1161562495 J * meandtheshell ~markus@85-124-233-119.work.xdsl-line.inode.at 1161562514 M * Bertl wb meandtheshell! 1161562544 M * meandtheshell Bertl: thank you :) 1161566139 M * Aiken Bertl compiling a 2.6.18.1 now, anything you would like tested or are you just after how it behaves in general for me on the alpha? 1161566386 M * Bertl everything is interesting, we are immediately before a release :) 1161566427 M * Bertl I know that there will be a warning regarding duplicate inclusion (which cannot be worked around easily and is harmless) 1161566458 M * Bertl something like this: include/linux/vs_base.h:129:2: warning: #warning duplicate inclusion (for 2.1.1-rc42) 1161566496 M * Bertl I also found a warning in mm/filemap.c which I have a patch for here (i.e. can be considered fixed) 1161566608 M * Aiken restarted the build and logging it this time 1161566615 M * Bertl any additional info, may it be from compiling or testing that would be interesting 1161567770 M * Aiken include/linux/vs_base.h:129:2: warning: #warning duplicate inclusion 1161567770 M * Aiken mm/filemap.c:2370: warning: initialization discards qualifiers from pointer target type 1161567800 M * Bertl yep, those are the two warnings I was talking about 1161567816 M * Bertl the second one is already fixed, the first one is problematic 1161567835 M * Bertl i.e. it is easy to fix for a specific arch, but not for all archs 1161567874 M * Bertl some archs include it twice, other archs only once, the duplicate inclusion goes via sched.h, which makes it even more complicated 1161567967 M * Aiken going to take awhile, the stripped vmlinux just produced is > 4 meg which brings me back to that old problem with my alpha :( 1161568450 M * Bertl hmm, I thought that was already fixed/worked around 1161568474 M * Bertl but I guess we are going in circles when I suggest to use compressed kernels :) 1161568520 M * Aiken the only milos that I can use won't work with a compressed image 1161568527 M * Aiken I tried 1161568540 M * Aiken can only get it to work with an uncompressed vmlinux 1161568560 M * Bertl thought so, maybe 'lab' would be an option? 1161568566 M * Aiken lab? 1161568574 M * Bertl linux as bootloader 1161568599 M * Bertl basically kexec with a tuned kernel 1161568620 M * Aiken I have a bad feeling if I go upstairs and turn the monitor on I am going to see an OOPS :( 1161568670 M * Aiken it lasted long enough for me to ssh in get the resuls from uname -a 1161568697 M * Bertl hmm, with 2.1.1-rc42? 1161568794 M * Aiken yes 1161568801 M * Bertl that's bad ... 1161568821 M * Aiken but also think about how many times a new kernel release has been broken on my alpha 1161568833 M * Aiken it could be the kernel or it could be rc42 1161568840 M * Bertl btw, regarding compressed images for milo, did you ever 1161568863 M * Bertl try to use gzip on the normal kernel image and leave the .gz extension as is? 1161568885 M * Bertl s/extension/suffix/ 1161568889 M * Aiken good question 1161568895 M * Aiken I know I tried gzip 1161568900 M * Bertl just stumbled over this: 1161568903 M * Aiken don't remember if I tried the .gz 1161568910 M * Bertl Gzip'd files are supported and these are recommended, particularly if you are loading from floppy which is rather slow. MILO recognises these by their .gz suffix. 1161569050 M * Aiken I have the output from netconsole which is something 1161569071 M * Bertl ah, great! please upload if possible ... 1161569240 M * Aiken http://paste.linux-vserver.org/537 1161569254 M * Aiken BUG: warning at kernel/irq/handle.c:177/__do_IRQ() 1161569265 M * Aiken my guess is mainline 1161569321 J * FaUl_ immo@shell.chaostreff-dortmund.de 1161569345 M * Bertl Aiken: nope, that looks like vserver 1161569418 M * Aiken any of the address you want hit with addr2line? 1161569431 M * Bertl nope, that's pretty clear, but! it's only a warning 1161569441 Q * FaUl Ping timeout: 480 seconds 1161569449 M * Bertl so it is not supposed to oops the machine, but I guess you got flooeded by such warnings 1161569484 M * Bertl let me check what actually causes the misbehaviour in line 177 1161569617 M * Aiken it is days like this where I think it would be nice if it was not at the other end of the house and at the top of a spiral staircase 1161569632 M * Aiken the error messages are still scrolling on the screen 1161569637 M * Aiken too fast to read 1161569695 M * Bertl yeah, seems we missed a change for alpha, will have a patch in a few minutes 1161569740 M * Aiken that will give me time to start lunch and to recover the alpha 1161569788 M * Bertl I think we missed v850 too, not that it would matter :) 1161570690 M * Aiken vmlinux.gz seems to work fine 1161570710 M * Bertl ah, great, so finally a solution for the 4M issue :) 1161570714 M * Aiken yes 1161570732 M * Aiken I am positive I have tried using gzip previously 1161570794 M * Aiken when you have a fix I am ready to try it 1161570813 M * Bertl should be ready in a minute 1161570932 M * Bertl okay, here they are: first one fixes the warning, second one the WARN_ON() (at least I hope so :) 1161570939 M * Bertl http://vserver.13thfloor.at/Experimental/delta-vs211-clean01.diff 1161570958 M * Bertl http://vserver.13thfloor.at/Experimental/delta-doirq-fix02.diff 1161571047 M * Aiken compiling 1161571099 M * Bertl great! tx! 1161571392 M * Bertl you have at least a guest on that alpha which is autostarted, right? 1161571399 M * Bertl s/a/one/ 1161571399 Q * Johnnie Read error: Connection reset by peer 1161571414 M * Aiken yes 1161571436 M * Bertl k, explains the 'scrolling warnings ...' 1161571540 J * Johnnie ~jdlewis@jdlewis.org 1161571711 M * Aiken vmlinux.gz failed 1161571728 M * Aiken it only seems to work if the vmlinux is < 4 meg 1161571736 M * Aiken which leaves me at my original problem 1161571747 M * Bertl indeed, how much memory do you have in that box? 1161571765 M * Aiken 192 meg 1161571791 M * Bertl hmm, and milo knows that? 1161571845 M * Aiken I'll have a look when I have a smaller kernel ready 1161571853 M * Aiken the latest milo I can use is 2.0.35 1161571863 M * Aiken any later versions of milo refuse to run 1161571870 M * Aiken ie 2.2.X milo 1161571870 M * Bertl okay, another thing which came to my mind is the 'load address' for the kernel 1161571894 M * Aiken milo complains about invalid format to vmlinux.gz 1161571932 M * Aiken but was working fine when I checked a bit earlier with a vmlinux that was < 4 meg 1161572097 M * Bertl does the message look like this?: 1161572100 M * Bertl #UNZIP invalid compressed format 1161572118 Q * ensc Killed (NickServ (GHOST command used by ensc_)) 1161572128 J * ensc ~irc-ensc@p54B4FA42.dip.t-dialin.net 1161572206 M * Aiken invalid compressed format 1161572895 Q * mire Remote host closed the connection 1161573177 M * Aiken Linux pebbles.bedrock 2.6.18.1-vs2.1.1-rc42 #6 Mon Oct 23 12:51:13 EST 2006 alpha Unknown Alcor GNU/Linux 1161573217 M * Aiken so far it is staying up, a guest is working - have ssh into both the host and a guest 1161573258 M * Aiken the guest is a still a COW 1161573337 M * Bertl okay, be aware that the CoW link breaking was improved 1161573346 M * Bertl i.e. chown/chmod will break links now too 1161573377 M * Bertl (that's why the 'not-yet-updated' test #116 of testfs.sh fails :) 1161573474 M * Aiken not sure if it was the older code or from the many crashes but needed a fsck.ext3 on the partition where my guests reside to recover abt 100 meg ealier today 1161573510 M * Aiken testme passes http://paste.linux-vserver.org/539 1161573558 M * Bertl excellent! 1161573702 M * Aiken this how you expect testfs to fail? http://paste.linux-vserver.org/540 1161573784 M * Bertl yep, precisely 1161573812 M * Bertl happens for all filesystems which are supported 1161573845 M * Aiken vxW: signal xid mismatch fffffc00002d75a0[#0,2] xid=#49152 1161573851 M * Aiken what about lots of those messages? 1161573881 M * Aiken 155 of them 1161573912 M * Bertl that's unusual 1161573930 M * Bertl let me check what it actually means 1161574062 M * Aiken those messages happen when I stop the guest 1161574073 M * Bertl that's funny, it basically means that 1161574100 M * Bertl context 49152 (dynamic context) is trying to send signals to a task on the host (pid=2) 1161574133 M * Aiken PID TTY STAT TIME COMMAND 1161574134 M * Aiken 1 ? S 0:00 init [5] 1161574134 M * Aiken 2 ? SWN 0:00 [ksoftirqd/0] 1161574134 M * Aiken 3 ? SW 0:00 [watchdog/0] 1161574183 M * Aiken pid 2 on the host? [ksoftirqd/0] 1161574282 M * Bertl so why would your guest send signals to ksoftirqd? 1161574325 M * Aiken a slightly bigger sample http://paste.linux-vserver.org/541 1161574339 M * Aiken I am wondering if it is working though all pids on the host 1161574377 M * Aiken current pid list http://paste.linux-vserver.org/542 1161574378 M * Bertl hmm, looks like 1161574401 M * Bertl could it be that there is something wrong with the pid isolation? 1161574428 M * Aiken the pid list from ps ax and the pid list from those messages correspond 1161574429 M * Bertl could you check what 'ls /proc' gives (no -l or other options, just the ls) 1161574435 M * Aiken host? 1161574438 M * Bertl nah, guest 1161574454 M * Bertl I'd suspect that there are pids listed which do not belong there 1161574485 M * Bertl or there is a loop on guest shutdown going through all possible pids, and sending a kill signal there 1161574487 M * Aiken http://paste.linux-vserver.org/543 1161574506 M * Bertl that looks quite fine 1161574514 M * Aiken is that warning a new warning or has it been there for quite awhile? 1161574526 M * Bertl brand new for your standards 1161574534 M * Aiken 0.30.208 1161574535 M * Bertl i.e. we added it around rc40 IIRC 1161574552 M * Aiken latest tools still .211? 1161574560 M * Bertl yep, I think so 1161574628 M * Aiken building .211 to see if that makes any difference 1161575188 M * Bertl hmm, do you use reboot_kill? 1161575270 M * Aiken I used vserver hoppy stop 1161575316 M * Bertl yeah, that's a flag, I wonder if it is/was enabled 1161575477 M * Aiken where would I check that? 1161575484 M * Aiken .211 tools do the same thing 1161575511 M * Bertl okay, let's start the guest and check /proc/virtual//status 1161575543 M * Aiken UseCnt: 7 1161575543 M * Aiken Tasks: 2 1161575543 M * Aiken Flags: 0000000602020010 1161575543 M * Aiken BCaps: 00000000344c04ff 1161575543 M * Aiken CCaps: 0000000000100101 1161575548 M * Bertl http://linux-vserver.org/Capabilities_and_Flags 1161575560 M * Bertl the flag we are looking for is 1<<37 1161575594 M * Bertl I see 33 and 34 but not 37 1161575632 M * Bertl btw, better switch to static contexts in the future, dynamic xids will go away 1161575663 M * Bertl (you can do that by simply echo-ing a value between 2-49151 into /etc/vservers//context ) 1161575721 M * Aiken dodne 1161575723 M * Aiken done 1161575754 M * Bertl now hoppi was it? should have a fixed xid :) 1161575777 M * Aiken hoppy has a fixed xid of 10 1161575821 M * Bertl I guess I know what caused those warnings, will try to recreate that here shortly 1161576623 M * Bertl Aiken: could it be that your vshelper is modified? 1161576761 M * Aiken not modified 1161576915 M * Aiken but it looks like rc 6 is being run 1161576923 M * Aiken which is strange for a stop as 6 is reboot 1161576960 M * Bertl hmm, does your guest contain any 'killing' there? 1161576981 M * Aiken Sending all processes the TERM signal... [ OK ] 1161576981 M * Aiken Sending all processes the KILL signal... [ OK ] 1161576997 M * Bertl okay, seems it does, which is usually fine 1161577015 M * Bertl but let's check how it does, and maybe what happens if you call that within a context 1161577018 M * Bertl like this: 1161577037 M * Bertl chcontext --xid 666 --secure -- 1161577061 M * Bertl by default, I'd assume it will run killall5 1161577080 M * Bertl which IMHO should check procfs for details, but maybe it simply doesn't 1161577109 M * Bertl nevertheless, that should not lead to the warnings you observed 1161577161 M * Aiken it runs killall5 1161577177 M * Aiken which leads to many xid mismatch for xid=#666 messages 1161577207 M * Bertl hmm, strange ... do you have time for a test kernel (trivial change)? 1161577218 M * Aiken yes 1161577270 M * Bertl okay, the warning is issued in kernel/signal.c ~ 594 1161577284 M * Bertl you should see a vxwprintk() there 1161577302 M * Bertl right below that (before the return) add 1161577325 M * Bertl WARN_ON((current->xid); 1161577345 M * Bertl that will do a stack trace in addition to the warning 1161577365 M * Bertl (so probably a stack trace for each pid you have, but I guess the machine should handle that) 1161577378 M * Aiken that it? 1161577386 M * Bertl if you are worried about the log amount, you can make that: 1161577409 M * Bertl WARN_ON((current->xid && (t->pid == 2)) 1161577414 M * Bertl +; 1161577585 M * Bertl but that's it, I'm interested in the stack trace, i.e. what functions show up there 1161577634 M * Aiken next silly question is which include is needed to define WARN_ON 1161577654 M * Bertl oops, sec 1161577680 M * Bertl asm/bug.h 1161577754 M * Aiken I had tried asm-generic/bug.h 1161577790 M * Aiken http://paste.linux-vserver.org/544 1161577875 M * Aiken won't compile with either WARN_ON 1161577878 M * Bertl *sorry* 1161577897 M * Bertl there is a duplicate bracket on the left 1161577910 M * Bertl WARN_ON(current->xid); 1161577912 M * Bertl or 1161577921 M * Bertl WARN_ON(current->xid && (t->pid == 2)); 1161577925 M * Bertl should work 1161577937 M * Aiken whoops, had not picked that 1161577979 M * Bertl typical copy paste error 1161578082 J * FireEgl FireEgl@Sebastian.Atlantica.US 1161578125 M * Bertl wb FireEgl! 1161578188 M * FireEgl =) 1161578275 M * Aiken Bertl http://paste.linux-vserver.org/546 1161578314 M * Bertl hum, xpid .. interesting 1161578448 M * Bertl nah, that looks more like a frame unrolling issue 1161578456 M * Aiken want anything hit with addr2line? 1161578480 M * Bertl no, I think I know how it happens (and probably where) but I don't know why yet :) 1161578678 M * Bertl ah, now I know why too :) 1161578717 M * Bertl expect a patch shortly 1161579600 M * Bertl Aiken: http://vserver.13thfloor.at/Experimental/delta-kill-fix01.diff 1161579614 M * Bertl this should fix the warnings 1161579864 M * Aiken rebooting 1161579995 M * Aiken seems to work 1161580012 M * Bertl excellent! 1161580041 M * Aiken vxW: signal xid mismatch fffffc00002d79e0[#0,1] xid=#10 1161580046 M * Aiken vxW: signal xid mismatch fffffc00002d79e0[#0,1] xid=#10 1161580046 M * Aiken vxW: signal xid mismatch fffffc00002d79e0[#0,1] xid=#10 1161580055 M * Aiken the one line test worked 1161580067 M * Aiken I still get those 3 lines when stopping a guest 1161580082 M * Bertl okay, that is where it tries to signal xid=1 1161580114 M * Bertl question is, does the guest have an init? 1161580128 M * Bertl or is it a fake init (i.e. blend through host init)? 1161580213 M * Aiken fake init 1161580225 M * Aiken so it could be trying to kill of the host init?? 1161580242 M * Bertl well, no, the warning shows that the kill is blocked 1161580251 M * Bertl but it tries to send a signal there, yes 1161580286 M * Bertl do we have a stack trace for them too, or did you remove the WARN_ON() already? 1161580303 M * Aiken I had removed it 1161580309 M * Aiken found another problem 1161580315 M * Bertl let's hear 1161580338 M * Aiken if I enter the guest the keyboard stopps responding properly in the terminal session 1161580366 M * Bertl ah, yes, we already observed that on ppc 1161580373 M * Bertl try the following: 1161580396 M * Bertl type something blind (including the enter) then press CTRL-D 1161580399 M * Aiken that is only recent bad behaviour 1161580406 M * Aiken that works 1161580423 M * Aiken ie ls ctrl-D giving me the output of ls 1161580432 M * Bertl yeah, disabling the 'new' (0.30.211) vlogin option is a final solution for now 1161580456 M * Bertl IIRC, you have to remove it from the vserver script itselfs 1161580458 M * Bertl *itself 1161580482 M * Bertl seems like some of the terminal tricks played there stop working on non x86 archs 1161580513 M * Aiken TERM=rxvt 1161580547 M * Bertl IMHO it's a mix of dietlibc breakage and 'unfortunate' userspace flags 1161580559 M * Bertl grep for the vlogin and just remove it for now 1161580561 M * FaUl_ Bertl: if everything works fine I finally found a shuttle for that sgi and sun to munich - it should be much easyer to get them from munich to you ;-) 1161580578 M * Bertl FaUl_: yep, definitely! 1161580593 M * Bertl i.e. I can probably pick them up there easily 1161580629 M * FaUl_ excellent 1161580636 N * FaUl_ FaUl 1161580679 M * FaUl so I hope a friend of mine takes it to darmstadt then and there it will be picked up by ray6 - it's not 100% now but it doesn't look to bad 1161580694 M * Bertl ah, cool! 1161580810 M * Bertl okay, as usual, thanks a lot to everybody for helping/testing/whatever! 1161580841 M * Bertl Aiken: was great, we fixed a lot of issues so far, will think about removing those remaining pid=1 warnings 1161580859 M * Bertl but now, I'm off to bed to catch some sleep :) 1161580862 M * FaUl anyway, i've to finallize my boot-process (dress up at all) so i can move myself to the uni 1161580876 M * Bertl FaUl: make that, and have fun! 1161580885 M * Aiken Bertl http://paste.linux-vserver.org/548 1161580901 M * Bertl ah, the trace, tx a lot! 1161580922 M * Bertl addr2line of the following would be nice;: 1161580925 M * Bertl fffffc0000339554 1161580934 M * Bertl fffffc0000339530 1161580945 M * FaUl Bertl: I'll try. is my first basic of electric-Lecture now, and my first basic of informatic afterwards ;-) 1161580969 M * Aiken fffffc0000339554 kernel/signal.c:1125 1161581000 M * Bertl FaUl: electrical engineering and computer sciences, not bad for a start :) 1161581001 M * Aiken fffffc0000339530 kernel/signal.c:1122 1161581126 M * Aiken 1125 error = group_send_sig_info(sig, info, p); 1161581138 M * FaUl Bertl: i'll see. that intridution into computer-technology is very promising ;-) 1161581164 M * FaUl as "cryptography and data-security" is 1161581224 M * FaUl anyway, its time for me now if i don't want to miss my train ;-) 1161581233 M * Bertl k, cya! 1161581318 M * Bertl Aiken: hmm, yeah, that probably needs a similar fix 1161581335 M * Bertl thanks again .. I'm off now to bed 1161581352 N * Bertl Bertl_zZ 1161581352 M * Aiken gn 1161582035 Q * _node Ping timeout: 480 seconds 1161582977 J * debugger Rui@217.129.151.190 1161582983 N * debugger rgl 1161585191 Q * Skram Ping timeout: 480 seconds 1161585403 J * dna_ ~naucki@4-226-dsl.kielnet.net 1161585409 J * Skram ~Mark@hermes.sentiensystems.com 1161588523 Q * dna_ Quit: Verlassend 1161589456 Q * _are_ Ping timeout: 480 seconds 1161590177 J * debugger Rui@217.129.151.190 1161590626 Q * rgl Ping timeout: 480 seconds 1161590863 J * prae ~Benjamin@host.187.57.23.62.rev.coltfrance.com 1161590887 Q * Zaki Remote host closed the connection 1161591191 Q * Aiken Ping timeout: 480 seconds 1161592117 J * renihs ~penguin@83-65-34-34.arsenal.xdsl-line.inode.at 1161592972 J * _are_ ~are@62.112.159.81 1161593082 Q * shedi Quit: Leaving 1161593852 Q * mcp Ping timeout: 480 seconds 1161594938 J * _cob ~cob@pc-csa01.science.uva.nl 1161595097 Q * ||Cobra|| Ping timeout: 480 seconds 1161595488 J * lilalinux ~plasma@dslb-084-058-192-088.pools.arcor-ip.net 1161595723 J * ms_ ~ms@arkansas.doc.ic.ac.uk 1161595785 M * ms_ morning. Just upgrade my server (debian), 2.6.18.1 kernel, and after all the vservers have started up, got a kernel warning: 1161595789 M * ms_ debugger: warning at kernel/pid.c:270/pid_task() 1161595811 M * ms_ err, that should have been "BUG:" (silly irc client) 1161595822 M * ms_ this corresponds to http://paste.linux-vserver.org/530 1161595832 M * ms_ but everything /seems/ to be fine 1161595845 M * ms_ just wondering if there's any further info about this issue? 1161596025 J * Zaki ~Zaki@212.118.110.193 1161597378 J * debugger_ Rui@217.129.151.190 1161597741 M * ms_ anyone awake yet? 1161597746 J * mire ~mire@65-167-222-85.adsl.verat.net 1161597836 Q * debugger Ping timeout: 480 seconds 1161598134 J * ||Cobra|| ~cob@pc-csa01.science.uva.nl 1161598350 Q * _cob Ping timeout: 480 seconds 1161598855 M * renihs me is awake but also no help :) 1161599780 Q * ||Cobra|| Read error: Connection reset by peer 1161599910 J * dna_ ~naucki@109-221-dsl.kielnet.net 1161599967 J * ||Cobra|| ~cob@pc-csa01.science.uva.nl 1161601381 M * _are_ I am awake, if that helps depends on the actual question. 1161601591 J * coocoon ~coocoon@dslb-084-056-162-181.pools.arcor-ip.net 1161601648 M * coocoon morning 1161601774 J * martink ~martink@89.248.242.50 1161602040 M * ms_ um, anyone know anything about http://paste.linux-vserver.org/530 ? 1161602078 M * martink Hi, is there any way how to completely disable network isolation for selected guests and thus have complete access to host's network? I would like to use quagga inside vserver guest and use vserver only as some kind of "better chroot". 1161602106 Q * DreamerC Remote host closed the connection 1161602134 M * daniel_hozac ms_: that should be fixed in recent patches. 1161602176 M * daniel_hozac martink: if the first IP address assigned to the guest is 0.0.0.0, it will have access to all IPv4 interfaces. 1161602240 J * DreamerC ~dreamerc@125-232-163-52.dynamic.hinet.net 1161602387 M * martink daniel_hozac: thank you! I hoped that it would be possible to use this way some ipv6 software too, as ipv6 is supported only in host. so this works for ipv4 only? or is there any workarount for ipv6 software inside quest without using those experimental ipv6 patches? 1161602434 M * daniel_hozac no, you'll need the IPv6 patches for IPv6. 1161602498 M * ms_ daniel_hozac: ahh, so the next debian kernel package out will fix it then? In the mean time, is there any danger to ignoring that warning? 1161602538 M * martink daniel_hozac: thank you 1161602721 Q * tso Ping timeout: 480 seconds 1161602731 M * phedny wow, company i work for is very satisfied with linux-vserver based system i installed 1161602754 M * phedny they now want another server to be converted to linux-vserver based system :) 1161602828 M * daniel_hozac ms_: i'd assume so, but you'd have to talk to waldi. 1161602851 M * ms_ ok, thanks 1161602873 M * daniel_hozac phedny: sounds great! 1161602902 M * phedny uh-hu, I feel excited for this new project :) 1161602933 M * Hollow haskell sucks 1161602935 M * Hollow ;) 1161602962 M * ms_ is it a standard problem that if you use vhashify and then apt-get/dpkg comes along and tries to upgrade stuff that you can often suffer from "permission denied" issues in the guests? 1161602971 Q * eyck Ping timeout: 480 seconds 1161603055 M * daniel_hozac yes. 1161603088 M * ms_ so presumably the use of vhashify is somewhat limited? 1161603145 M * cehteh mhm .. i havent had that yet 1161603227 M * cehteh well ... you could/should configure the hashify then to exclude the files in question .. and when you already run it trouble, break the link and fix permissions 1161603241 M * daniel_hozac dpkg tries to chmod the file before removing it, which just doesn't work. 1161603247 M * daniel_hozac (if it's setuid) 1161603820 M * waldi daniel_hozac: which patch version? 1161603838 M * daniel_hozac 2.0.2.2-rc3 should be fixed, at least. 1161604194 J * mcp ~hightower@wolk-project.de 1161604325 J * Ben81 ~Ben81@tipi0e.lri.fr 1161604638 J * rgl Rui@217.129.151.190 1161604986 Q * debugger_ Ping timeout: 480 seconds 1161605983 Q * coocoon Quit: KVIrc 3.2.0 'Realia' 1161606560 Q * dna_ Ping timeout: 480 seconds 1161607195 J * dna_ ~naucki@109-221-dsl.kielnet.net 1161607278 M * ms_ ?type (let (.:) = (.) . (.) in unzip .: zip) 1161607278 M * ms_ forall a b. [a] -> [b] -> ([a], [b]) 1161607278 M * ms_ wtf :-P 1161607278 M * ms_ (.) . (.) 1161607278 M * ms_ who's the first to make a joke about boob syntax? 1161607287 M * ms_ agh, sorry, ww 1161607519 M * doener Haskell? 1161607537 M * ms_ yeah, sorry about that 1161607571 M * doener slightly above the 4 lines paste limit ;) 1161607639 M * Hollow hehe 1161607645 M * Hollow another who has to learn haskell? ;) 1161607663 M * ms_ no, I teach it! ;) 1161607666 M * Hollow today my lecture spent two ours to define the float datatype and howto define function, wtf.. 1161607668 M * ms_ haskell is cool 1161607680 A * Hollow yawned way too much 1161607767 M * doener Hollow: let me know when you start to play music with Haskell ;) 1161607794 M * Hollow hehe 1161607808 M * doener basically our lectures did some DNA stuff, then music and finally sorting with Haskell... 1161607829 M * doener oh! and around christmas we had to print an ascii christmas tree with it... 1161607838 M * Hollow wow! 1161607839 M * Hollow *g* 1161607853 M * Hollow beschäftigungstherapie 1161607890 M * doener I believe that functional programming languages can be cool/useful, but the stuff we did was... boring... 1161607912 M * doener Hollow: nah, rather ABM ;) 1161607944 M * Hollow yeah, sth like that ;) 1161607970 Q * mcp Read error: Connection reset by peer 1161607970 J * _mcp ~hightower@wolk-project.de 1161609092 J * Piet hiddenserv@tor.noreply.org 1161609442 J * tso ~tso@244-005.dyn-fa.pool.ew.hu 1161611026 Q * ms_ Ping timeout: 480 seconds 1161611626 Q * rgl Read error: Connection reset by peer 1161613791 Q * sladen Ping timeout: 480 seconds 1161613833 J * sladen paul@starsky.19inch.net 1161614502 M * kevinp Is there a way to look at from the host what process is listening on port 80 in each vserver guest? 1161614522 M * kevinp and what IP address it is bound to? 1161614557 M * doener chcontext --xid 1 netstat -tnlp | grep :80 1161615274 M * kevinp very good, any way to tie in the owner/xid to each one as well? 1161615316 M * kevinp for example, I see some that are 0.0.0.0:80, I want to see which of the guests are doing that 1161615338 M * kevinp (I think somehow one guest is blocking another guests port 80) 1161615370 M * kevinp Like one IP address was once assigned in a different guest or something 1161615399 M * _are_ well, escept with iniot every pid should be unique -> grep vps ax for that pid 1161615447 M * _are_ wow, should start waking up, sorry for that many typos. I fixed 3 already before sending :-/ 1161615473 M * kevinp all show tcp 0 0" 1161615481 M * kevinp all show "tcp 0 0" 1161615503 M * kevinp maybe with lsof? 1161615516 Q * prae Ping timeout: 480 seconds 1161615534 Q * Rich_Estill_afk Quit: Leaving 1161615546 J * Rich_Estill_afk ~restill@c-24-11-195-139.hsd1.mi.comcast.net 1161615643 M * kevinp is xid 1 special to show all xids or something? 1161615649 M * doener yep 1161615660 M * _are_ context 1 is a read-onlys context able to see all things 1161615665 M * doener it can see all processes/network connections/... 1161615670 M * kevinp good to know 1161615701 M * kevinp when I try lsof -i instead of netstat I only get what looks like the host's processes 1161615831 J * eyck eyck@ghost.anime.pl 1161615858 M * doener kevinp: ls -d /proc/virtual/[0-9]* | xargs -n1 basename | xargs -i -n1 bash -c "echo XID {}; chcontext --xid {} netstat -tlnp 2>&1 | grep :80" 1161615872 M * doener that cannot show which process is using that socket though 1161615916 M * kevinp that should give me the info I need, awesome, thanks! 1161616314 N * Bertl_zZ Bertl 1161616318 M * Bertl morning folks! 1161616322 M * doener morning Bertl 1161616421 M * Bertl any news? 1161616663 J * prae ~Benjamin@host.187.57.23.62.rev.coltfrance.com 1161616697 M * Bertl welcome prae! 1161617161 J * shedi ~siggi@inferno.lhi.is 1161617749 Q * Ben81 Quit: Leaving 1161617755 J * ms_ ~ms@arkansas.doc.ic.ac.uk 1161617804 M * Bertl wb ms_! 1161618064 M * ms_ mmm. tnx. not sure what happened there 1161618108 M * ms_ if two processes in different guests wish to link to the same shared library, how many times does the library get loaded. 1161618109 M * ms_ ? 1161618155 M * Bertl depends on the disk layout 1161618165 M * Bertl if the guests are unified, it is mapped only once 1161618172 M * ms_ oh right. 1161618172 M * kevinp I currently have 1 ip address assigned to my host and all the guests look like this --> eth0:honeypot 1161618195 M * kevinp How do I set up another IP address on the host? Can I create a eth0:0? 1161618205 M * kevinp without messing things up? 1161618212 M * ms_ Bertl: but processes in different contexts can't actually create shared memory between each other can they? 1161618224 M * Bertl kevinp: sure, note that newer tools do not use the 'older' aliases by default 1161618250 M * Bertl ms_: nope, no writeable shared memory between guests atm 1161618279 M * Bertl kevinp: i.e. the addresses will typically go to eth0 (not eth0:something) 1161618279 M * ms_ atm?? Surely allowing such things would offer bad consequences wrt the separation of processes? 1161618332 M * Bertl ms_: well, we are heading towards an even more modular concept, where you will be able to _share_ certain spaces (in this case memory) between guests 1161618356 M * ms_ why? what problem(s) does this solve? 1161618371 M * Bertl none, it just gives more flexibility if you need that 1161618411 M * ms_ but by default it'll all be off right? 1161618427 M * Bertl yep, the default guest is well isolated, of course 1161618439 M * ms_ I'm just concerned that bad configuration would now mean that isolation is not as expected... 1161618465 M * Bertl nah, be asured that will not be the case 1161618475 A * ms_ is suitable assured ;-) 1161618483 M * ms_ -e +y ;( 1161618582 M * kevinp thanks Bertl 1161618590 M * Bertl np 1161618685 M * ms_ are you going to sort vhashify so that it doesn't break apt-get/dpkg ? 1161618702 M * ms_ (I realise the breakage may well be dpkg's fault) 1161618723 M * Bertl you probably refer to the chmod before replacing? 1161618739 J * stefani ~stefani@tsipoor.banerian.org 1161618748 M * Bertl if so, that should be already fixed in recent devel versions 1161618753 M * ms_ oh cool 1161618754 M * Bertl hola stefani! 1161618760 M * matti Bertl: :-) 1161618764 M * stefani salut. 1161618768 M * matti Bertl: Hello, how are you? 1161618772 M * ms_ Bertl: and yes, that was what I was refering to. Tnx 1161618783 M * Bertl matti: fine, tx, and you? 1161618794 M * matti Bertl: Fine, thanks :) 1161618805 M * matti :) 1161619239 J * ekc ~ekc@netblock-66-245-252-180.dslextreme.com 1161619257 M * Bertl welcome ekc! 1161619277 M * ekc hello 1161619501 M * ekc Can a malicious vserver poison the ARP caches on a LAN (other hosts,vservers,routers)? Is there a way to prevent that? 1161619525 M * Bertl as arp is not allowed by default, I'd say no 1161619551 M * ekc whats' the capabilites flag for arp? 1161619572 M * Bertl well, at userspace level, you need CAP_NET_RAW for that at least 1161619615 M * ekc ok 1161619863 J * bonbons ~bonbons@83.222.36.111 1161619995 Q * bronson__ Ping timeout: 480 seconds 1161620172 Q * ekc Remote host closed the connection 1161620590 M * Bertl hey bonbons! 1161620749 M * kevinp It looks like I have a vserver that is holding on to port 80, maybe a hidden process? 1161620766 M * hardwire it can only hold on to port 80 for its instance only 1161620785 M * hardwire and shouldn't effect localhost or any other vserver unless they are sharing IP's 1161620785 M * Bertl kevinp: unlikely, check with vps and chcontext --xid 1 lsof ... 1161620786 M * kevinp I can't see it inside the vserver, but I can from outside? 1161620812 M * hardwire the heat is off at work today 1161620816 M * hardwire its 42 degrees in my office 1161620821 M * hardwire typing slower than usual 1161620844 N * _mcp mcp 1161620851 N * mcp _mcp 1161620877 M * kevinp I'll try and get some more info to show you 1161621020 M * Bertl yes, please do so ... 1161621165 J * teukka ~tmatilai@201.254.102.13 1161621187 M * Bertl welcome teukka! 1161621205 M * teukka thanks =) 1161621226 A * hardwire bows 1161621326 M * prae (lag) hi Bertl 1161621433 J * jesse_ ~wenchien@59-105-176-11.adsl.static.seed.net.tw 1161621473 N * jesse_ wenchien 1161622166 Q * _are_ Ping timeout: 480 seconds 1161622252 N * _mcp mcp 1161622437 Q * prae Quit: Quitte 1161622891 M * bonbons Hey Bertl! 1161623560 M * kevinp looks like the :80 problem was actually a inept administrator inside the vserver, thanks for everyone's help 1161623572 M * kevinp (they has set https to listen to 80) 1161623755 M * Bertl good to hear! 1161623764 M * Bertl tx for clarifying! 1161624300 M * gdm wi 94 1161625364 J * bronson__ ~bronson@66.160.177.217 1161626314 J * dodgi ~no@host81-132-0-237.range81-132.btcentralplus.com 1161626478 M * dodgi sorry i have problem 1161626490 M * dodgi Im in instaling stage3 1161626497 M * dodgi and i see --interface eth0:192.168.1.253/24 \ 1161626514 M * dodgi i dont have any ip 192. in ifconfig 1161626519 M * dodgi what should i write? 1161626532 M * Bertl well, depends on what ips you want for your guests 1161626555 M * Bertl this would assigne a private ip 192.168.1.253 to the guest 1161626566 M * dodgi but i have only one ip 1161626576 M * dodgi like 80.34.250.3 1161626581 M * Bertl and what is the guest supposed to do/provide? 1161626594 M * dodgi err? 1161626610 M * dodgi what u mean? 1161626618 M * Bertl well, for example, should it be reachable from outside? 1161626627 M * dodgi yes 1161626657 M * Bertl then you need to either use a public ip, or utilize DNAT to map certain ports 1161626677 M * Bertl otherwise the 'world' will not be able to reach it, right? 1161626690 M * dodgi can i use this same which im using now? 1161626698 M * dodgi yes right 1161626714 M * Bertl sure, but if you assign that to the guest, services will clash with the host ip 1161626726 M * dodgi alright 1161626735 M * dodgi i will go read sorry for messing 1161626762 M * Bertl np, that's why I asked 'for what purpose you want that guest' 1161627911 M * hardwire he wasn't judging you :) 1161627912 Q * derjohn2 Read error: Connection reset by peer 1161627947 J * derjohn2 ~aj@dslb-084-058-252-217.pools.arcor-ip.net 1161628249 M * hardwire meh! 1161629160 M * hardwire ok.. how does super mount work? 1161629168 A * hardwire is totally lost in this respect 1161629197 M * Bertl hmm, last time I checked, it wasn't in mainline 1161629753 M * hardwire meh 1161629755 M * hardwire secure mount 1161629824 M * Bertl well, it 'jsut' works when you enable it ... 1161629827 M * Bertl *just 1161629837 M * Bertl no magic, no tricks 1161629932 M * hardwire I guess I don't understand the procedure 1161629969 M * Bertl well, ask some questions, maybe we can answer them? 1161630016 M * hardwire how do I bind mount a host directory outside of the vdir .. into a virtual host 1161630030 Q * bronson__ Ping timeout: 480 seconds 1161630039 M * Bertl ah, you are talking about the userspace utility! 1161630041 J * bronson__ ~bronson@adsl-64-161-106-11.dsl.snfc21.pacbell.net 1161630058 M * hardwire I have no idea what I am talking about 1161630065 M * hardwire just that I need to.. and I have no idea how 1161630077 M * Bertl well, you put the 'host' path and the guest path into the fstab (config) 1161630089 M * hardwire weird 1161630136 M * hardwire with type none fs and bind in option right? 1161630200 M * Bertl yep 1161630266 M * hardwire and then.. start it? 1161630277 M * kevinp Bertl: Where's your changelog again? There needs to be a link to that from the out of date page on the site (http://linux-vserver.org/ChangeLog-2.1) 1161630282 M * Bertl hardwire: yup 1161630292 M * kevinp just curious what changed in rc42 1161630298 M * hardwire Bertl: ok 1161630310 M * Bertl kevinp: derjohn2 is maintaining the changelogs now 1161630382 M * hardwire vnamespace -e 49000 mount -t none -o bind /var/cache/rsnapshot/daily.1/vservers/anc-www-01/ /var/lib/vservers/anc-www-01/mnt/ 1161630385 M * hardwire danke 1161630399 M * hardwire guess I just needed to know that it would work w/o fubar 1161630633 M * hardwire why is /tmp as tmpfs preferred? 1161630642 M * hardwire just sos I knows 1161630670 M * Bertl because of performance reasons 1161630687 M * Bertl tmpfs is quite fast for temporary files 1161630695 M * hardwire I can see that on larger systems with many vservers 1161630705 M * hardwire Bertl: well yeh.. its ram :) 1161630728 M * Bertl usually, yes 1161631244 M * hardwire its optimized to store in swap as well? 1161631368 M * Bertl yup, tmpfs is just living in the page cache 1161631444 Q * lilalinux Remote host closed the connection 1161631464 M * waldi # vserver dns enter 1161631464 M * waldi vlogin: select(): Function not implemented 1161631468 M * waldi hui 1161631486 M * bonbons Bertl: is there some showstopper for 2.1.1? 1161631503 M * daniel_hozac waldi: wtf? 1161631535 M * Bertl bonbons: currently I'm investigating a procfs issue 1161631556 M * bonbons ok, what kind of issue? 1161631567 M * Bertl http://paste.linux-vserver.org/538 1161631572 M * waldi oldselect([...][...]) = -1 ENOSYS (Function not implemented) 1161631580 M * waldi it uses the wrong syscall 1161631589 M * daniel_hozac waldi: arch? 1161631592 M * waldi powerpc 1161631603 M * waldi ppc64 1161631603 Q * sladen Ping timeout: 480 seconds 1161631622 M * Bertl bonbons: we are hitting the BUG_ON() checking for I_CLEAR (probably caused by procfs hiding) 1161631659 M * Bertl bonbons: it seems to happen on x86_64 SMP only atm, couldn't recreate it on x86 yet 1161631671 M * Bertl bonbons: which makes is even stranger 1161631688 M * bonbons ok... 1161631728 J * sladen paul@starsky.19inch.net 1161631749 M * daniel_hozac waldi: looks like include/asm-powerpc/unistd.h still lists oldselect as the regular select. 1161631827 M * waldi hu? 1161631862 M * waldi hmm 1161632009 M * waldi yep, 82 is listed as oldselect by strace 1161632017 M * waldi but why does it not longer work ... 1161632111 M * waldi anyway, this needs to be fixed ... 1161632261 M * waldi SYSX(sys_ni_syscall,sys_ni_syscall,ppc_select) 1161632271 M * waldi this is syscall 82 1161632282 M * daniel_hozac yeah. 1161632352 M * waldi i think this should be fixed in the kernel 1161632371 M * daniel_hozac i agree. 1161632371 M * hardwire hmm 1161632383 M * hardwire I think vserver is too abstract for me 1161632394 M * waldi use xen 1161632397 M * daniel_hozac either _newselect should be select, or select should be implemented on all arches. 1161632410 M * hardwire waldi: I can use vserver.. its just too weird for me 1161632420 M * hardwire for my purposes vserver works best 1161632482 M * waldi daniel_hozac: it should always use _:newselect 1161632493 M * waldi daniel_hozac: select is the old kernel interface 1161632495 M * hardwire like whats vcontext actually do on the host 1161632498 M * hardwire its weird to me 1161632510 M * daniel_hozac waldi: ok. 1161632513 M * hardwire err 1161632515 M * hardwire vnamespace 1161632520 M * daniel_hozac hardwire: it enters a namespace. 1161632529 M * Bertl hardwire: that's not even part of Linux-VServer actually 1161632531 M * daniel_hozac or cleans it, or sets it. 1161632537 M * Bertl hardwire: that's a mainline feature :) 1161632549 M * daniel_hozac well, entering isn't supported in mainline, is it? 1161632549 M * waldi dietlibc maps select to _newselect on s390 but not on ppc 1161632654 M * waldi oh, even i386 needs _newselect now 1161632672 M * waldi but i'm not sure how the interface differs 1161632709 M * Bertl dietlibc 0.30? 1161632718 M * waldi yes 1161632745 M * Bertl hmm, then somebody should check the cvs and/or notify fefe 1161632814 M * waldi and I need to write a bug 1161632837 M * hardwire Bertl: mainline as in running processes on the host as a certain XID? 1161632846 M * hardwire for.. typically. security and filesystem supported reasons? 1161632863 M * hardwire it just so happens that linux-vservers are a fancy chroot using an XID when they run eh? 1161632885 M * Bertl hardwire: you were talking about vnamespace, remember? 1161632896 M * hardwire *blink* 1161632898 M * hardwire yup 1161632899 M * Bertl that is just a tool which allows to 'enter' namespaces 1161632911 M * hardwire how does it relate to the vserver utility 1161632912 M * Bertl and private namespaces are part of mainline since 2.4 1161632923 M * hardwire what is mainline? 1161632934 M * hardwire the.. kernel? 1161632946 M * Bertl mainline means the kernel from ekrnel.org 1161632954 M * hardwire :) 1161632969 J * _are_ ~are@62.112.159.81 1161632972 M * hardwire had to make sure 1161632981 M * Bertl so, you can 'experience' namespaces with vanilla kernels 1161632986 M * hardwire well I guess I will check that tout when I have time 1161633057 M * Bertl you can see it like different 'views' into the same 'real' setup 1161633175 M * waldi hrm, I don't find the select implementation in the glibc 1161633190 M * waldi Bertl: glibc uses _newselect for all linux arches 1161633192 Q * ms_ Ping timeout: 480 seconds 1161633201 M * hardwire meh 1161633254 M * Bertl waldi: so that should probably be updated for dietlibc too 1161633258 M * waldi yes 1161633946 M * waldi bug written 1161634010 M * m4z hey, is there a known reverse dns issue with vserver 2.0.1? 1161634042 M * daniel_hozac ... how are the two items related? 1161634060 M * m4z i can do dns from within a guest just fine, but reverse dns (like apache's HostnameLookups) doesn't work at all 1161634090 M * m4z and i have so no idea why 1161634246 M * Bertl did you check with dig? 1161634321 M * Bertl and what about using tcpdump on the host to check the dns communication? 1161634323 Q * teukka Read error: Connection reset by peer 1161634336 M * m4z not yet but the apache config is from a working system... i'll try that after mysql is emerged (yes, gentoo, linux for the patient (;) 1161634450 M * m4z i have some fascist firewall rules and changed all relevant drops to being logged so i think i can tell all the guest tries to do is use itself as the nameserver although the correct nameserver is in resolv.conf 1161634485 M * Bertl well, that would point to /etc/resolv.conf being wrong then? 1161634538 M * Bertl m4z: maybe your firewall has some dns udp vs tcp issue? 1161634555 M * Bertl i.e. it allows queries via udp but not tcp (or vice versa?) 1161634575 M * m4z i doubt i can do anything wrong with "nameserver 192.168.0.1" in resolv.conf 1161634592 M * Bertl that looks like a local ip to me :) 1161634601 M * m4z hmmm, but then the tcp queries would be logged by iptables 1161634614 M * Bertl what ip does the guest have? 1161634618 M * m4z yes, thats the local box doing dns 1161634629 M * m4z the guest is .20 1161634637 M * m4z the host .2 1161634664 M * Bertl and you don't happen to have some weird masquerading/nat-ing on that host? 1161634711 M * m4z no, just forwarding port 80 and 443 to the guest, the rest is for the host 1161634793 M * m4z that's a gentoo-guest in a gentoo-host. i have tried debootstrapping a sarge and with that there even dns doesn't work (but there are some other problems i have to look into rsn) 1161634829 M * m4z like whenever i start a vserver, the hosts eth0 comes up... 1161634859 M * m4z don't know if thats wanted behavior, i'm just fiddling with it for some days now 1161634916 M * Bertl well, that's your configuration 1161634935 M * Bertl if you tell the tools to mess with your interfaces, they'll do so :) 1161634965 M * m4z hmm, i'll have to learn much more i guess (; 1161635066 J * teukka ~teukka@201.254.110.242 1161635204 P * teukka 1161635383 M * m4z well, i'll continue tomorrow - good night! 1161635464 M * Bertl k, cya!" 1161635895 M * derjohn Bertl, any new about the rc42? No further objections came in? 1161635918 M * Bertl well, we have some new issues .. so the final release is delayed 1161636095 Q * Rich_Estill_afk Quit: Leaving 1161636121 Q * dna_ Quit: Verlassend 1161636384 M * kevinp Bertl: Occasionally untarring a large tarball inside the host server locks up my session. This has happened on multiple vserver hosts runng different versions of 2.x and I can't reproduce it. I start a new session and try it again and it works. I have never seen this on a non vserver host. Any ideas? 1161636427 M * Bertl how do you untar it? and what do you mean by lockup? 1161636457 M * kevinp tar -xzvf filename.tar.gz, session completely freezes, only way out is to close my terminal window 1161636476 M * kevinp Even if I kill the tar process from another session, it never comes back 1161636489 M * kevinp the tar process never finishes either 1161636500 M * Bertl so it keeps hanging around? 1161636512 M * daniel_hozac anything in dmesg? 1161636530 M * kevinp I have one right now what info you want? 1161636581 M * Bertl first, ps auxwww would be interesting 1161636607 M * Bertl if that shows something else than 'D', an strace could be interesting too (attach with -p) 1161636703 N * Osgi|AW Osgiliath 1161636705 M * Osgiliath hi there 1161636762 M * kevinp http://paste.linux-vserver.org/551 for ps auxwww 1161636779 M * derjohn kevinp, which kernel/vserver vsersion/rc ? 1161636788 M * kevinp rc42 on this one 1161636793 M * kevinp sorry rc41 1161636819 M * kevinp 2.6.18.1-vs2.1.1-rc41 1161636821 M * Bertl S looks good, so it's probably waiting for something 1161636837 M * Bertl let's try to attach to it from inside the guest, with strace 1161636844 M * derjohn hm, I would have bet it similar to the meanwhile solved mysql-server problem on rc <= 36 .. but rc41 is fixed in that manner. 1161636848 M * Osgiliath hi Bertl 1161636875 M * Osgiliath hi morrigan (don't know if you're away) 1161636878 M * kevinp There is no guest to go into, I was doing it from the host 1161636908 M * Bertl so the tar is hanging on the host? 1161636912 M * kevinp correct 1161636915 M * derjohn kevinp, did you set limits ? 1161636919 M * Bertl let's use strace there then 1161636923 M * Bertl hey Osgiliath! 1161636939 M * Bertl Osgiliath: any news regarding your oracle setup? 1161636940 M * kevinp How do I do a strace on a running process? 1161636942 M * derjohn tyr cat /proc/virtual//limits to see if your untarring hit a max. 1161636956 M * Bertl derjohn: it's on the host 1161636963 M * kevinp derjohn, I'm in the host not a guest 1161636978 M * Osgiliath it's installing .... and it goes alright for now 1161636997 M * kevinp I'm in the process of creating a new guest, copied from another host 1161637002 M * derjohn kevinp, eh, the host locks up? I guess it waits for disks .... any broken raid disk maybe ? 1161637011 M * Osgiliath the listener started correctly, now it's building the database 1161637020 M * Bertl derjohn: then it would be in 'D' state 1161637031 M * derjohn kevinp, gdb program 1234 1161637038 M * kevinp /proc/mdstat shows everything is fine 1161637134 M * kevinp so, gdb program 6819 or gdb tar 6819? 1161637148 M * Bertl kevinp: try strace 1161637167 M * Bertl strace -fF -p 6819 1161637172 M * kevinp ok 1161637230 M * kevinp it's gone! 1161637286 M * kevinp it definitely didn't finish, not all the files are there, but there is no tar process running anymore 1161637308 M * kevinp and of course, my session is still locked at the same place 1161637316 M * Bertl did the strace give you soemthing? 1161637340 M * Bertl it looks more like a network effect to me 1161637355 M * Bertl i.e. you ahve a stream of data (tcp) running 1161637373 M * Bertl and at some point, you lose connectivity (or routing changes drastically) 1161637386 J * Aiken ~james@tooax6-044.dialup.optusnet.com.au 1161637407 M * Bertl this will look like a lockup, and will keep the session around until it times out 1161637415 M * kevinp attach: ptrace(PTRACE_ATTACH, ...): No such process 1161637428 M * Bertl at which point, the processes will get a HUP 1161637474 M * kevinp hmm, I don't believe I have ever had this on a non-vserver host, and all the servers are in the next room 1161637495 M * kevinp and my other terminals to that server aren't interrupted 1161637500 M * Bertl well, if it happens to you on the host, it's the normal entwork stack 1161637574 M * Bertl okay, I think I'm off for a nap now, kind of tired ... 1161637581 M * Bertl will probably back later ... 1161637590 N * Bertl Bertl_zZ 1161637594 M * kevinp looks like your right (as usual), my terminal just came back with a connection timed out error 1161637709 M * doener kevinp: autossh + screen works nicely to survive short network problems (eg. my daily dsl connection drop) 1161637924 Q * bonbons Quit: Leaving 1161638023 A * derjohn looks up autossh 1161638250 M * doener sth. like: autossh -- $HOST -t screen -RS `date +%s` is nice. Will create an (almost) unique screen session and reconnect you to that session in case that you loose your connection 1161638278 M * doener the session name is just used to cope with multiple screen sessions, so that you are reconnected to the right one 1161639171 M * Hollow started a preliminary documentation structure, please comment: http://wiki.linux-vserver.org/Documentation 1161639176 Q * Piet Ping timeout: 480 seconds 1161639256 M * Loki|muh Hollow: isn't Arch a Distribution? 1161639263 J * Piet hiddenserv@tor.noreply.org 1161639285 J * rgl Rui@217.129.151.190 1161639287 M * rgl hello 1161639292 M * doener yep, needs to go down 1161639297 M * Hollow Loki|muh: yep, i just noticed it.. 1161639306 M * Loki|muh ok :) 1161639336 M * doener Hollow: btw, why do I get the annoying Google Maps warnings? 1161639364 M * Hollow because the api key is issued for linux-vserver.org, not wiki.linux-vserver.org... :/ 1161639396 M * doener Hollow: but I don't even see any Google Maps content on the edit page 1161639417 M * Hollow i guess the plugin loads nevertheless... 1161639445 M * Hollow once the migration is finished wiki.* will go away anyway 1161639448 M * doener It does not on the existing pages 1161639461 M * doener only if you get onto the edit page 1161639470 M * Hollow well, on the developers page it does.. 1161639477 M * Hollow that's where the map is 1161639477 M * rgl how do I setup a guest interface to use an eth0 alias like, eth0:5 ? I tried to use it inside interfaces/1/dev, but vserver start echos Cannot find device "eth0:5", but it still shows the right IPs inside the guest... what I'm doing wrong? :( 1161639493 M * Hollow but i guess we could easily live without this map 1161639501 M * doener Hollow: yeah, and there it is expected and somewhat ok 1161639509 M * doener *lol* "In order to see the map that would go in this space, please use a compatible web browser" 1161639514 M * daniel_hozac rgl: echo 5 > .../interfaces/1/name 1161639524 M * doener instead of the developer page 1161639526 M * daniel_hozac rgl: and eth0 in .../interfaces/1/dev 1161639527 M * Hollow doener: lol? which browser? 1161639535 M * Hollow ah.. yeah 1161639537 M * doener Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.8.0.7) Gecko/20060830 Firefox/1.5.0.7 (Debian-1.5.dfsg+1.5.0.7-2) 1161639539 M * Hollow if the api key is wrong.. 1161639546 M * rgl daniel_hozac, ah! thx :) 1161639569 M * Hollow well, i tried to switch api keys with $_SERVER['HTTP_HOST'] in mediawiki config, but it didn't work out somehow.. 1161639575 M * doener stupid Google 1161639651 M * Hollow even the fact the you need an api key annoys me a lot.. 1161639686 M * Hollow anyway, do you think the structure makes sense so far? 1161639728 M * doener looks good to me 1161639740 M * rgl daniel_hozac, humm it now shows: "RTNETLINK answers: File exists" any ideia why? 1161639754 M * daniel_hozac rgl: because you had already assigned that IP address to the interface? 1161639785 M * Hollow ok, guess we still need a categories for util-vserver and vcd documentation (in addition to those doc integrated to other pages) 1161639791 M * rgl daniel_hozac, if you mean on the host? then yes. I should not do that? 1161639808 M * daniel_hozac rgl: if you do, you don't need to tell the utils to do it as well. 1161639833 M * rgl daniel_hozac, humm, how will they see the alias interfaces? 1161639852 M * daniel_hozac rgl: you don't assign interfaces at all, you assign IP addresses. 1161639875 M * daniel_hozac whatever interface it is on doesn't really matter that much. 1161639899 M * rgl daniel_hozac, humm, so I don't need to have a interfaces/0/* ? 1161639904 M * daniel_hozac you only tell the utils about the interface so that it can create/remove it on start/stop. 1161639916 M * daniel_hozac rgl: of course you do, that's where you assign IP addresses. 1161639933 J * ms_ ~ms@arkansas.doc.ic.ac.uk 1161639973 M * rgl daniel_hozac, humm, ok, I have an interfaces/0/dev with eth0. but how about eth0:5 alias? 1161639990 M * daniel_hozac what about it? i already told you how to specify that. 1161640007 M * rgl daniel_hozac, unless I add them into interfaces/1/dev it wont show up in guest ifconfig 1161640036 M * daniel_hozac yes, it will, as long as the _IP address_ assigned to eth0:5 is assigned to the guest. 1161640080 M * rgl daniel_hozac, ah, so I don't need to have that interfaces/1/name? 1161640084 M * daniel_hozac if you don't want the utils adding/removing the IP addresses, create the file interfaces/x/nodev. 1161640124 M * daniel_hozac you do if you want the utils to create eth0:5 when you start that guest. 1161640260 Q * Osgiliath Quit: The Sign Of Wrath Awaked 1161640284 M * rgl daniel_hozac, I don't get it :( 1161640335 M * rgl daniel_hozac, I've created the interfaces/1/dev with eth0 and 1/ip with the ip, but when I stop the vserver, it destroys all host aliases :| 1161640346 M * rgl er all host eth0 aliases 1161640378 M * rgl and when I start it, it wont creates the aliased interface for eth0:5 :| 1161640497 J * Osgiliath ~abigor@kurzweil.no-ip.org 1161640522 M * rgl oh, but it shows up in "ip addr ls" 1161640523 M * doener rgl: if the ip address is already assigned on the host, you use ip and nodev 1161640543 M * doener if the tools shall create and destroy the address, you use ip and dev 1161640564 M * doener and for the latter if you want an alias (not needed at all), you use ip, dev and name 1161640606 M * kevinp derjohn where are the latest changelogs kept? just checking to see if there is a reason to go to rc42 1161640668 M * derjohn kevinp, there are no changelogs. sry,but one one pages or informed me about the changes. but I remember there was a reeeeeeal problem with the init system. 1161640673 M * rgl doener, how about, ip nodev and name? 1161640683 M * daniel_hozac rgl: and how does that make sense? 1161640694 M * rgl doener, for not creating the alias device? 1161640700 M * derjohn kevinp, under some circumstances the guest didnt start correctly. But I followd the discussion with only one ear ... 1161640726 M * kevinp ok, no startup problems for me so far, so I'll wait a little longer 1161640727 M * derjohn kevinp, the next changelog is hopefully the devel-release 1161640738 M * doener rgl: pretty pointless 1161640740 M * kevinp that would be great 1161640763 M * daniel_hozac kevinp: history dumping fixed, init can exit, and some config restructuring (enable legacy networking instead of disable :)) 1161640768 M * doener rgl: what would the name do? there's nothing to be created or destroyed 1161640782 M * doener rgl: the isolation is completely ip address based 1161640802 M * doener rgl: whatever interface (alias) the ip address is assigned to the vserver will see 1161640819 M * rgl doener, humm, so I should not have the ip on the host at all? 1161640852 N * bronson__ bronson 1161640853 M * doener rgl: that's up to you... if you want it to stay around all the time, set it up yourself and use nodev. 1161640876 M * doener rgl: if you want it to be automatically created and destroyed on vserver start/stop, use ip, dev and optionally name 1161640884 M * bronson I'm trying to copy a vserver but I get: vserver-copy: Vserver file "/etc/vservers/rinspin.conf" does not exist 1161640891 M * bronson True, it doesn't. 1161640897 M * kevinp thanks daniel_hozac 1161640898 M * bronson But /etc/vservers/rinspin/* exists. 1161640916 M * bronson Is vserver-copy looking for an antiquated file or something? 1161640989 M * bronson My command line: vserver-copy -v --domain trestle.com --ip 192.168.1.9 rinspin trestle 1161641234 M * bronson Anyone know why vserver-copy isn't working for me? 1161641312 J * debugger Rui@217.129.151.190 1161641314 M * kevinp I guess the real question is whether or not vserver-copy is even for use with 2.x 1161641320 M * kevinp I think the answer is no 1161641346 M * bronson kevinp: you're right. 1161641347 M * bronson http://www.mail-archive.com/vserver@list.linux-vserver.org/msg09507.html 1161641363 M * kevinp Some people have written scripts to do the same thing, but no one has done it in the mainline tools yet 1161641397 M * bronson So... Is there an easy way for me to copy vservers? Or should I just bite the bullet and build new ones each time? 1161641451 M * kevinp I don't know how other's do it, but I just wrote a script that uses tar and grabs the files from my standard skel and makes a new one in a different folder 1161641472 N * debugger rgl_ 1161641482 M * kevinp I think other's use the -skel option on the build then copy files 1161641486 M * rgl_ doener, daniel_hozac, I think I get it. thx :) 1161641512 M * rgl_ doener, daniel_hozac, thou, I don't understand why all host aliases are removed when I stop a guess 1161641531 M * kevinp bronson: just two folders to worry about /etc/vservers/vservername and /vservers/vservername 1161641563 M * doener rgl_: are the addresses all on the same subnet? If so, and the vserver has the primary address for that network, all secondaries will be removed with it 1161641635 M * bronson kevinp: so just copy them, then edit /etc/vservers/vservername/* to fit? 1161641651 M * bronson Sounds reasonable. 1161641664 M * kevinp exactly 1161641669 M * rgl_ doener, indeed, they are all on the same subnet. the primary is the one that does have the "secondary" word on "ip addr ls"? that is: "scope global secondary eth0:1" is not the primary? 1161641674 M * bronson kevinp: cool, thanks. 1161641692 M * rgl_ doener, err "that does _not_ have" 1161641703 M * doener rgl_: yep, as it is not secondary, it is primary ;) 1161641732 M * rgl_ doener, humm, so I can't assign the primary to any guest? 1161641777 Q * rgl Ping timeout: 480 seconds 1161641838 M * doener you can, either make that one nodev or, on recent kernels, enable some sysctl setting to change the kernel behaviour, so that a secondary address is promoted to become the primary address instead of removing all secondaries 1161641849 M * doener (I don't remember which setting that is) 1161641870 M * derjohn n8 folks, /me away now 1161642151 M * Hollow off to bed too, have a good night! 1161642652 Q * dodgi Ping timeout: 480 seconds 1161642720 Q * meandtheshell Quit: exit (0); 1161643032 M * bronson /etc/vservers/NAME/cache and /etc/vservers/NAME/run are both broken links for me. 1161643042 M * bronson But it doesn't seem to affect anything! 1161643643 M * bronson What is the cache? Sez "Path of the storage area for cached information about this vserver" ... but what information is cached? 1161643680 J * kaner_ kaner@strace.org 1161643696 M * doener bronson: is the vserver running? might be, that the symlink is only valid then 1161643706 Q * kaner Ping timeout: 480 seconds 1161643713 M * bronson doener: ah, that's why. It's not running. 1161643718 M * bronson I'm in the middle of trying to copy it. 1161645580 M * bronson Here are some notes on how to copy vservers: http://wiki.u32.net/index.php?title=Linux-VServer 1161645584 M * bronson fwiw 1161645925 J * dodgi ~no@host81-132-0-237.range81-132.btcentralplus.com 1161645931 M * dodgi Hi 1161645935 N * Bertl_zZ Bertl 1161645942 M * Bertl back now ... 1161645977 M * dodgi yup :) 1161645983 M * dodgi Bertl all working ;> 1161645994 M * Bertl good! 1161646004 M * dodgi Bertl: could u just tell me can i run proftpd or sshd o it? 1161646130 M * Bertl well, I don't use proftpd, but sshd is running on every guest here 1161646144 M * Bertl and IIRC, proftpd works just fine too 1161646172 M * dodgi could u give me some paper how to do it i mean about sshd? 1161646217 M * Bertl hmm, what's the problem you see there? i.e. what do you want to do? 1161646245 M * dodgi localhost ~ # emerge openssh 1161646245 M * dodgi !!! ARCH is not set... Are you missing the /etc/make.profile symlink? 1161646245 M * dodgi !!! Is the symlink correct? Is your portage tree complete? 1161646287 M * Bertl hmm, that looks like a gentoo issue to me, but probably just some missing stuff 1161646303 J * s0undt3ch_ ~s0undt3ch@81.193.57.83 1161646318 M * Bertl Hollow already went to bed, he probably would know that out of his head 1161646341 M * Bertl dodgi: you know, there is a 'special' guest layout, yes? 1161646343 M * dodgi Bertl can i copy make.con from normal server to this vserver? 1161646371 M * Bertl I'd assume so, everything working on the host should work in the guest too 1161646389 Q * s0undt3ch Killed (NickServ (GHOST command used by s0undt3ch_)) 1161646389 N * s0undt3ch_ s0undt3ch 1161646431 M * dodgi cool 1161646432 M * dodgi brb 1161646789 M * dodgi shyt :/ 1161647625 J * kaner kaner@strace.org 1161647734 Q * kaner_ Read error: No route to host 1161647890 M * rgl_ g'night guys 1161647899 M * Bertl night! 1161647913 M * rgl_ hey H ;) 1161647917 M * Bertl dodgi: what's the problem? 1161647936 Q * rgl_ Quit: Fui embora