1114905718 M * Doener ok, for my purposes (tabsize in kernel code) two autocmd on BufEnter seem to be the best solution... 1114905769 M * Doener it basically keeps me from using ts!=8 on kernel stuff and ts!=4 for everything else, but that's just fine for me 1114906426 J * prae ~prae@sherpadown.net 1114906524 M * Bertl Doener: yup ... agreed ;) 1114907385 N * X MrX 1114907402 M * Bertl morning MrX! 1114907413 M * Bertl Doener: anything which keeps us from going rc1 ? 1114907473 M * Doener no, don't think so 1114908024 M * Doener Bertl: btw, what's the reason for keeping all the original accounting code as comments? 1114908032 M * Doener like this: 1114908033 M * Doener ->------>-------vma->vm_mm->locked_vm += grow; 1114908033 M * Doener +>------>-------// vma->vm_mm->locked_vm += grow; 1114908034 M * Doener +>------>-------vx_vmlocked_add(vma->vm_mm, grow); 1114908065 A * Doener .oO( shouldn't use direct xterm copy'n'paste with vim... ) 1114908066 M * Bertl hehe .. reason is simple, this way I see which was handled 1114908118 M * Bertl (well, and how the original looked like) 1114908191 M * Doener ok 1114908980 M * Doener ah, i guess i finally found out what the proc_doutstring entry in your todo list means... 1114908991 M * Bertl yeah? ;) 1114909000 M * Doener the /proc/sys/kernel/osrelease and friends aren't virtualized 1114909015 M * Doener and proc_doutstring is the function that is used to handle those 1114909024 M * Doener i _guess_ this is related somehow ;) 1114909033 M * Bertl very likely ;) 1114909046 M * Bertl wanna address this? 1114909052 M * Doener %s/ut/uts/ 1114909059 M * Doener i'm already at it :) 1114909071 M * Bertl k, then I wait with -rc1 ;) 1114910445 M * Bertl ARE NOW ANY DEBIAN FOLKS AROUND? ;) 1114910837 M * Bertl *sigh* 1114910905 M * Doener hm, what kind of debian folk do you need? 1114911003 M * Bertl some which can tell me how to rebuild a f*cking source package ... 1114911041 M * Doener debian sources? 1114911044 M * Bertl http://packages.debian.org/testing/source/dietlibc 1114911054 M * Bertl given this, how do I get a .deb? 1114911099 M * Doener dpkg-buildpackage -rfakeroot -uc -b 1114911111 M * Bertl where when and how? 1114911151 M * Bertl in th dir, where those three files are? 1114911183 M * Bertl do I have to unpack them? apply the patch? 1114911243 M * Doener ah, you can download them via "apt-get source dietlibc"... if you just got the files, i guess untar+patch should do the same 1114911259 A * Bertl really doesn't know why he is doing debian work *big* *sigh* 1114911292 M * Bertl okay, np, so I have a dir, with the 'debian' directory 1114911300 M * Bertl now what? 1114911341 M * Doener in the directory above debian/ (dietlibc-0.28 or so i guess) enter the above command 1114911353 M * Bertl /usr/bin/dpkg-buildpackage: line 175: fakeroot: command not found 1114911374 M * Doener apt-get install fakeroot 1114911388 M * Doener and: apt-get build-dep dietlibc 1114911397 M * Doener the latter one installs all the build dependencies 1114911456 M * Bertl /usr/bin/fakeroot: line 150: debian/rules: Permission denied 1114911557 M * Doener sec... 1114911567 M * daniel_hozac chmod +x? 1114911578 M * Bertl for debian/rules? 1114911588 M * daniel_hozac i would assume so. 1114911602 M * Bertl hmm, yup, helps 1114911605 M * daniel_hozac (i'm not really a Debian guy myself) 1114911616 M * Bertl LOL 1114911680 M * Bertl FATAL: can't create bin-sparc64/start.o: No such file or directory 1114911714 M * Bertl nevertheless the same version is installed ;) 1114911776 M * Doener ok, i just looked up how to properly unpack those files... 1114911786 M * Bertl k, let's hear ... 1114911801 M * Doener dpkg-source -x dietlibc_0.28-3.dsc 1114911815 M * Bertl okay, worked 1114911832 M * Doener ok, let's see if it builds now... 1114911838 M * Bertl now the buildpackage? what does -uc do? 1114911862 M * Bertl -b = --build 1114911911 M * Doener -uc == don't sign the .changes file... 1114911921 M * Doener whatever that means... i just learned to use it that way ;) 1114911936 M * Bertl sorry that I'm asking ;) 1114911954 M * Bertl FATAL: can't create bin-sparc64/start.o: No such file or directory 1114911972 M * Doener interesting... 1114911986 M * Bertl so it doesn't build on sparc64, although the binary does exist ;) 1114912164 M * Bertl is the fakeroot essential, or could I build without? 1114912175 M * Doener as root you can probably build without... 1114912402 M * Bertl hmm, yup, but same error ... 1114912424 M * Doener hm, the bin-sparc64 directory should have been created, right? 1114912495 M * Doener Bertl: which gcc version do you have there? 1114912658 M * Doener hmm... there's no sparc64 dietlibc package in the debian repository AFAICT... only a sparc package... 1114912988 M * Bertl yup, it's 32bit userspace on a 64bit machine ... 1114912999 M * Bertl gcc --version 1114912999 M * Bertl gcc (GCC) 3.3.5 (Debian 1:3.3.5-8) 1114913072 M * Doener ok, for a 32bit package do: ARCH=sparc dpkg-buildpackage ... 1114913175 M * Bertl FATAL: can't create bin-sparc64/start.o: No such file or directory 1114913216 M * Doener hmm... this is interesting... the same line tries to compile for sparc on my x86 box... 1114913245 M * Bertl it does here too .. emphasis on _tries_ ;) 1114913308 M * Doener hm, it should use bin-sparc then... not bin-sparc64 1114913328 M * Bertl /usr/bin/make dyn prefix=/usr/lib/diet \ 1114913328 M * Bertl MYARCH='sparc' VERSION='0.28-3' 1114913347 M * Bertl touch build-stamp 1114913348 M * Bertl fakeroot debian/rules binary 1114913356 M * Bertl .. 1114913361 M * Bertl make[1]: Entering directory `/root/DIET/dietlibc-0.28' 1114913365 M * Bertl gcc -I. -isystem include -pipe -nostdinc -m64 -Os -Wall -W -Wchar-subscripts -Wmissing-prototypes -Wmissing-declarations -Wno-switch -Wno-unused -Wredundant-decls -c sparc64/start.S -o bin-sparc64/start.o 1114913386 M * Bertl which is _definitely_ a sparc64 compile ;) 1114913606 M * Bertl btw, the dietlibc itself compiles fine on that machine ... 1114913746 M * Doener next try: dpkg-buildpackage -asparc -uc -b -rfakeroot 1114913857 M * Bertl FATAL: can't create bin-sparc64/start.o: No such file or directory 1114913874 M * Doener ok, i give up... this doesn't make sense... 1114913896 M * Bertl it happens in the install path 1114913911 M * Bertl /usr/bin/make install prefix=/usr/lib/diet DESTDIR='/root/DIET/dietlibc-0.28/debian/dietlibc' 1114914002 M * Bertl trying now with modified rules 1114914086 M * Bertl ahh, works! 1114914109 M * Bertl so nobody tried before to build it on a 64bit machine with 32bit userspace ... 1114914147 M * Bertl well, scratch the part from 'on' till 'userspace' 1114914199 M * Bertl okay, thanks a lot for the help! 1114914567 M * Doener you're welcome 1114914589 A * Doener is thinking hard how to tackle the uts stuff... 1114914697 M * Bertl hum hum, debian uses reversed -p0 patches?! 1114914889 M * Doener hm, -p0 yes. reversed, not that i could see... 1114914913 M * Bertl yeah, well, seems it is doing strange stuff with the sources ... 1114914923 M * Doener if you're referring to the debian/diffs/* patches, those are applied once on the first build 1114914947 M * Doener the patch-stamp file is touched then and used to verify whether the sources are already patche 1114914950 M * Doener d 1114914960 M * Bertl ah, okay, that probably explains it 1114915177 M * Bertl so I can fix something in the sources (patched ones) and rebuild? 1114915209 M * Doener not sure... i had some troubles there IIRC... 1114915256 M * Bertl hmm, yeah, that messes up the sources completely :( 1114915263 M * Bertl back to the start ... 1114915303 M * Doener to be sure, create a patch, put it into the diffs directory and build then 1114915336 M * Bertl yup, I'm pretty sure that's not the smartest way to do it .. but yes ... 1114915382 M * Bertl and I did complain about the complicated rpm development ... 1114915411 M * Bertl no wonder ola doesn't want to apply patches/do fixes ;) 1114915425 M * Doener there are a lot helper tools for debian package maintance... but i don't know any of them ;) 1114915450 M * Bertl damn I forgot to save the fixed rules 1114915861 M * Doener ok, building qemu kernel for the uts fix/hack ;) 1114916963 M * Doener http://www.13thfloor.at/~doener/vserver/patches/diff-2.6.11.7-vs2.0pre3-proc_doutsstring.diff 1114917019 M * Doener compiles, boots and works as expected (in qemu) 1114917072 M * Bertl hmmm ... that requires a lot of stack ... 1114917091 M * Doener yep... but i didn't find any better way to do it 1114917288 M * Doener hmm... i could simply move the switch to proc_dostring i guess... 1114917346 M * Bertl ahem, better not ;) 1114917382 M * Doener ah right... it does more than kern ctl stuff... OOPS! 1114917412 M * Bertl hmm ... hmm ... 1114917444 M * Bertl it shows that we have to figure some way (in the future?) to virtualize sysfs entries 1114917541 M * Bertl what about the 'write' case? 1114917545 M * Doener hehe... that's why it took me so long to come up with anything... 1114917635 M * Bertl I guess the best way would be to provide an optional function, which fills in the data/size 1114917902 M * Bertl ABI_FAKE_UTSNAME <- interesting ;) 1114917942 M * Doener yep, but unused as it seems 1114917947 M * Bertl yup 1114918062 M * Doener uts stuff is not writeable 1114918073 M * Bertl what about 1114918074 M * Bertl typedef int virt_handler (ctl_table *ctl, int write, struct file * filp, 1114918074 M * Bertl void **datap, size_t *lenp); 1114918139 M * Doener you can change the permissions so f.e. "echo 123 > /proc/sys/kernel/osrelease" doesn't complain, since the opening suceeds, but the actual write still fails, because the ctl_table permissions are taken into consideration 1114918154 M * Bertl and ... 1114918155 M * Bertl virt_handler *virtualize; /* Context Virtualization */ 1114918179 M * Bertl then change the proc_dostring() to call that, if it is defined? 1114918234 M * Doener hmm... basically we would just want to virtualize the data part, right? 1114918234 M * Bertl then make a handler which contains your switch() 1114918263 M * Bertl well, AFAICT the data and len are all which ever might change 1114918286 M * Bertl the write and filp are provided to figure the context and operation 1114918296 M * Doener at least, those are the only ones used by proc_dostring 1114918395 M * Doener what about: typedev virt_handler(ctl_table *ctl, void **datap, size_t *lenp); 1114918409 M * Doener and have two local variables in proc_dostring 1114918433 M * Bertl that's the idea, but the function might need to know the read/write 1114918435 M * Doener the virt_handler, if set, sets the data and lenght... 1114918445 M * Bertl and it also might need to know the xid ... 1114918474 M * Doener xid could be fetched from current, couldn't it? 1114918490 M * Bertl probably 1114918504 M * Doener why would it need the read/write= 1114918519 M * Bertl because it could pass a dummy address for writes? 1114918611 M * Bertl what about xid_t xid, int write ? 1114918632 M * Bertl or the other way round ... 1114918659 M * Bertl data and maxlen local variables as you said 1114918716 M * Doener ok 1114918816 M * Doener hm... where would i put vx_uts_virt_handler(...)? 1114918825 M * Doener kernel/vserver/cvirt.c? 1114918832 M * Bertl yup 1114918993 Q * MrX Quit: Killed Terminal 1114919179 M * Doener hmm... do we need it to return int anymore? 1114919207 M * Bertl hmm, maybe as error check, but basically no 1114919224 M * Doener i'll have it return 0 for now... 1114919233 M * Bertl okay ... 1114920017 M * Doener ok, compiling... 1114920357 M * Bertl chcontext --xid 100 -- uname -a 1114920358 M * Bertl New security context is 100 1114920359 M * Bertl Linux ankaa 2.6.11.8-vs2.0-pre3 #6 SMP Sat Apr 30 23:12:06 CEST 2005 sparc64 GNU/Linux 1114920373 M * Bertl yay! 1114920378 M * Doener woohoo :) 1114920469 M * Bertl but (and this sounds really familiar) 1114920471 M * Bertl chbind --ip 127.0.0.1 -- true 1114920471 M * Bertl Bus error 1114920500 M * Doener yep... 1114920530 M * Bertl but ... now we have some debug info ... 1114920541 M * Bertl $ chcontext --xid 100 -- true 1114920547 M * Bertl vxD: vc: VCMD_00_0[0], 63,0000000000000000,1 1114920547 M * Bertl vxD: vc: VCMD_63_1[1], 100,00000000effffc20,1 1114920562 M * Bertl $ chbind --ip 127.0.0.1 -- true 1114920569 M * Bertl Bus error 1114920578 M * Bertl that's it ... so it happens in userspace ;) 1114920642 M * Bertl Program received signal SIGBUS, Bus error. 1114920642 M * Bertl 0x000110f4 in gethostbyname_r () 1114920776 M * Bertl uhum .. libcruft ;) 1114920899 M * Doener ok... all syntax errors fixed (i hope ;) 1114920909 M * Doener now waiting for the build to finish... 1114920929 M * Doener btw, did you hear anything from timo regarding the status/outcome of his project? 1114920936 M * Bertl btw, you probably compile on an UP system, no? 1114920949 M * Doener yep 1114920971 M * Bertl if you specify up to -j3 it speeds up compilation if you have a hot file cache ... 1114921012 M * Doener i did some "time make" runs some time back, that said that -j2 is the fastest for me 1114921032 M * Bertl hmm, timo sent me some stuff, IIRC in a proprietary format I didn't want to read ... 1114921098 M * Bertl ah, no ... 1114921183 M * Bertl well, he wanted to send me something ;) 1114921226 M * Doener ok, kernel booted... 1114921361 M * Bertl excellent! 1114921443 M * Doener and it works, too! :) 1114921455 M * Bertl even better! 1114921478 M * Bertl so let's have a look at the masterpiece ... 1114921554 M * Doener http://www.13thfloor.at/~doener/vserver/patches/diff-2.6.11.7-vs2.0pre3-proc_doutsstring-2.diff 1114921673 M * Bertl nitpiking: len, maxlen 1114921739 M * Doener in vx_uts_virt_handler, right? 1114921765 M * Bertl hmm, *not there*, in proc_dostring() 1114921786 M * Doener hm, in vx_uts_... it's maxlen <> lenp 1114921823 M * Doener what is it in proc_dostring()? 1114921832 M * Doener ah... should go on one line, right? 1114921848 M * Bertl yup, and the virt_handler part has to move up ... 1114921872 M * Bertl if (!table->data || !table->maxlen || !*lenp || 1114921873 M * Bertl (*ppos && !write)) { 1114921881 M * Doener yup 1114921897 M * Bertl btw, I guess it would not hurt to do: 1114921905 M * Bertl data = table->data; 1114921909 M * Bertl maxlen = table->maxlen; 1114921924 M * Bertl if (table->virt_handler) 1114921930 M * Bertl table->virt_handler(table, write, filp->f_xid, &data, &maxlen); 1114921947 M * Doener k 1114921955 M * Bertl this would save you the lenp assignment ... 1114921974 M * Bertl and give a sane default for the data if the handler doesn't act 1114922000 M * Bertl another option would be to explicitely zero them out 1114922217 M * Doener ok, building... 1114922233 M * Doener booting... 1114922310 M * Doener testing... 1114922446 M * Doener works... uploading 1114922543 M * Doener http://www.13thfloor.at/~doener/vserver/patches/diff-2.6.11.7-vs2.0pre3-proc_doutsstring-3.diff 1114922567 M * Bertl looks good ... 1114922586 M * Bertl thanks a lot! 1114922591 M * Doener my pleasure! 1114922606 M * Doener so we can finally remove that todo list entry... 1114922619 M * Doener took only half a year to figure out what it was about :) 1114922627 M * Bertl yup, go ahead, get rid of it! you earned it! 1114922749 M * Doener to the wiki-history it went... ;) 1114922925 M * Bertl uh-oh ... 1114922930 M * Bertl result->h_addr_list=(char**)(buf+strlen(name)+1); 1114922931 M * Bertl result->h_addr_list+=sizeof(unsigned long)-((unsigned long)(result->h_addr_list)&(sizeof(unsigned long)-1)); 1114922934 M * Bertl result->h_addr_list[0]=(char*)&result->h_addr_list[2]; 1114922941 M * Bertl doesn't look very 'portable' to me ... 1114922992 M * Doener doesn't look readable to me ;) 1114923006 Q * prae Quit: Pwet 1114923161 M * Bertl well, I guess it's 'supposed' to do some padding/alignment ;) 1114923206 M * Bertl but assuming that sizeof(char *) == sizeof(unsigned long) is a little .. hmm bold 1114923264 M * Doener hm.. aren't pointers as big as longs? 1114923283 M * Doener at least i've heard rumours about that... ;) 1114923715 M * Bertl hmm .. any evidence? 1114923806 M * Doener hm, i just found some documents saying that before C99 there was no relation at all between them, and with C99... well... opinions seem to differ ;) 1114923825 M * Doener so you're right 1114923847 M * Bertl but actually that seems not to be the issue here ... 1114924164 M * Bertl hehe, this is tricky ... 1114924257 M * Bertl you know what long *x = NULL; x+=2; printf("%d\n",x); gives? 1114924274 Q * ntrs_ Ping timeout: 480 seconds 1114924276 M * Bertl (except for a few warnings) 1114924293 M * Doener hm, on 64bit probably 16... 1114924310 M * Doener on 32bit 8 1114924320 M * Bertl yup, and on 32bit 8 .. and what did the dietlibc folks intend? 1114924336 M * Bertl to adjust for alignment! LOL 1114924422 M * Doener it's a system library, it _has_ to be complicated ;) 1114924771 A * Bertl wonders if dietlibc has been _ever_ tested on non x86 archs ... 1114924783 M * Bertl src/chbind --ip 127.0.0.1 -- true 1114924783 M * Bertl ipv4root is now 127.0.0.1 1114924831 M * Doener great! 1114924863 M * Bertl dietlibc-0.28]$ grep -r 'sizeof(unsigned long)' * | wc 1114924863 M * Bertl 24 184 1724 1114924921 M * Bertl but luckily it seems that this was the only palce where they played those alignment tricks ... 1114925009 M * Bertl Linux-VServer Test [V0.12] (C) 2003-2005 H.Poetzl 1114925009 M * Bertl chcontext is working. 1114925009 M * Bertl chbind is working. 1114925009 M * Bertl Linux 2.6.11.8-vs2.0-pre3 sparc64/0.30.207/0.30.207 [Ea] 1114925009 M * Bertl VCI: 0001:0025 267 03110076 1114925011 M * Bertl --- 1114925014 M * Bertl [000]# succeeded. 1114925016 M * Bertl [001]# succeeded. 1114925019 M * Bertl [011]# succeeded. 1114925021 M * Bertl [031]# succeeded. 1114925024 M * Bertl [101]# succeeded. 1114925026 M * Bertl [102]# succeeded. 1114925029 M * Bertl [201]# succeeded. 1114925031 M * Bertl [202]# succeeded. 1114925034 M * Bertl yay! 1114925064 M * Doener we definately had a good night :) 1114925081 M * Bertl yup! time for breakfast and then bed ... 1114925086 M * Bertl (at least for me ;) 1114925100 M * Doener heh :) enjoy your meal... and sweet dreams! 1114925633 M * Bertl http://vserver.13thfloor.at/Experimental/delta-diet.diff (just for the record, probably fixes x86_64 and sparc/64) 1114925890 M * Doener ok, doesn't tell me anything ;) didn't manage to learn any assembler yet 1114925922 M * Bertl heh, I'm learning a new assembler every second day ;) 1114926057 M * Doener hm? 1114926128 M * Bertl well, it's not so that every arch is completely new ... 1114926154 M * Doener true, but don't you run out of archs to learn pretty soon? 1114926182 M * Bertl if you know how a cpu works .. and how assembler is written ... it's basically a question of having a manual at hand ... 1114926209 M * Bertl and no, they are very inventive in creating 'new' old cpus ;) 1114926226 M * Doener heh 1114926289 M * Bertl what really is fascinating is the 32 vs 64 bit address space 1114926331 M * Bertl if you consider it, for a 64bit arch, the entire 32bit address space is as laarge as a single byte ;) 1114926406 M * Doener yeah, but getting 1GB of memory in that scaling is pretty expensive then ;) 1114926442 M * Bertl let's see in 10 or 20 years ;) 1114926553 M * Doener ok, i'll have breakfast now, too... 1114926556 M * Bertl okay, I'm off to bed now ... 1114926557 M * Doener cya later 1114926569 M * Doener have a good 'night' Bertl! 1114926573 M * Bertl may you all live in interesting times! ;) 1114926578 N * Doener Doener|gone 1114926585 N * Bertl Bertl_zZ 1114934028 J * erwan_ho ~erwan@lns-vlq-39f-81-56-133-136.adsl.proxad.net 1114935679 Q * Medivh jupiter.oftc.net arion.oftc.net 1114935679 Q * DaPhreak jupiter.oftc.net arion.oftc.net 1114935679 Q * locksy jupiter.oftc.net arion.oftc.net 1114935679 Q * albeiro jupiter.oftc.net arion.oftc.net 1114935679 Q * micah jupiter.oftc.net arion.oftc.net 1114935679 Q * pusling jupiter.oftc.net arion.oftc.net 1114935779 J * locksy ~locksy@mrtg.sisgroup.com.au 1114935779 J * DaPhreak ~phreak@lms.rz.uni-greifswald.de 1114935779 J * albeiro albeiro@albeiro.usercloak.oftc.net 1114935779 J * pusling ~pusling@195.215.29.124 1114935779 J * micah micah@micha.hampshire.edu 1114935779 J * Medivh ck@paradise.by.the.dashboardlight.de 1114935784 Q * DaPhreak Read error: Connection reset by peer 1114935794 J * DaPhreak ~phreak@lms.rz.uni-greifswald.de 1114935805 J * _Medivh ck@paradise.by.the.dashboardlight.de 1114935815 Q * Medivh Read error: Connection reset by peer 1114935831 Q * locksy Ping timeout: 480 seconds 1114935836 J * locksy ~locksy@mrtg.sisgroup.com.au 1114941029 J * DuckMaster ~Duck@dyn-83-157-154-66.ppp.tiscali.fr 1114941317 Q * DuckKing Read error: Operation timed out 1114941793 J * kalou__ ~kalou@ABoulogne-102-1-4-72.w193-253.abo.wanadoo.fr 1114941910 Q * kalou Ping timeout: 480 seconds 1114945135 J * ComplexHo ~ComplexHo@funk.gotadsl.co.uk 1114947918 J * atsab_ ~as@lotes.vtu.lt 1114947918 Q * atsab Read error: Connection reset by peer 1114949772 J * ntrs ~ntrs@Dardeene-68.188.50.87.charter-stl.com 1114950070 J * atsab ~as@lotes.vtu.lt 1114950087 Q * ntrs Read error: Connection reset by peer 1114950480 Q * atsab_ Ping timeout: 480 seconds 1114950780 J * ntrs ntrs@Dardeene-68.188.50.87.charter-stl.com 1114951288 Q * click Read error: Operation timed out 1114956550 Q * erwan_ho Remote host closed the connection 1114956655 Q * bro Ping timeout: 480 seconds 1114956975 M * kalou__ what is vserver chkconfig for ? 1114957003 M * kalou__ I upgraded to latest util-vservers, but the layout of the filesystem seems to have changed a lot 1114957056 M * kalou__ I tried: vserver xfree chkconfig, here is the result: readlink: readlink ("/usr/local/etc/vservers/xfree/run": Invalid argument 1114957085 M * kalou__ 'vserver ... suexec' is supported for running vservers only; aborting ... 1114957091 M * kalou__ rather lost here. 1114957105 M * meebey you should stop the vservers before upgrade 1114957133 M * meebey I have no idea what changed though, but having running vservers without control is bad :-P 1114957139 M * kalou__ ok. This is a Debian based vserver, and chkconfig is Redhat related. 1114957155 M * kalou__ But I'm unable to start it either, with the same error messages: 1114957180 M * kalou__ +: var/lock: No such file 1114957206 M * kalou__ save_ctxinfo: readlink ("/usr/local/etc/vservers/xfree/run"): Invalid argument 1114957236 M * meebey it searches the run state inside the vserver configs? 1114957239 M * kalou__ I had to move /vservers/xfree to /usr/local/etc/vservers/xfree_vdir 1114957247 M * meebey uh yeah I remember that the structure changed 1114957255 M * meebey can't help you on that, sorry 1114957278 M * meebey waaat? 1114957279 M * kalou__ meebey .. ok thank you .. :) Which utils version are u using ? 1114957288 M * meebey you moved the vserver inside the config dir? that sounds insane 1114957300 M * kalou__ probably did something wrong 1114957319 M * meebey maybe a symlink, if the new structure needs that, but not moving :) 1114957325 M * meebey there should be a README 1114957336 M * meebey describing what changed 1114957342 M * kalou__ oki yes of course .. let's try a symlink and go throug the README :-( 1114957347 M * kalou__ silly am I ! 1114957495 M * kalou__ hmm. not many docs about upgrading from stable tools to alpha tools on 2.6 1114957501 M * kalou__ let's try in the wiki 1114957857 N * Doener|gone Doener 1114958043 M * Doener kalou__: you should have a symlink called vdirbase (or sth. like that) /usr/local/etc/vservers/.defaults ... that one should point to /vservers 1114958095 M * kalou__ oki 1114958109 M * Doener and /usr/local/etc/vservers/xfree/vdir should be a symlink to /etc/vservers/.defaults/vdirbase/xfree ... 1114958123 M * Doener but that is only relevant if you have a new style configuration... 1114958135 M * Doener legacy configurations should work ootb 1114958167 M * kalou__ /usr/local/etc/vservers symlinked to /etc/vservers I presume 1114958198 M * Doener probably not... 1114958211 M * Doener by default, the tools use /usr/local as prefix... 1114958230 M * kalou__ ok .. now vserver zoo start complains about : Can not find vserver-setup at /usr/local/vserver/zoo/ 1114958234 M * Doener if you want the config to be in /etc/vservers, have have to set that when configuring the tools 1114958284 M * kalou__ ok. The format of zoo.conf changed between stable and alpha, or can I just keep the conf files ? 1114958306 M * Doener you can keep them... the new configuration scheme works without them 1114958325 M * Doener if you don't have a new config style directory, it is assumed that you're using a legacy config 1114958470 M * kalou__ hmm. I can see test -e /usr/local/etc/vservers/zoo/legacy 1114958704 M * Doener hm, didn't ever see that... maybe the configuration has changed... 1114958720 M * Doener anyway, my gf arrived, got to go... later... 1114958723 N * Doener Doener|gone 1114958979 Q * ComplexHo Quit: Leaving 1114959398 M * SiD3WiNDR Doener|gone: HEY, girlfriends are no reason for leaving IRC! 1114959400 A * SiD3WiNDR giggles 1114959840 Q * monrad Ping timeout: 480 seconds 1114960354 J * monrad ~monrad@213083190130.sonofon.dk 1114963876 J * bro ~vanity@lanparty.lv 1114964196 N * Bertl_zZ Bertl 1114964213 M * Bertl morning folks! 1114964973 J * click click@dsl-static-122-208.aal.tiscali.no 1114965889 M * Bertl welcome click! 1114966754 N * Doener|gone Doener 1114966827 M * Bertl wb Doener! 1114966828 M * Doener evening folks! 1114966860 M * Doener SiD3WiNDR: they definitely _are_ a reason for leaving irc ;) 1114967087 M * Bertl he is just jealous of your 'real' life ;) 1114967443 M * Doener Bertl: did you find anything we'd need for an rc1? 1114967606 M * Bertl well, I consider making a pre4 with some minor changes to the NFS code ... 1114967624 M * Bertl there are _some_ places where we _assume_ unconditional xid tagging 1114967670 J * hillct ~hillct@client200-5.dsl.intrex.net 1114967690 M * hillct hi all 1114967737 M * Doener welcome hillct 1114967775 M * hillct hi Doener 1114967782 M * hillct I have another stupid problem 1114967803 M * Doener Murphy's coming home... ;) 1114967814 M * hillct no 1114967821 M * hillct this is something I did wrong 1114967827 M * hillct and never really tested 1114967830 M * hillct I'm trying to connect from the host server to a vhost running mysqld 1114967862 M * hillct I've granted appropriate privileges to a user running on the IP of the host server 1114967873 M * hillct but I get the mysql error: 1114967885 M * hillct ERROR 1130: Host '66.193.14.121' is not allowed to connect to this MySQL server 1114967900 M * Bertl what about _allowing_ it? 1114967912 M * hillct in mysql? 1114967928 M * Bertl well, yeah or wherever you disallow it ;) 1114967931 M * hillct sure. granted access for user hillct at the IP of the host server 1114967936 M * Doener ah yeah... that is some strange mysql thing... it somehow seems to dislike local connection not made on 127.0.0.1... 1114967949 M * hillct aha 1114967957 M * hillct so it appears to come from localhost? 1114967960 M * hillct sort of 1114968032 M * hillct that's what I'm trying to figure. what host does the connection appear to come from when connecting from host to vhost? 1114968098 M * Doener < hillct> ERROR 1130: Host _'66.193.14.121'_ is not... ;) 1114968106 M * hillct yah 1114968118 M * Doener but mysql seems to notice something somehow... 1114968120 M * hillct which is the host I specifically granted access for 1114968122 M * hillct :) 1114968140 M * hillct yah, there's got to be some strange nuance here... 1114968176 M * hillct because it's clearly not being affected by iptables 1114968186 M * hillct which I initially thought was odd 1114968216 M * hillct but earlier, Bertl said connections from host to vhost are in fact loopbacks 1114968249 M * Bertl they will happen via the loopback device, yes 1114968263 M * Doener yes, they are... but i doubt that mysql can check that the traffic went over lo 1114968270 M * Bertl but they are 'normal' ip traffic, as far as applications are concerned 1114968273 M * hillct the mysql error indicates the connection is being established on the TCP level but are being rejected by the mysql server 1114968287 M * hillct hmm 1114968291 M * Doener well, i'll give it another try... gimme some minutes... 1114968323 M * Bertl (i.e. you can do the same connect from ip via lo on every linux system) 1114968345 M * Doener hillct: hm, what bind-address did you set in my.cnf? 1114968355 M * hillct hmm 1114968362 M * hillct I didn't 1114968370 M * hillct mysqld is running on the vhost 1114968378 M * hillct so it sohuld only be bound to the vhost address 1114968386 M * hillct right? 1114968398 M * hillct I do have a mysqld running on the host box as well 1114968406 M * hillct that one is specifically bound to the host IP 1114968406 M * Doener yes, but at least debian sets it to 127.0.0.1 1114968414 M * Doener maybe mysql does some special check for that... 1114968414 M * hillct hmm 1114968419 A * hillct does a sanity check 1114968519 M * hillct oh cute 1114968523 M * Doener hillct: hmm... take another look at the ip address in the error... 1114968529 M * Doener it's the vserver's ip address 1114968556 M * Doener i guess you allowed access from 66.193.14.120, not 66.193.14.121 1114968565 M * Doener (if i correctly remember your setup) 1114968566 M * hillct duh... 1114968569 M * hillct so it is 1114968576 M * hillct geez 1114968578 M * hillct I suck 1114968612 M * Doener well, i didn't notice either when i tried some time ago... 1114968629 M * Doener i just bind-mounted to socket somewhere back then... 1114968635 M * Bertl most folks do not look at the ugly details (if something goes wrong ;) 1114968685 M * hillct oh. gee. tell us how you really feel ;) 1114968701 A * hillct will pay more attention in future 1114968818 M * Doener btw, is this also something which will be addressed with ngnet... 1114968831 M * hillct cool 1114969430 Q * click Ping timeout: 480 seconds 1114970319 M * romke Bertl: http://www.rafb.net/paste/results/MxVBgs59.html 1114970332 M * romke "kernel BUG at kernel/sched.c:584!" 1114970351 M * Bertl what's 1.2v? 1114970450 M * Bertl anyway, looks to me like this one is missing: 1114970451 M * Bertl http://vserver.13thfloor.at/Experimental/FOR-1.9.6/delta-unhold-fix01.diff 1114970465 M * romke Bertl: its 1.9.5 + 2.6.11.7 1114970485 M * Bertl yup, expected, please use vs2.0-pre3 1114970531 M * romke Bertl: thx, I'll upgrade 1114970534 M * Bertl ah, and btw, there is 2.6.11.8 ;) 1114970700 J * eXplasm ~explasm@p549FF335.dip.t-dialin.net 1114970706 M * eXplasm hi ho :D 1114970729 M * eXplasm in which file can i edit the variable for "prefix: /usr/local" 1114970759 M * eXplasm ? 1114970759 M * eXplasm :> 1114970782 M * Bertl in the config.h file ... 1114970805 M * Bertl but probably better just to specify it as ./configure argument with --prefix 1114970806 M * eXplasm okay, thx 1114970868 M * Bertl something like this: 1114970870 M * Bertl ./configure --prefix=/usr --localstatedir=/var --sysconfdir=/etc --with-vrootdir=/vservers 1114970922 P * hillct 1114971457 M * eXplasm hm, okay 1114971465 M * eXplasm vserver are running 1114971491 M * eXplasm but i ever get: "WARNING: can not find configuration, assuming legacy method" 1114971511 M * eXplasm it that because i have used a very old util-vserver in the past? 1114971515 M * eXplasm had 1114971651 M * Bertl probably, it's a question of the config ... 1114971665 M * Bertl the new config is tree based, where the old (legacy) config is a single file 1114971704 M * Bertl (you can find details about the new config in your util-vserver's doc dir or on the 'flower page') 1114971738 M * Doener Bertl: on that child_reaper todo list entry again... (thanks to #kernelnewbies ;) what exactly is it you/we want to achieve there... 1114971765 M * Doener that the childs get reparented to the vserver's init process if there is one instead of the host's init process? 1114971816 M * Bertl hmm, they should get now ... 1114971855 M * Bertl IIRC the idea was to reparent processes in an 'init' based vserver, once the vserver's init died unexpectedly 1114971876 M * eXplasm hmm, whats the vattribute parameters to enable icmp on the vservers? 1114971892 M * Bertl Doener: but somehow I got the feeling this was solved by accident 1114971924 M * Bertl --ccap icmp IIRC 1114971961 M * Bertl but let's check the flower page ... 1114972003 M * Bertl yup, -ccap icmp it is ... 1114972008 M * Bertl *-- 1114972221 M * eXplasm can you give me the link to the "flower page" please? :) 1114972234 M * Doener http://linux-vserver.org/ 1114972240 M * Doener http://www.nongnu.org/util-vserver/doc/conf/configuration.html 1114972632 M * eXplasm oehm, which build methods can i use with debian sarge?! i cant find any man pages for it 1114972707 M * Doener AFAIK only debootstrap and skeleton... apt-rpm and yum are not available on debian by default... 1114972722 M * Doener if you manage to get those installed, the other build methods should work as well 1114972758 M * Bertl eXplasm: IIRC debootstrap and skeleton (by default) 1114972785 M * Bertl hmm, should read first ;) 1114972826 M * Doener heh :) 1114972828 M * eXplasm E: No such script: /usr/lib/debootstrap/scripts/deb3.1 << which packet is required?? debootstrap is installed 1114972861 M * Doener you should tell it, which debian version you want to use... 1114972870 M * Doener i.e. -- -d {woody,sarge,sid} 1114973154 M * eXplasm ah, okay 1114974025 M * Bertl Doener: http://vserver.13thfloor.at/Experimental/FOR-2.0/delta-nfsd-feat01.diff 1114974283 M * eXplasm hmm, the vserver build, doesnt create an directory in the /vservers ... where is the content?! 1114974312 M * Bertl well, on debian the locations are a little weird ... 1114974327 M * Bertl IIRC the vservers are somewhere in /var/lib/... 1114974355 M * eXplasm omfg 1114974377 M * Bertl well, they say that's where they have to go .. ya' know standards and such ;) 1114974404 M * Doener hmm... if he had /usr/local stuff and such, he probably isn't using the debian util-vserver package ;) 1114974443 M * Bertl good point, but if he had used my config line, then they would be in /vservers no? 1114974463 M * Doener true... but maybe he modified it... 1114974475 M * eXplasm hmm, is a difference between the the apt package and the sources from the page? 1114974484 M * Doener let's try: ls -l /etc/vservers/.defaults/vdirbase 1114974499 M * Bertl if it is /etc/vservers ;) 1114974508 M * Bertl I would suggest to use 1114974518 M * Bertl vserver-info - SYSINFO 1114974529 M * Doener eXplasm: the package is debianized... don't ask me what exactly has been modified... 1114974533 M * Bertl but please, don't paste it here ;) 1114974553 M * eXplasm ;) 1114974631 J * prae ~prae@sherpadown.net 1114974655 M * eXplasm hmm, okay, the content is in /etc/vservers/.defaults/vdirbase/vsTEST ... but why ".defaults"? 1114974666 M * Doener Bertl: except for the fact that i don't know anything about the whole nfs stuff, the patch looks good ;) 1114974689 M * Bertl hehe ... yeah, still have to test it somewhat ... 1114974689 M * Doener eXplasm: that's a directory used by the tools to save configuration stuff... 1114974726 M * Doener eXplasm: now do: ls -l /etc/vservers/.defaults/vdirbase 1114974747 M * Doener the vdirbase is a symlink, i.e. ls -l will show you the real location... 1114974751 M * Bertl eXplasm: the vdirbase is a symlink 1114974860 M * eXplasm puh, okay 1114974874 M * eXplasm damn hot in my room :D 1114974879 M * Doener welcome prae! 1114974891 M * Doener yeah, same here... 1114975272 M * eXplasm bah, i dont like this symlink mazes -.- 1114975272 Q * alexx Read error: Connection reset by peer 1114975360 M * eXplasm ah, the magical config gile :D /lib/util-vserver/util-vserver-vars, brb 1114975417 M * Bertl it's not really a config file ... 1114975552 M * eXplasm i want the config in /etc/vservers, and the content in /vservers .... how at the old version .... the new util-vserver makes all complexer... damn 1114975574 M * Bertl did you see the configure line I pasted before? 1114975591 M * Bertl it should do exactly that on your debian system 1114975615 M * eXplasm you mean the ./configure parameters? 1114975622 M * Bertl yup 1114975651 M * eXplasm yes, of course i have used this 1114975668 M * Doener did you remember to uninstall first? 1114975671 M * Bertl did you uninstall the previous tools before? 1114975684 M * Bertl then rebuild and install the newly configured? 1114975727 M * eXplasm hmm ... it should be ... but i try to install it completely new, wait 1114976005 M * eXplasm hmm whats the better choice ... the apt package or source? 1114976026 M * Bertl is the apt package 0.30.207 ? 1114976048 M * eXplasm Version: 0.30.204-5 1114976059 M * Bertl I hope that answers your question ;) 1114976074 M * eXplasm ;) 1114976768 J * shuri sjnesjd@dsl.speedline209.226.electronicbox.net 1114976827 M * shuri hi all 1114976838 M * Doener hey shuri! 1114976842 M * Bertl hey shuri! LTNS! 1114976877 M * shuri Bertl i confirm that latest quota patch work with latest stable 1114976893 M * shuri LTNS? 1114976902 M * Bertl excellent! (Long Time No See) 1114976908 M * shuri hehe 1114976912 M * shuri i was outside 1114976916 M * shuri for work 1114976934 M * shuri 2 days in a Datacenter! 1114977003 M * eXplasm it works :D /etc/vservers/vsBASE and /vservers *g* but i missing the "config" with hostname, ip and this "main" things o_O 1114977052 M * shuri eXplasm time to read the flower page :) 1114977061 M * eXplasm :D ok 1114977124 M * shuri for ip if i remember 1114977132 M * shuri mkdir vserver-name/interfaces/iface 1114977162 M * shuri create files ip , mask , broadcast , dev 1114977179 M * eXplasm for each option, a own file? :D 1114977183 M * shuri yes 1114977192 M * eXplasm thats a little bit stupid o_O 1114977200 M * eXplasm but okay ;> 1114977224 M * shuri anyone can confirm that? 1114977264 M * Bertl eXplasm: best way is to use some arguments at server creation time ... 1114977278 M * Bertl (you can create a 'dummy' skeleton server) 1114977480 M * shuri the vserver build take care of this 1114977594 M * Doener i'm off to bed now... g'night 1114977599 N * Doener Doener_zZz 1114977633 M * shuri gnite Doener_zZz 1114977635 M * Bertl night Doener_zZz! 1114978008 M * Bertl okay, off for dinner now ... back later ... 1114978030 N * Bertl Bertl_oO 1114978051 M * eXplasm hmm, whats is vshelper?! 1114978154 Q * Hollow Remote host closed the connection 1114979460 J * hws123 ~sebastian@p508CE446.dip.t-dialin.net 1114979467 M * hws123 hi 1114979531 M * hws123 I am trying to copy an existing vserver but I can't start it afterwards. How can I debug this? 1114980196 Q * hws123 Quit: Leaving 1114980725 J * anon1 ~xxx@nts-100.7-185-64.nts-online.net 1114980731 M * anon1 hi guys 1114981613 M * eXplasm hi 1114981615 M * eXplasm hmm 1114981616 M * eXplasm vattribute 49159 --ccap icmp 1114981616 M * eXplasm vattribute: execvp("49159"): No such file or directory 1114981618 M * eXplasm o_O 1114983141 M * eXplasm traceroute: raw socket: Operation not permitted 1114983144 M * eXplasm autsch 1114983959 N * Bertl_oO Bertl 1114983978 M * Bertl evening folks! 1114983993 M * Bertl hey anon1! 1114984018 M * Bertl eXplasm: a) you should not use dynamic contexts ... 1114984034 M * Bertl b) which kernel? 1114984041 M * eXplasm 2.6. 1114984042 M * eXplasm why not? 1114984182 M * Bertl hmm, yeah, I was more looking for something like 2.6.x.y-vsA.B.C ;) 1114984237 M * Bertl and you are missing the --xid in your vattribute command 1114984280 M * eXplasm yeah, i have found it ;> but traceroute doesnt work ... 1114984281 M * eXplasm 2.6.11.5-vs1.9.5-ub-vs 1114984476 M * Bertl did you verify that the vserver has the icmp ccap? 1114984493 M * eXplasm no, how can i check it? 1114984602 M * Bertl http://linux-vserver.org/HowTo+Read+ProcFS 1114984613 M * Bertl it's the status information 1114984818 M * eXplasm hmm and where can i find the icmp flag? 1114984836 M * eXplasm thats on: CCaps: 0000000000000101 or? 1114985046 M * Bertl yup 1114985125 J * Hollow ~Hollow@home.xnull.de 1114985153 M * eXplasm and how can i "extract" them from this binary string?! =D 1114985197 M * Bertl it's not binary, it's hexadecimal ;) 1114985218 M * eXplasm ah okay ;) i havent seen the bcap 1114985249 M * Bertl but that doesn't answer your question ;) 1114985262 M * eXplasm ;) 1114985273 A * Bertl has to look into the source too ... 1114985288 M * Bertl #define VXC_RAW_ICMP 0x00000100 1114985299 M * Bertl so yes, it seems to have the required cap ... 1114985326 M * Bertl nevertheless traceroute fails? 1114985355 M * eXplasm ich such dringend son funpic, linux vs windows wo die konfiguration einer netzwerkkarte beschrieben wird, hat zufällig einer grad den link da? :D 1114985364 M * eXplasm grr 1114985367 M * eXplasm sry 1114985371 M * Bertl np ;) 1114985384 M * eXplasm xchat's amsg is network independend :D 1114985499 J * FaUl ~immo@ip88.164.1211G-CUD12K-01.ish.de 1114985513 M * FaUl hi 1114985541 M * FaUl hey cemil :) 1114985757 J * hillct ~hillct@65.76.245.91 1114985757 Q * eyck_ Read error: Connection reset by peer 1114985770 M * hillct GAH!!!! 1114985772 M * hillct so 1114985797 M * hillct now that I fought my way through this, I guess I should ask if someone else has documented it 1114985819 M * hillct I was just configuring bind9 for use on a vserver host 1114985822 M * hillct lots of fun 1114985837 M * hillct not complex, although tricky to locate the documentation 1114985843 M * hillct OR 1114985852 M * hillct I was just looking in the wrong place 1114985940 J * eyck eyck@81.219.64.71 1114985966 M * FaUl hmm, i've got some strange behaviour with my 2.4.30-vs1.2.10 and networking. i've set some inerfaces eth0 with different vlans eth0 eth0.2 eth0.3 etc. each vserver should use _one_ interface and - if needed connect via the gateway to another (this is importent to enforce that packet-filter-policys) 1114985972 M * mugwump I think there's a section for bind9 in Paul Sladen's FAQ 1114985987 M * hillct saw the part about recompiling without caps 1114985999 M * hillct my issue was an rndc problem 1114986013 M * mugwump oh? 1114986019 M * hillct by default FC2 wants RNDC to talk on localhost 1114986028 M * hillct which fucks things up massively 1114986030 M * FaUl but if i connect lets say vserver shell with eth1.30 (ip 192.168.30.2) i can still connect to every interface on this host via the loopback-device 1114986041 M * mugwump sure, you've got localhost = private IP, not 127.1, right? 1114986087 M * hillct mugwump me? or FaUl? 1114986115 M * mugwump I was talking to you, hillct 1114986120 M * hillct no 1114986127 M * hillct I'm using 127 1114986134 M * mugwump FaUl: the routing doesn't quite work like you're expecting, I think 1114986146 M * FaUl mugwump: how does it work? ;) 1114986172 M * mugwump Well, nothing (except iptables) will stop a vserver on one IP in your box talking to another vserver on your box 1114986199 M * mugwump hillct: remove all references to 127.0.0.1 from the vserver's /etc/hosts, set localhost to its private IP 1114986216 M * FaUl wah, this means I have to compile a kernel with iptables-suport? i'll try this 1114986243 M * FaUl mugwump: is this a bug (as in "will be fixed sometime?') 1114986269 M * FaUl na s/?')/")? 1114986310 M * FaUl arg,h ignore parsing this regex :) 1114986319 M * mugwump FaUl: it's not really a bug, more of a caveat. No doubt the picture will be slightly different with Next Generation Networking. 1114986353 M * FaUl mugwump: will next generation networking get into stable (2.4)-release? 1114986358 M * mugwump hillct: anything with a hardcoded 127.0.0.1 can be fixed, but things tend to work better if you don't rely on that too much 1114986375 M * hillct mugwump is there a vserver config item for private IP (ie: defining device localhost in /etc/vserver/NameHere/interfaces 1114986379 M * hillct ) 1114986393 M * hillct or just locally to the vserver itself? 1114986400 M * mugwump FaUl: no, but it will get into vserver 2.1, which will also be called stable but it's 2.6 1114986433 M * FaUl hmm, i don't like 2.6 on a server because its even more buggy then 2.4 :) 1114986455 M * mugwump hillct: just the /etc/hosts within the vserver needs to change; so, instead of having two interfaces (lo/ethN) with an IP each, you have just one with one IP. 1114986472 M * hillct ah 1114986478 M * FaUl anyway, ill go to bed right now, school tomorrow... 1114986490 M * hillct 2.6 works OK for me 1114986545 M * FaUl hillct: i'd problems with bonding, vlan-support and ipv6 together on 2.6 so i downgraded to 2.4 again 1114986581 M * hillct ah 1114986583 M * FaUl n8 anyway now ;) 1114986595 M * romke Bertl: 2.0-pre3 on 2.6.11.7, chcontext --ctx ps aux -> shows list of proceses, then stops with state S< -- should it be like that? 1114986688 M * mugwump that just means it's waiting for something other than IO, such as work to do... use strace -p to see what it's doing 1114986702 M * mugwump (the < means high priority) 1114986738 M * Bertl romke: hmm, no ... 1114986813 M * romke Bertl: it has something to do with vserver stop 1114986862 M * Bertl romke: interesting, why would that be called? 1114986975 M * romke Bertl: ok, i can now reproduce it: 1. vserver stop 2. this hangs just beffore "sending all processes TERM signal", 3. on other console do: chcontext --ctx ps aux, it'll hang too 1114987017 M * Bertl hmm .. very interesting ... 1114987030 M * Bertl which tool version do you use? 1114987069 M * romke 205 1114987086 M * Bertl hmm, maybe you should update to 0.30.207 ... 1114987098 M * romke propably again my fault, will try :> 1114987102 M * Bertl not sure this is related though ... 1114987255 M * romke Bertl: does 207 need util-vserver-0.30-vpsfix01.diff ? 1114987337 M * Bertl nope, should all be inclusive ... 1114987379 Q * prae Quit: Pwet 1114987389 M * romke ok 1114987757 M * romke Bertl: 207 does not have fix with spinlock.h in dlimit.h -> http://www.rafb.net/paste/results/5KmRaj97.html 1114987780 M * Bertl hmm, yeah, right ... but you know how to fix that, no? 1114987862 M * romke Bertl: moving that include after #ifdef __KERNEL__ was only way to skip error? 1114987884 M * romke Bertl: (see that paste) 1114987904 M * Bertl yup ... 1114987924 M * romke aah, ok... 1114987934 M * Bertl once enrico picks up the recent kernel includes it will be gone 1114987969 M * Bertl unfortunately the 207 release crossed with this fix 1114987993 M * romke heh, I'll manage to live with that :P 1114988650 M * romke Bertl: same thing with 207 1114988699 M * Bertl okay, please do 1114988727 M * Bertl chcontext --xid strace -fF ps aux 1114988745 M * Bertl and upload the output (you can save it with -o ) 1114988920 M * romke Bertl: http://procyon.romke.net/~romke/dev/psaux.strace 1114988930 M * Bertl tx 1114988976 M * Bertl hmm, looks to me like it exited? 1114989060 M * romke Bertl: yes, ps exited, but then i don't get shell prompt back, on third console I see: /sbin/vshelper shutdown 103 *twice* 1114989097 M * Bertl hmm, well, maybe it simply logs you out? 1114989198 M * romke Bertl: http://procyon.romke.net/~romke/dev/chctx.strace 1114989240 M * Bertl yup, you cab bit strace into a context switch 1114989250 M * Bertl s/cab bit/can not/ 1114989261 M * romke Bertl: then why vshelper shutdown doesn't exits? 1114989262 M * Bertl no idea how I actually hit those keys ;) 1114989270 M * romke hehe 1114989286 M * Bertl romke: is your vserver init based? 1114989298 M * romke Bertl: yes 1114989311 M * Bertl then it's probably a bug in the vshelper ... 1114989351 Q * ndim Ping timeout: 480 seconds 1114989358 M * romke Bertl: it looks like after stopping all proceses in selected ctx vshelper hungs 1114989412 M * Bertl well, the vshelper has to wait for the context to exit 1114989426 M * Bertl and the context is supposed to exit after all processes were killed 1114989427 M * romke Bertl: I must use another console to kill it, and all is ok then (ps in that ctx shows only ps, start again work, ...) 1114989451 M * Bertl let's take a look at the vshelper and what it does ... sec 1114989894 M * Bertl hmm .. vshelper shutdown is not ahndled at all .. as far as I can tell 1114989942 M * romke Bertl: it looks like vshelper from "vserver ... stop" blocking others from exiting this context, and that's why other processes entering this context cannot exit after executing command (such as ps) 1114989982 M * Bertl well, let's --debug and later strace the vserver stop then ... 1114990014 M * romke Bertl: --debug where? 1114990020 M * Bertl vserver --debug ... 1114990037 M * Bertl I'll try to recreate it here ... somehow ... 1114990048 M * romke Bertl: in start of context ? 1114990063 M * Bertl vserver --debug hansi stop 1114990087 M * Bertl you said the simple stop hangs, no? 1114990096 M * romke aah, ok 1114990103 M * romke wait a sec 1114990290 J * ndim hun@helena.bawue.de 1114990396 P * hillct 1114990423 M * romke Bertl: http://procyon.romke.net/~romke/dev/debug <- till hang - what to do now? kill vshelper? 1114990440 M * Bertl no let it hang ... 1114990504 M * romke Bertl: it's hanging right now :| 1114990515 M * Bertl good ... 1114990540 M * Bertl you rc script does something .. which makes it hang ... 1114990550 M * Bertl (inside the vserver) 1114990566 M * Bertl the vserver stop has no timeout yet 1114990575 M * Bertl so it keeps waiting for that too ... 1114990589 M * Bertl now if you go to a different console/logon on the host 1114990590 M * romke Bertl: I can show you what It does after showing "On the next boot fsck will be skipped" 1114990607 M * Bertl and do vps auxwww 1114990622 M * Bertl we should see what is still active in this context 1114990646 M * Bertl also a vkill --xid should make it proceed 1114990689 M * romke Bertl: nothing is active in that context 1114990706 M * romke (as vps auxwww say) 1114990731 M * Bertl hmm, unusual ... 1114990744 M * Bertl what does /proc/virtual//status say? 1114990784 M * romke UseCnt: 2 1114990784 M * romke Tasks: 0 1114990784 M * romke Flags: 0000000202000210 1114990784 M * romke BCaps: ffffffffd44c04ff 1114990784 M * romke CCaps: 0000000000000101 1114990786 M * romke Ticks: 0 1114990828 M * Bertl okay, let's look at /proc/virtual//limit then ... 1114990870 M * romke Bertl: http://www.rafb.net/paste/results/we3iaV66.html 1114990911 M * Bertl hmm, funny seems really to be the shutdown helper which is sitting around 1114990958 M * Bertl what guest distro is this? 1114990982 M * Bertl (how large is the vserver guest?) 1114991011 M * romke Bertl: both host and guest PLD Linux 1114991047 M * romke Bertl: guest is 87MB 1114991058 M * romke Bertl: wanna tar.bz2 of it? 1114991084 M * Bertl yup, please (or if you use ext2 a dump? 1114991096 M * romke Bertl: xfs 1114991122 M * Bertl okay, then tar.bz2 please ;) 1114991160 M * romke preparing right now