109526269 M * Bertl try with vps -auxww 1109526271 M * Tauras I guess it is vps proceses 1109526345 M * Tauras yep init process is 6969 ctx 49153 1109526400 M * Tauras during stop it goes: New security context is 49153 1109526418 M * Tauras but cannot stop 1109526478 M * Bertl try vkill -x 49153 -k 9 1109526637 M * Tauras I killed just before you wrote with vkill -c 49153 -s 9 6969 1109526658 M * Tauras so it is not running any more :-) thanks a lot 1109526667 M * Bertl you're welcome! 1109526754 M * Tauras what is legacy file? 1109526808 M * Bertl as linux-vserver started, it had a one file config (the one you are using now) 1109526839 M * Bertl since some timenow the config uses a directory based configuration scheme, so the old, depreciated config is called 'legacy config' 1109526871 M * Bertl this config file is usually referred to as legacy (config) file 1109526884 M * Tauras you can see debug startup in pastebin.com 1109526897 M * Bertl url? 1109526915 M * Tauras http://www.pastebin.com/247063 1109526920 M * Bertl got it 1109526974 M * Bertl hmm, seems the --debug option didn't make it into the legacy file ... funny ... 1109527005 M * Bertl lets stop the entire thing once again (should now work) 1109527043 M * Tauras kill init problem 1109527043 M * Bertl but it looks to me as if your init is sucessfully started, so what are your issues? that the init isn't starting anyservices? 1109527069 M * Bertl okay, kill it as before, then let's try: 1109527094 M * Bertl bash -x /usr/lib/util-vserver/legacy/vserver sarge1 start 1109527306 M * Tauras do you want me to paste evferything? 1109527321 M * Bertl no please upload it as before 1109527332 M * Tauras yes I meant it 1109527437 M * Tauras what is this /etc/vservers/sarge1.sh? 1109527448 M * Tauras http://www.pastebin.com/247069 1109527460 M * Bertl a script which is used on startup and exit 1109527619 M * Bertl okay, that looks correct, your vserver's init is started, it seems to sit around then ... 1109527649 M * Tauras during enter: chcontext: vc_new_s_context(): Operation not permitted 1109527674 M * Tauras vps -ax shows vserver processes 1109527696 M * Tauras it is just I cannot enter vserver 1109527696 Q * maharaja Server closed connection 1109527711 M * Bertl try vcontext --migrate --xid 555 /bin/bash 1109527719 J * maharaja maharaja@ipax.at 1109527727 M * Bertl wb maharaja! 1109527744 M * Tauras I'm inside 1109527766 M * Bertl well, yes and no ... 1109527817 M * Bertl it seems that debian 'still' used relatively old versions, so either you update the kernel and tools, or you switch to a non-legacy config, there was an issue with legacy config which was fixed some time ago 1109527882 M * Tauras u mean to update util-vserver? 1109527916 M * Bertl and the kernel aptches, we are currently at 2.6.11-rc5-vs1.9.4.8 (tools 0.30.204) 1109527925 M * jd86 Bertl: can you point me to a document that explains how to seup qoutas (hard disk spac elimiting per vserver) so when i get this other box up i can do it? 1109527968 M * Tauras ok, i'll try vanilla kernel on other pc 1109527987 M * Tauras thank you for your time and help 1109527994 M * Bertl you're welcome! 1109528011 M * Bertl jd86: google says: http://linux-vserver.org/index.php?page=Quota+and+Disk+Limits ;) 1109528032 Q * Tauras Quit: using sirc version 2.211+KSIRC/1.3.10 1109528355 M * jd86 Bertl: i read the whole thing but my question now is, do i have to do any additonal patching to the kernel? or just enable the 24/24 "thing" under verservers 1109528358 M * jd86 s/verservers/vservers 1109528380 M * Bertl with 2.61.9.x the disk limits are included in the base patch 1109528388 M * Bertl 2.6/1.9.x that is 1109528418 M * jd86 alright is there anything horribly changed from 1.9.4 to whatever latest is? 1109528441 M * Bertl a few features where added and some 'bugs' fixed ... 1109528446 M * Bertl *were 1109528458 M * jd86 so i should manually patch to the latest? 1109528473 M * Bertl 1.9.5 should be out soon ... 1109528497 M * jd86 hrm 1109528500 M * jd86 ugha 1109528683 M * jd86 whatabout util-vserver 0.30.196, much newer then that? 1109528724 M * Bertl 0.30.204 has a bunch of fixes and support for some newer build methods IIRC 1109528738 M * jd86 all have to snatch that 1109528881 M * jd86 where can i snatched the latest? 1109528895 M * Bertl http://linux-vserver.org/alpha+util-vserver 1109528954 M * jd86 so .214 should be a good version 1109528971 M * Bertl probably, we'll see ;) 1109528997 M * jd86 haha 1109529105 M * jd86 updating away 1109529270 M * jd86 ahh its really .204 1109529285 M * jd86 messing me up :( 1109530091 Q * Medivh Server closed connection 1109530117 J * Medivh ck@paradise.by.the.dashboardlight.de 1109530922 M * prae it's possible to exec with chcontext a prog into vs when we know CTX number ? 1109530991 M * Bertl sure 1109531267 M * prae how ? :p I test lot of options, but it reject them 1109531286 M * prae chcontext --xid 49187 -- lsof 1109531292 M * Bertl chcontext --ctx -- sleep 100 & 1109531295 M * prae chcontext: vc_new_s_context(): Operation not permitted 1109531314 M * Bertl that just means that you have an old kernel/tools 1109531323 M * Bertl and are using a legacy config file ;) 1109531355 M * Bertl and yes, --xid is now default! 1109531365 M * prae it's util-vserver-0.30.203 and kernel 2.6.10 & vs1.9.4 1109531382 M * jd86 do you guys need a mirror for the util-vserver or anything? 1109531406 M * Bertl not at the moment I guess, but can we come back to that at a later time? 1109531426 M * Bertl (if so, just msg me some information how to contact you) 1109531435 M * jd86 Bertl: you talking to me? 1109531454 M * Bertl if you were talking to me too, then yes ;) 1109531476 M * jd86 you get the msg? 1109531489 M * Bertl yup, tx 1109531500 A * prae shake chcontext 1109531515 M * prae *shake, shake, seniora* 1109531634 M * Bertl prae: hmm, where is the kernel from? 1109531674 M * prae Bertl, don't slaps me ! ... I wasn't root :p 1109531676 A * prae shame 1109531861 J * DuckKing ~Duck@dyn-83-157-150-192.ppp.tiscali.fr 1109531905 M * prae chcontext --xid 49187 -- sleep 100 1109531905 M * prae vcontext: vc_create_context(): Invalid argument 1109531926 M * jd86 -- shake -- --shake -- -- shake -- 1109531952 M * prae jd86: always :) 1109532046 Q * DuckMaster Read error: Operation timed out 1109532098 M * Bertl well, folks! let's investigate this a little closer 1109532113 M * Doener hm... we've seen this before IIRC... 1109532126 M * Bertl because both of your report 2.6.10-vs1.9.4 and I'm pretty confident the issue is already fixed in 1.9.4 ... 1109532158 M * Doener Bertl: could you refresh my memory? what did we fix there? 1109532180 M * Bertl when using legacy config, the vserver falls back to the legacy tools too 1109532201 M * Bertl this uses the old syscall commands ... 1109532253 M * Bertl IIRC we added a check to 'allow' certain commands (which where not allowed before) to make that work ... 1109532260 M * prae it's strange, when I use a xid already used (for my vserver), chcontext doesn't like this 1109532274 A * Bertl is now loogin for the relevant patch ... 1109532287 M * prae xid 0 and 49187 failed, but other is ok 1109532302 M * Bertl ahem 0? 1109532309 M * Doener hm, wasn't that a fix for an EPERM error? 1109532332 M * Bertl ad 49187 does this context exist? 1109532352 M * prae yes 1109532361 M * prae used by one of my vserver 1109532367 M * Bertl prae: because if it isn't already running you are not allowed to create a context with xids higher than 49151 1109532380 M * Bertl vserver-stat does list it? 1109532408 M * prae yes 1109532414 M * prae CTX PROC VSZ RSS userTIME sysTIME UPTIME NAME 1109532414 M * prae 0 100 1.8G 99.7K 24m26s19 8m25s60 8h26m02 root server 1109532414 M * prae 49187 3 5.7M 674 0m00s57 0m00s17 36m44s81 cortex 1109532456 M * Bertl okay, lets get the output of 'cat /proc/virtual/49187/*' and upload it somewhere please ... 1109532486 M * Bertl (you are free to censor sensitive information) 1109532520 M * prae http://pastebin.com/247105 1109532528 M * Bertl tx 1109532556 M * prae no sensitive information for the moment :) 1109532771 M * prae you suggest me to force "xid" below 49151 for my new vserver ? 1109532815 M * Bertl if you use static context ids you have to go below 49152 1109532829 M * Bertl but let's keep the current dynamic for now 1109532988 M * Bertl could you strace -fF the chcontext and upload that somewhere too? and do you ahve VSERVER_DEBUG enabled? 1109532998 M * prae no prob 1109533007 M * prae chcontext with my own options ? 1109533024 M * prae VSERVER_DEBUG, no, i'm no sure, I check 1109533040 M * Bertl yep your options 1109533046 M * prae # CONFIG_VSERVER_DEBUG is not set 1109533048 M * prae sorry :( 1109533059 M * Bertl np, probably not relevant anyway 1109533141 M * prae http://pastebin.com/247112 1109533208 M * Bertl tx 1109533319 M * prae vcontext --migrate --xid 49193 -- lsof 1109533321 M * prae works 1109533342 M * Bertl yep, that's supposed to work ... 1109533465 M * prae I suppose that "execve" into line 917 has a problem 1109533532 M * Bertl hmm, that looks like a bug in the tools indeed ... 1109533541 J * zimdog ~zimdog@c-67-164-190-201.client.comcast.net 1109533546 M * Bertl welcome zimdog 1109533557 M * Bertl Doener: it is calling VCMD_ctx_create 1109533558 M * zimdog hello Bertl 1109533616 M * Bertl prae: could you download and compile 0.30.204 and try again without changing or restarting the vserver? 1109533645 M * Bertl (make sure to compile it 'debian' specific or to remove the old tools first) 1109534142 M * prae 0.30.204 ? ok, wait 1109534165 M * prae (same version under my sys ?) 1109534224 M * Bertl yes 0.30.204 on your system (without vserver restart) after removing existing tools 1109534321 M * prae arf! sorry, I believed that I had already 0.30.204 :p 1109534331 M * prae wait, 1109534401 M * Bertl and get the tar sources not some debian package (sorry deb-folks!) 1109534549 M * prae of course :) 1109534946 M * zimdog Is there any docs on moving /vservers to another partition? 1109535264 M * prae Bertl: 1109535265 M * prae chcontext --xid 49194 -- sleep 5 1109535266 M * prae vcontext: vc_create_context(): Invalid argument 1109535271 M * prae with .204 1109535271 M * Bertl zimdog: hmm, should be fairly simple 1109535295 M * Bertl prae: and xid 49194 does exist, right? 1109535300 M * prae yes 1109535302 M * prae vserver-info 2> /dev/null| grep util 1109535302 M * prae util-vserver: 0.30.204; Feb 27 2005, 21:07:12 1109535309 M * prae 49194 3 5.7M 658 0m00s40 0m00s10 0m11s95 cortex 1109535349 M * zimdog Bertl: May havve found my problem. I was doing a service vservers stop and it was leaving the servers mounted I guess. I did a vserver stop and then it let me mv it 1109535378 M * Bertl yep, that's kind of expected .. ;) 1109535432 M * Bertl prae: 1109535433 M * Bertl 20:08 < prae> chcontext: vc_new_s_context(): Operation not permitted 1109535449 M * Bertl in this case you didn't have a context right? 1109535473 M * prae what did you mean ? 1109535478 M * Bertl or was this with even older tools? 1109535500 M * Bertl I'm really confused because your tools seem to act very weird ... 1109535514 M * Bertl the vc_new_s_context() call is a legacy syscall 1109535531 M * Bertl while the vc_create_context() one is a new-style syscall 1109535562 M * Bertl so either your tools are not sure what to do/call or you changed the tools at some point, or the configs/contents did change at some point ... 1109535601 M * Bertl in any case if you use a legacy config, the tools should use the legacy calls, or the 'correct' new calls which both doesn't happen ... 1109535613 M * Bertl but I'm going to test that here right now in a few minutes ... 1109535935 Q * ntrs_ Read error: Connection reset by peer 1109536004 M * prae you suppose that an older version of util-vserver is somewhere in my system ? 1109536029 M * Bertl it could be, let me check here, if I can reproduce the issues, maybe it's a bug in recent tools ;) 1109536038 M * maharaja maja 1109536045 M * maharaja ups :) 1109536058 M * Bertl too much pokeymons? ;) 1109536064 M * prae Bertl: ok :) 1109536426 M * Bertl # vserver XXXX start 1109536426 M * Bertl WARNING: can not find configuration, assuming legacy method 1109536426 M * Bertl Starting the virtual server XXXX 1109536426 M * Bertl Server XXXX is not running 1109536433 M * Bertl (this looks similar for you, right?) 1109536526 M * Bertl # vserver XXXX enter 1109536526 M * Bertl WARNING: can not find configuration, assuming legacy method 1109536530 M * Bertl [root@vserver:XXXX /] 1109536542 M * Bertl (should look similar too, please try) 1109536564 M * Bertl # chcontext --xid 49152 true 1109536564 M * Bertl vcontext: vc_create_context(): Invalid argument 1109536572 M * Bertl (indeed fails .. funny thing ;) 1109536618 M * Bertl # vcontext --migrate --xid 49152 true 1109536625 M * Bertl (works as expected ...) 1109536681 M * prae Bertl: WARNING: can not find configuration, assuming legacy method 1109536681 M * prae Starting the virtual server XXXX 1109536681 M * prae Server XXXX is not running 1109536681 M * prae (this looks similar for you, right?) 1109536684 M * prae heuuu .. no 1109536695 M * prae I haven't Warning messages 1109536711 M * prae root@carbone:~# vserver cortex start 1109536711 M * prae Starting system log daemon: syslogd. 1109536711 M * prae Starting periodic command scheduler: cron. 1109536726 M * prae root@carbone:~# vserver cortex enter 1109536726 M * prae sh-2.05b# 1109536731 M * Bertl hmm, you have old style or new style config? 1109536742 M * prae hmmm new style I'm suppose 1109536744 M * Bertl (i.e. config file or config tree) 1109536747 M * prae tree 1109536752 M * prae so, new style :) 1109536773 M * Bertl okay, I tried with legacy now, but it seems to be the same for new style config 1109536820 M * Bertl but I guess it's working according to specs in both cases actually ... 1109536834 M * Bertl chcontext --help says: 1109536844 M * Bertl chcontext allocate a new security context and executes 1109536847 M * Bertl a command in that context. 1109536851 M * Bertl By default, a new/unused context is allocated 1109536868 M * Bertl (this is broken english, but it says nothing about 'entering' a context) 1109536987 M * Bertl Doener: I guess the real issue is that chcontext is not any longer supposed to be used for entering running contexts, that it works with static xid might be a bug on the kernel side ... 1109537030 M * Bertl (although the name is somewhat misleading then) 1109537299 M * Doener hm, should give -EEXIST... except if VXF_STATE_SETUP is set... 1109537311 M * Bertl # chcontext --xid 1001 true 1109537312 M * Bertl vxD: vc: VCMD_00_0[0], 63 1109537312 M * Bertl vxD: vc: VCMD_09_1[0], 1001 1109537312 M * Bertl vxD: vc: VCMD_11_1[0], 1001 1109537314 M * Bertl - 1109537323 M * Bertl # chcontext --xid 49152 true 1109537323 M * Bertl vxD: vc: VCMD_00_0[0], 63 1109537323 M * Bertl vxD: vc: VCMD_09_1[0], 49152 1109537323 M * Bertl vcontext: vc_create_context(): Invalid argument 1109537333 M * Bertl - 1109537344 M * Bertl with 1.9.4.8 and .30.204 1109537378 M * Bertl but it's interesting actually ... 1109537477 M * Bertl hmm, probably a kernel issue after all ... 1109537524 M * Doener hm, no __loc_vx_info should already return -EBUSY... (if in setup state) 1109537542 M * Bertl check if ((xid >= MIN_D_CONTEXT) && (xid != VX_DYNAMIC_ID)) 1109537543 M * Bertl return -EINVAL; 1109537552 M * Bertl in vc_ctx_create() 1109537552 M * Doener yep, that's for dynamic 1109537572 M * Doener i'm talking about static ones atm 1109537598 M * Doener if (!(new_vxi->vx_flags & VXF_STATE_SETUP)) { 1109537598 M * Doener ret = -EEXIST; 1109537613 M * Doener that should make the call fail for static context, shouldn't it? 1109537678 M * Bertl yep .. it should ... 1109537697 M * Bertl unless the setup flag is still set 1109537723 M * Doener then we would bail out in __loc_vx_info with -EBUSY 1109537732 M * Doener if (vxi->vx_flags & VXF_STATE_SETUP) { 1109537732 M * Doener vxdprintk(VXD_CBIT(xid, 0), 1109537732 M * Doener "loc_vx_info(%d) = %p (not available)", id, vxi); 1109537803 M * Bertl okay, let me enable debuging for that and do a trace ... 1109537979 M * Bertl http://vserver.13thfloor.at/Stuff/vc_create_context.txt 1109538497 M * Doener how is it possible that both checks 'succeed'? either VXF_STATE_SETUP is set or it is not set... 1109538649 M * Bertl or we have a bug somewhere ;) 1109538897 M * Bertl VXF_STATE_SETUP is not set 1109538917 M * Bertl so we end up with 1109538936 M * Bertl vxi = 'the located context' 1109538954 M * Bertl and we do a get on that and __loc_vx_info() returns that as expected 1109538995 M * Bertl now the interesting part is, that vc_create_contex() should not use the __loc_vx_info() in this way anywas 1109539041 M * Bertl now let's see who uses __loc_vx_info() 1109539083 M * Bertl locate_or_create_vx_info() and vc_ctx_create() 1109539106 M * Bertl the first one is for legacy purposes, and does the job just as expected ... 1109539128 M * Bertl the second one would require a stricter checking ... 1109539155 M * Bertl so maybe we add a new arg called 'create' to that? 1109539188 M * Bertl __loc_vx_info(int id, int *err, int create) 1109539223 M * Bertl we can then block some other 'confusing' checks too ... 1109539264 A * Bertl is doing that right now, expect a diff in a few minutes 1109539291 M * Doener or create a 'create only' function and use whatever we need in each place? 1109539309 M * Bertl would be an option too .. thought about that ... 1109539317 M * Bertl maybe we should do that actually ... 1109539348 M * Bertl does __create_vx_info() sound good? 1109539393 M * Doener yep. better than ate_vx_info (comparing to alloc/loc_vx_info ;) 1109539420 A * Bertl .o( good that I already had dinner ;) 1109540173 M * Bertl okay, test compile ;) 1109540661 M * Bertl I'm pretty sure it will break existing tools ;) 1109540734 M * Bertl okay, it compiled, uploading patch in a jiffie 1109540976 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-1.9.5/delta-create_info-feat01.diff 1109540993 M * Bertl now let's see what still works ;) 1109541032 M * Bertl testme.sh runs perfectly! 1109541106 M * Bertl (with old and new tools, 0.30{,193,204} 1109541194 M * prae :) 1109541252 M * Bertl context startup (vserver) works fine with old and new tools 1109541331 M * Bertl ah, entering with chcontext gives an oops :/ 1109541365 M * Doener existing context or new context? 1109541372 M * Bertl existing, new works fine 1109541513 M * Bertl hmm, seems to happen in the history recoding ... 1109541624 M * Doener + vxi = ERR_PTR(-EINVAL); 1109541624 M * Doener + goto out_unlock; 1109541633 M * Doener and then: 1109541633 M * Doener + vxh_create_vx_info(id, vxi); 1109541634 M * Bertl yep, right ... 1109541652 M * Bertl thanks! 1109541656 M * Doener :) 1109541706 M * Bertl vxh_create_vx_info(id, IS_ERR(vxi)?NULL:vxi); 1109541710 M * Bertl better? 1109541748 M * Doener yep 1109541792 M * Doener but probably that branch should move out of __create_vx_info anyway, shouldn't it? 1109541817 Q * DaPhreak Quit: Orks are da best 1109541822 M * Bertl hmm, please elaborate? 1109541910 M * Doener forget it, i've been reading __loc_vx_info... 1109541950 M * Bertl well, we could probably rewrite that to use ERR_PTR too, what do you think? 1109541982 M * Doener good idea 1109541990 M * Bertl interested in doing so? 1109542027 M * Doener you know how to delegate work ;) ok, will do 1109542051 J * mef ~mef@pcp09895218pcs.ewndsr01.nj.comcast.net 1109542056 M * Bertl welcome mef! 1109542061 M * mef hi bertl... 1109542075 M * mef keeping busy? 1109542076 M * Bertl prae: okay, so there is good and bad news for you 1109542120 M * Bertl the good news is that we pretty fixed up that issue, the bad news is that chcontext will now stop working with static xids too (once the kernel uses that patch) 1109542121 Q * eyck Read error: Connection reset by peer 1109542126 M * Doener about keeping backwards compability... should we update the legacy stuff? 1109542130 J * eyck eyck@81.219.64.71 1109542138 M * Bertl mef: sure always ... 1109542155 M * Bertl Doener: in what way? 1109542172 M * Doener make chcontext work as it used to do... enter or create a context 1109542205 M * Bertl would be an userspace tools change .. maybe that is the way to go ... at least for legacy servers 1109542223 M * mef bertl: have to go, but just sent you email. Cheers! 1109542240 M * Bertl k, cya 1109542242 Q * pulsar Read error: Operation timed out 1109542243 M * prae Bertl: now, all xids are only dynamics ? 1109542255 Q * mef Quit: 1109542282 M * Doener Bertl: hm, we would just have to change the handler for the legacy syscall? what do i miss? 1109542301 M * Bertl prae: no, but in your current setup, switching to static xids should 'work around' your issues (for now) 1109542330 M * Bertl Doener: that the tools are not using the legacy calls anymore ;) 1109542336 M * prae so, chcontext works with ? ... statics or dynamics ? 1109542344 M * prae (now, with the latests patch) 1109542375 M * Bertl with the latest patch, it doesn't work for either if the context is running, it will work for static if the context doesn't exist ... 1109542408 M * Bertl without that latest patch (i.e. in your original setup) 1109542421 M * Bertl you can 'exploit' a bug and switch to static context ids ;) 1109542465 M * Bertl but what you actually want to do is use 'vserver exec ...' 1109542476 M * Doener Bertl: the legacy chcontext uses the legacy call and the new one uses a simple migrate if the context exists... 1109542494 M * Bertl obviously not ... 1109542497 M * jd86 Bertl: ok i have another vserver box setup with a seperate filesystem (ext3 for vservers) and i'd like to try the quota stuff per vserver 1109542505 M * jd86 Bertl: if you have time to help me with it :-\ 1109542521 M * Bertl sure .. start with mounting it with tagxid 1109542527 M * Bertl (and verify that in /proc) 1109542532 M * jd86 ok hold on 1109542567 M * jd86 mount tells me its mounted with rw,noatime,tagxid 1109542576 M * Bertl # chcontext --xid 1001 true 1109542576 M * Bertl vxD: vc: VCMD_00_0[0], 63 1109542576 M * Bertl vxD: vc: VCMD_09_1[0], 1001 1109542576 M * Bertl vcontext: vc_create_context(): Invalid argument 1109542584 M * jd86 thats what i'm doing now? 1109542588 M * Bertl Doener: new tools with patch now 1109542611 M * Bertl jd86: no, you must read 'between' the lines (I'm multitasking) 1109542620 M * jd86 Bertl: :), more shakespere 1109542626 M * Bertl precisely! 1109542628 M * jd86 heh 1109542654 M * Bertl now check via /proc/mounts (we do not trust 'mount') 1109542666 M * jd86 yes 1109542668 M * jd86 its in there. 1109542687 M * Doener Bertl: could you bash -x chcontext ...? 1109542695 M * Bertl jd86: okay, then let's put a vserver with static xid on that ... 1109542728 M * jd86 h/o let me see what the file is called to make it a static one. 1109542754 M * prae Bertl: but, we can continue to exec a prog under other xid (in main) with vcontext --migrate ? 1109542765 M * Doener hm, seems the tools don't return what they're expected to return... 1109542845 M * Doener Bertl: the tools expect EEXIST if the context exists (which is actually the right thing to return i guess ;) 1109542876 M * Bertl http://pastebin.com/247148 1109542893 M * Bertl okay, so we should change that then ... 1109542905 M * Doener yep, vcontext returns 254 on EEXIST 1109542940 M * jd86 Bertl: ok i got one with 2 for a xid on there 1109542966 M * Bertl prae: if you want to start a process as if it was started from inside a vserver, you should look at the vserver exec and suexec commands 1109542993 M * Bertl jd86: okay, already on that partition? 1109543002 M * jd86 huh 1109543013 M * jd86 the vserver is on that partition..... if thats what ur guessing 1109543016 M * Bertl (your vserver, between the lines) 1109543038 M * Bertl just read my lines prefixed with jd86, okay? ;) 1109543043 M * jd86 yep 1109543066 M * jd86 i had read yours but i didnt know what you meant 1109543070 M * jd86 i got a vserver on there, yes 1109543084 M * Bertl okay, start and stop it twice to make sure everything is working as expected 1109543111 M * jd86 heh, cant do that exactly.. it gives me icky error (i just realized it, maybe i should create another one in the meantime) 1109543113 M * jd86 proc error 1109543130 M * Bertl hmm, you should configure your proc security I guess ;) 1109543135 M * jd86 about how /proc must be mounted 1109543142 A * jd86 doesnt remember doing this last time 1109543144 M * jd86 how do i do that then? 1109543164 M * Bertl usually the vprocunhide script does that on bootup 1109543178 M * jd86 ooooh hold on 1109543188 M * Bertl http://linux-vserver.org/Proc-Security 1109543210 M * jd86 i have a /etc/init.d/vservers on my other computer but not on this new one... what is that part of? 1109543225 M * Bertl util-vserver (alpha tools) 1109543237 M * jd86 well i have 204.. 1109543244 M * jd86 does it normally plop it right in there? 1109543248 M * Bertl rpm -ql util-vserver-sysv 1109543248 M * Bertl /etc/rc.d/init.d/vprocunhide 1109543248 M * Bertl /etc/rc.d/init.d/vservers-default 1109543248 M * Bertl /usr/lib/util-vserver/vserver-wrapper 1109543248 M * Bertl /usr/lib/util-vserver/vsysvwrapper 1109543251 M * rs re 1109543258 M * Bertl welcome back rs! 1109543267 M * rs evening bertl! 1109543271 M * rs how are you ? 1109543272 M * jd86 hrm 1109543275 M * Doener jd86: maybe you didn't specify a prefix? 1109543283 M * Bertl rs: fine, thanks! 1109543285 M * Doener try /usr/local/etc/init.d 1109543297 M * Bertl rs: and you? got around some testing? 1109543302 M * jd86 Doener: i went gentoo style, i dont have a /usr/local/etc/ 1109543325 M * Doener so you're using the ebuild, right? 1109543337 M * jd86 i used one that i switched from the old ver to 204 1109543358 M * rs no, I no longer can access the dev platform from outside the network for moment 1109543371 M * Bertl ah, security measures? 1109543391 M * Bertl so that evil rs can't work from home?! ;) 1109543393 M * Doener the .204 ebuild contains that initscript...even a IMHO heavily improved version 1109543404 M * rs no, network changes, my old access are no longer valid because IPs have changed :) 1109543404 M * jd86 wheres the .204 ebuild? 1109543411 M * jd86 is there one now? 1109543416 M * rs it will take some days to fix 1109543421 M * prae Bertl: but, vcontext has a specificity to run prog into MAIN under xid (ex: i have lsof into MAIN, not under vserver); 1109543425 M * prae Bertl: Will this modification prevent to me to do this? 1109543429 M * jd86 argh 1109543432 A * jd86 slaps self silly. 1109543505 M * Bertl prae: well, no, vcontext will work as usual, but context will as you noticed, not put you completely into the vserver, just the process isolation is done with this command 1109543627 M * prae ok, I see that (when I execute "ls" with vcontext, I see my files into MAIN ) 1109543661 M * prae it's good for me :) 1109543743 M * prae good works :) 1109543778 M * jd86 ok 1109543781 M * jd86 here we go :-) 1109543872 M * jd86 when i try to enter a vserver on there, it says 'vserver ... suexec' is support for running vservers only; aborting... 1109543897 M * jd86 oh i guess its not really starting 1109543899 M * jd86 ugha 1109543964 A * jd86 sighs 1109543987 M * jd86 ok whats "RTNETLINK answers: File exists" mean? 1109544072 M * jd86 oh my its really trashed now 1109544163 M * Bertl it means that your ip already is assigned 1109544171 M * jd86 eh and that means? 1109544172 M * Bertl (check with ip addr ls) 1109544175 M * jd86 this is totally trashed 1109544202 M * Bertl the problem is/was probably caused by the badly configured proc in the first place 1109544242 M * jd86 when i tell it to start all it says is Caching service dependencies... and thats about all 1109544283 M * Bertl well, that's a gentoo specific issue 1109544301 M * Bertl but I heard that removing some files in a /var/run* dir can help here 1109544338 M * jd86 alright 1109544343 M * jd86 this is mess. 1109544363 M * jd86 i trashed my tty1 1109544373 M * Bertl 'reset' 1109544386 M * Bertl (the command not the button ;) 1109544386 M * jd86 no like it has something when i tried to start it 1109544394 M * jd86 and its like press ctrl d to or enter password for maintence 1109544401 M * prae Bertl, jd86 and Doener, good night :) 1109544405 M * jd86 night 1109544407 M * Doener night prae! 1109544412 M * Bertl night prae! and thanks for the help! 1109544441 M * prae thanks to you :) 1109544447 M * prae see'ya tomorrow 1109544461 Q * prae Quit: Pwet 1109544630 A * jd86 slaps self. UGH. 1109544646 M * jd86 whats vshelper 1109544654 M * Doener Bertl: what error code shall we use for "no dynamic xid available"? 1109544829 M * jd86 ok i'm having major prob with this vshelper thing 1109544837 M * jd86 i added a thing to /etc/sysctl.conf and it is still not working. 1109544866 M * Doener did you reload the sysctl settings? 1109544871 M * jd86 saying the configured vshelper /usr/lib/util-vserver/vshelper does not match the 'vshelper' script of the util-vserver package. 1109544874 M * jd86 i rebooted 1109544890 M * jd86 i mean what other vshelper is there? 1109544899 M * Bertl well, probably the script is right, and the scripts don#t match ;) 1109544908 M * jd86 well wheres the other 1109544915 M * Bertl util-vserver keeps a separate copy 1109544916 M * Doener should read: kernel.vshelper = /usr/lib/util-vserver/vshelper 1109544921 M * Doener in sysctl.conf 1109544950 M * jd86 i have kernel.vshelper = "/usr/lib/util-vserver/vshelper" as the script told me to put in there 1109544954 M * Bertl hmm, usually it's in /sbin/vshelper, no? 1109544955 M * Doener then run sysctl -p to set the values 1109544965 M * Doener Bertl: not in the gentoo ebuild 1109544976 M * Doener jd86: with quotes it doesn't work IIRC 1109544976 M * jd86 sysctl -p shows the stuff on the screen 1109544995 M * jd86 now how do i update it? 1109544997 M * Doener should also set them 1109544999 M * Bertl i.c. where do they put hotplug? (just out of curiosity) 1109545019 M * Doener /sbin 1109545046 M * jd86 ARGH 1109545048 M * jd86 I'M GOING TO KILL SOMETHING. 1109545051 A * jd86 breathes in. 1109545054 A * jd86 breaths out 1109545059 M * Bertl good ;) 1109545092 A * Doener hides 1109545127 M * jd86 ugh it went so smooth on the other boxy 1109545132 M * jd86 this one is just jelous 1109545198 M * Bertl very likely! 1109545246 M * jd86 werll the other box is entirely ~x86 and this is x86 1109545266 M * Bertl sparc or what? 1109545271 M * jd86 ok i'm still getting this RTNETLINK answers: File exists. 1109545274 M * jd86 huh? 1109545278 M * jd86 no ones all unstable 1109545281 M * jd86 and others stable (gentoo thing) 1109545287 M * Bertl ah, okay ;) 1109545299 M * jd86 so what do i do about this RTNETLINK thing? 1109545308 M * Bertl what ip is assigned to that context/vserver? 1109545315 M * jd86 192.168.0.60 1109545331 M * Bertl this is the one in the config? 1109545337 M * jd86 yessum 1109545344 M * jd86 how do i check whats it in? 1109545345 M * jd86 just to make sure 1109545347 M * Bertl when do you get the RTNETLINK message? 1109545356 M * jd86 when starting it 1109545358 M * jd86 and it does nothing more 1109545388 M * Bertl could you paste the output of 'vserver --debug start' to pastebin.com (or similar) 1109545412 M * jd86 plz hold 1109545438 M * jd86 it wont let me out it to a file? grwl 1109545461 M * Bertl > /tmp/file 2>&1 1109545498 M * jd86 http://jim.psunit.com:989/~jim/output 1109545540 M * Bertl hmm, well, close, but no banana! 1109545551 M * jd86 what you couldnt see the bannas? 1109545562 M * jd86 or could you see the site? 1109545573 M * Bertl yes, but look at it yourself ;) 1109545582 M * jd86 ehh 1109545588 M * jd86 looks like a quote from shakespere... 1109545639 M * jd86 whats wrong? 1109545648 M * Bertl (somehow you managed to get a usage message, that's all I see) 1109545661 M * jd86 oh my h/o 1109545662 M * jd86 lol 1109545672 M * jd86 huh? 1109545675 M * jd86 usage message? 1109545678 M * jd86 i dont know what your talkin bout 1109545688 M * Bertl Usage: /usr/sbin/vserver {start|stop|suexec|restart|condrestart|exec|enter|chkconfig|running|status} 1109545695 M * jd86 at the bottom 1109545700 M * Bertl that's what your dump shows ... 1109545700 M * jd86 i thought u meant i did the debug thing wrong 1109545706 M * jd86 wait is that all? 1109545712 M * Bertl yes, you probably did it wrong ;) 1109545715 M * jd86 noo its a long thingy 1109545726 M * jd86 o 1109545727 M * jd86 oh 1109545728 M * jd86 stupid me 1109545731 M * jd86 har har har 1109545777 M * micah Bertl: I have tried to apply the newest vserver 1.9.4 patches against the Debian 2.6.8 kernel and there are a lot of rejects, so I think it might take some work to fix them :P 1109545781 M * jd86 reload Bertl 1109545833 M * Bertl micah: right, I saw them Snow-Man provided me with a debian patch ;) 1109545836 M * micah Bertl: I was thinking of trying to apply the vserver patches to the debian kernel before any of the debian patches were applied and then apply each one of the debian patches at a time to see if I could tell where the problems was 1109545845 M * micah Bertl: ah, the one for 2.6.10 1109545858 M * Bertl no, 2.6.8-13 or so 1109545863 M * jd86 Bertl: i updated the output refreshz0r :) 1109545867 M * micah Bertl: oh! Snow-Man made 2.6.8 one?! 1109545873 M * micah that is great news 1109545882 M * Bertl micah: no, just the debian aptches in one patch ;) 1109545885 M * Bertl jd86: tx 1109545927 M * micah Bertl: ah, ok, so there is no vserver patch that works with 2.6.8-13 now 1109545947 M * Bertl jd86: okay, the vserver script runs fine, the messages come from your gentoo guest .. you probably need to remove the netowrking stuff there 1109545971 M * Bertl micah: not yet ... 1109545973 M * jd86 Bertl: remove.. i mean its not in the runleve.... 1109546146 M * jd86 ok i got it to start 1109546161 M * jd86 but whats this mean: /sbin/rc: Line 7: is_vserver_gues: command not found. 1109546162 M * jd86 OOH 1109546163 M * jd86 i know what that is 1109546171 M * jd86 stupid thing i tried 1109546176 M * jd86 oh shite 1109546180 M * jd86 it didnt actually start 1109546185 M * jd86 waaa 1109546210 J * Ruzgar ~Lock@85.97.90.179 1109546215 T * Ruzgar Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com Irc.Zurna.Com 1109546216 P * Ruzgar 1109546250 M * jd86 wow hes as cool as ... well a dog shitting on ur foot 1109546261 M * jd86 do i have ot have proc in my /etc/fstab on the guest? 1109546343 M * jd86 argh. 1109546353 T * Bertl http://linux-vserver.org/ | latest stable 1.2.10, devel 1.9.4.8, ng9.0 -- He who asks a question is a fool for a minute; he who doesn't ask is a fool for a lifetime -- share the gained knowledge on the wiki, and we'll forget about the minute ;) 1109546444 F * ChanServ +o Bertl 1109546489 M * Bertl 85.97.90.179 1109546501 F * Bertl +b *!*@85.97.90.179 1109546560 F * Bertl -o Bertl 1109546585 M * Bertl jd86: proc should be 'automatically' mounted 1109546619 J * ntrs ntrs@Dardeene-68.188.50.87.charter-stl.com 1109546658 M * jd86 i'm getting that rtnet link thing 1109546659 M * jd86 more. 1109546725 M * jd86 whats the RTNETLINK have to do with anything? 1109546758 M * jd86 i think its got to do with proc not being automatically mounted. 1109546760 M * Bertl that is if you do 'ip addr ...' stuff 1109546775 M * Bertl ip (from iproute2) uses netlink to communicate 1109546780 M * Bertl welcome ntrs! 1109546858 M * jd86 Bertl: well i dont think rc will run at all without proc being mounted. 1109546914 M * jd86 i know what i'll do. 1109546987 M * Bertl ++ /usr/lib/util-vserver/secure-mount -a --chroot --fstab /etc/vservers/gentoo-template/fstab --rootfs no 1109547006 M * Bertl (this is executed by the script and it should mount all your stuff from fstab) 1109547704 Q * jd86 Read error: Operation timed out 1109548130 M * Bertl Doener: I found a way to fix it up completely! 1109548149 M * Doener woohoo :) 1109548156 M * Bertl (and now I understand what enrico meant ;) 1109548185 M * Bertl well, took just half a year ;) 1109548303 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-1.9.5/delta-create_info-feat02.diff 1109548329 M * Bertl you can now enter static and dynamic context with chcontext ;) 1109548603 M * Doener Bertl: hm, that's basically the EINVAL -> EEXIST change, right? 1109548616 M * Bertl yes, but in both cases, for static and dynamic 1109548638 M * Bertl before dynamic was rejected with EINVAL 1109548664 M * Bertl now it depends on the xid, if it exists, it's rejected with EEXIST, otherwise with EINVAL 1109548676 M * Doener ah, i c 1109548700 M * Bertl now I'm lookin into putting back the SETUP check