--- Day changed --- Log opened Thu Jun 12 00:00:01 2003 03:54 -!- radical [~kernet@202.88.184.39] has joined #uml 04:03 -!- avoozzzl is now known as avoozl 06:47 -!- _slh__ [~chatzilla@68.19.227.139] has joined #uml 06:49 -!- radical [~kernet@202.88.184.39] has quit [Ping timeout: 488 seconds] 07:58 < Run> ... 09:15 < Run> damn 09:15 < Run> This uml crashes a bit too quick :/ 09:17 < Run> Anyone here who can help me? 09:17 < Run> I was compiling something in uml... but the compiler never returned to the command prompt. 09:18 < Run> So, I stopped the uml (which was running inside gdb) 09:18 < Run> and then I find this: 09:18 < Run> (gdb) bt 09:18 < Run> #0 0x400c80a5 in nanosleep () from /lib/i686/libc.so.6 09:18 < Run> #1 0xa001ceda in idle_sleep (secs=-4) at arch/um/kernel/time.c:115 09:18 < Run> How is that possible? It is supposed to be compiling! 09:39 -!- jdike [~jdike@129.170.249.105] has joined #uml 09:40 < jdike> hi guys 09:40 < Run> hi 09:41 < Run> jdike: I am still using 2.5.69 with uml-patch-2.5.69-1 and the patch that I mailed to user-mode-linux-devel 09:42 < Run> ... and the kernel (uml) freezes -heh- often. Even when I do nothing whatsoever. 09:42 < Run> I was wondering if you have some new patch on top of this that might fix that in the meantime. 09:42 < avoozl> even if you do not run it inside gdb? 09:42 < Run> yes 09:43 < avoozl> i've only tried 2.4.20 :( 09:44 < jdike> Run: well, can you get stack traces from the hangs 09:45 < Run> #0 0x400c80a5 in nanosleep () from /lib/i686/libc.so.6 09:45 < Run> #1 0xa001ceda in idle_sleep (secs=-4) at arch/um/kernel/time.c:115 09:45 < Run> Thats it :/ 09:46 < Run> Yet, I couldn't login in anymore - it never returned from a running application - and uml_mconsole wasn't able to do anything (control-alt-delete not, sysrq not and reboot not. 'halt' froze too, never returned). 09:47 < Run> That is, uml_mconsole said 'OK' to everything - but nothing happened. 09:47 * green have only 2 problems with 2.5.70-um, those are: 1. hangs on shutdown. 2. Leaks memory on exec 09:48 < Run> Maybe its fixed in .70 then. 09:48 < Run> Where do I get the uml-patch-2.5.70 ? 09:49 < Run> green: Do you know anything about NPTL/TLS? 09:50 < Run> The only reason I am trying to use uml is because the kernel hangs with my program, I am trying to figure out why. 09:50 < Run> The program uses nptl, thus. 09:52 < Run> Well, it's not really important anyway... I think it only happens due to a bug in my program. So, when I fix that bug, then the kernel won't barf either :/ 09:52 < Run> People could use this as a DoS I suppose... but I have no time to keep pulling peoples jackets. 09:52 < Run> However. 09:52 < Run> Now that I have uml running, I wanted to test my program inside uml while using nptl. 09:53 < Run> And... for somereason it doesn't use it :/ 09:53 < Run> green: You got some time to help? 09:56 < Run> >:-/ 09:56 * Run feels ignored. 09:56 < Run> jdike: Are you coming with a follow-up comment, or do you give up on this problem? 09:57 < Run> jfdike: You never answered my yes/no question by the way. 09:57 < Run> s/f// 09:57 < jdike> Run: what question? 09:58 < Run> If you changed something in a recent patch that might fix sudden hangs on an idle machine. 09:58 < jdike> Run: no 09:58 < Run> I am working on this with a friend (other network) - he mentioned a '.bk7' patch or something. 09:59 * Run doesn't know what that patch does. 10:06 < green> Run: I know there is NTPL thing, that's it 10:07 < green> and there is not 2.5.50-um patch, I just maintain my own uml tree that I try to be as close to Linus' tree as possible 10:08 < green> close to current Linus' bk tree, I mean 10:10 < Run> NPTL, not NTPL... 10:11 < Run> Maybe I'd like to try your 2.5.70 tree then, it is not online? 10:12 < Run> NPTL is the Native POSIX Thread Library, for which heavy kernel support was build into 2.5.41. 10:13 < Run> The old thread stuff (LinuxThreads) will be dumped as of glibc-2.4. glibc-2.3 is a hybrid that switches depending on the availability of ABI version chosen. 10:13 * green always get the order of letters in NPTL incorrectly ;) 10:14 < green> Run: you can try to pull my tree from bk://namesys.com/bk/reiser4-linux-2.5 , it haev some unrelated changes, but this is unimportant as long as you do not select "reiser4" filesystem support 10:14 < Run> I am really having problems understand why inside uml this application doesn't use nptl, but instead falls back to linuxthreads. 10:14 * Run looks 10:15 < Run> http://namesys.com/bk/reiser4-linux-2.5 10:15 < Run> or ftp: ? 10:15 < green> if you use this never RH implementation, I think they now only use nptl if the kernel (as in uname -a) reports "nptl" or if AT_SYSINFO pointer is present. something like this 10:15 < green> no. I mean "bk://...", of course you need bitkeeper for that 10:16 -!- thy_ [koubt9am@u-pl5.ms.mff.cuni.cz] has joined #uml 10:17 < Run> I never heard of bk: before :/ 10:19 < green> http://www.bitmover.com 10:19 < Run> Thanks: 10:19 < Run> Tue Jan 07 2003 Jakub Jelinek 2.3.1-32 10:19 < Run> - update from CVS 10:19 < Run> - don't use TLS libs if kernel doesn't set AT_SYSINFO 10:19 < Run> (#80921, #81212) 10:20 < Run> Now if only I knew what that AT_SYSINFO is and how I can make the kernel do with it what is needed... 10:21 < jdike> Run: look at fs/binfmt_elf.c 10:22 < david> hello 10:22 < Run> jdike: What would I look for? AT_SYSINFO doesn't appear in that file. 10:22 < jdike> hi david 10:22 < Run> /usr/src/kernel/uml/linux-2.5.69>grep AT_SYSINFO fs/binfmt_elf.c 10:22 < Run> /usr/src/kernel/uml/linux-2.5.69> 10:23 < jdike> Run: hmmm, it's part of the ELF format 10:23 < jdike> Run: grep for AT_ 10:27 < Run> /include/asm-i386/elf.h:103:#define AT_SYSINFO 32 10:32 < Run> Ok, it is there... 10:35 < Run> It would help if there as a skas3.patch for 2.5.70 :/ 10:35 < Run> Is there? 10:36 < jdike> Run: no 10:36 < Run> :( 10:36 < Run> I need to run 2.5.something... but then I can't run uml. 10:37 < Run> Or can I? 10:38 < Run> I am now running 2.4.20-skas3 10:38 < Run> Guess I'll reboot... 10:39 < Run> Oh, the uml crashed again by the way. 10:39 < Run> While I was typing in this window. 10:39 < Run> UML:~/c++/libcwd/testsuite> 10:39 < Run> Program terminated with signal SIGSEGV, Segmentation fault. 10:39 < Run> The program no longer exists. 10:39 < Run> The UML:... thing was the last prompt that was sitting there. 10:40 < Run> ...rebooting to 2.5.70 10:40 -!- Run [carlo@node-d-5886.a2000.nl] has quit [Remote host closed the connection] 11:38 -!- DnsInfector [~DnsInfect@81.53.249.207] has joined #uml 11:39 < DnsInfector> hi all 11:40 < david> hello 11:41 < DnsInfector> anyone knows if USAGI patch applies correctly over UML one ? 11:43 -!- smcavoy [~smcavoy@207.61.160.163] has joined #uml 11:45 < jdike> DnsInfector: what is it? 11:46 < david> jdike: IPv6 11:46 < david> DnsInfector: yes, it works fine 11:46 < david> DnsInfector: did you even try it? 11:48 < DnsInfector> yes but not within UML 11:48 < DnsInfector> actually my FW is a USAGI host 11:48 < david> DnsInfector: so, try to apply it to the UML kernel, figure out if there are any rejects and try to work around them 11:49 < DnsInfector> ok i'll give a try then 11:49 < david> last time I tried it worked, but that was ~8mths ago 11:49 < DnsInfector> 8months... 11:49 < DnsInfector> i don't things have changed so much within the last year 11:50 < DnsInfector> +think 11:50 < DnsInfector> i was looking at http://user-mode-linux.sourceforge.net/dl-sf.html 11:50 < DnsInfector> and the skas patch lokks quite old 11:50 < DnsInfector> is there anything fresheer ? 11:51 < smcavoy> when I try and boot a FS that was built with rootstrap I keep getting 'fsck failed.' because device /dev/ubd/0 is RO. Any ideas? 11:51 < jdike> smcavoy: add 'rw' to the command line? 11:52 < david> DnsInfector: the host-skas3 patch is the latest 11:52 < david> DnsInfector: nothing changed with it since then 11:52 < DnsInfector> so i may say "good job" 11:52 < DnsInfector> if nothing changed since Nov 11 2002 11:52 < david> well, actually 11:53 < david> if you have a kernel with the ptrace 'fix' for the 2.4.20 kmod exploit 11:53 < david> you need a one-line modification to the host-skas3 patch 11:53 < david> kernels.usermodelinux.org/host/ 11:53 < jdike> david: that's fixed in 2.4.21-latest, isn't it? 11:53 < jdike> david: and stock skas3 works fine again? 11:54 < david> jdike: honestly, I've not looked 11:54 < david> jdike: there is still a bunch of ptrace stuff broke in it, so I kept in the mm->dumpable to make sur 11:54 < david> e 11:54 < david> jdike: in theory it should be fixed, but I don't have a box that is easily rebooted right now to check 11:55 < smcavoy> jdike: just "rw" or something like "rw=yes" ? 11:56 < david> just 'rw' 11:56 < DnsInfector> that's a long time that i must change my stock MDK kernel with a more up to date one 11:56 < DnsInfector> i'll apply skas patch too 11:59 < smcavoy> hrmm... now I'm getting 'Cannot open root device "ubd0" or 62:00' 12:06 < david> smcavoy: does your rootfs have a /dev/ubd/0? 12:06 < david> maybe you need 'devfs=mount' 12:12 -!- DnsInfector is now known as DnsInfectorAw 12:18 < smcavoy> david: my bad, read only permissions were set for the user that was actually running the uml :) 12:27 -!- DnsInfectorAw is now known as DnsInfector 12:28 < smcavoy> should I be able to resize a ext3 fs with resize2fs? 12:29 < smcavoy> I tried 'resize2fs [fs] 3000' and it gaves me 'resize2fs: No space left on device while trying to resize john_fs' the partiton it's on has over 8gb free? 12:30 < david> smcavoy: should work - did you actually expand the image? 12:31 < smcavoy> david: no, how might that be done? 12:34 < david> dd if=/dev/zero of=your_fs bs=1k count=1 seek=$[2*1024*1024] 12:34 < david> that would make it 2Gb 12:34 < david> then you can expand it 12:37 < smcavoy> well that resized the file, but resize2fs is still giving the same error 12:45 < avoozl> arent you supposed to add a an M after 3000 12:45 < avoozl> otherwise its in 'units' which is pretty obscure 12:49 < smcavoy> avoozl: it doesn't like letters after the #'s 12:56 < avoozl> smcavoy: thats odd. it says here in the manpage that the value should be suffixed by s K M or G 12:56 < avoozl> i might have a more recent resize2fs though, i have a 1.34-WIP 21-may-2003 13:09 < smcavoy> I've got 1.27, standard with Debian 3.0 13:09 < caker> hello 13:10 < jdike> caker: I see you're up and running now 13:10 < caker> jdike: tomorrow the beta ends, and I'll flip the switch - but real launch isn't until Monday (morning) 13:10 < caker> jdike: very exciting 13:11 < jdike> caker: Ah, OK 13:11 < jdike> caker: Yup 13:11 < caker> jdike: thank you :-) 13:11 < jdike> caker: How are you pre-signed-up customers? 13:11 < jdike> s/you/you for/ 13:11 < caker> jdike: had two today signup! 13:11 < david> smcavoy: debian 3.0 is 2.2 kernel based so the tools may not have support for ext3 anyway 13:12 < jdike> caker: what's your capacity, about? 13:12 < caker> jdike: 32 Linode64's and 18 Linode128's with the initial two host machines 13:13 < jdike> caker: what did you end up doing about your arp problems? 13:13 < caker> jdike: ping cron job .... 13:14 < jdike> caker: heh 13:14 < jdike> excuse me while I barf :-) 13:14 < caker> jdike: don't tell anyone :-) 13:21 < david> caker: oh, I figured out a reasonable method for UML redundency 13:21 < david> caker: drbd :-) 13:21 < caker> david: (reading the site now) 13:22 < david> caker: two boxes, network RAID-1, if one box dies just bring the UMLs up on the other 13:22 < caker> that's better than the NAS idea I was brewing 13:22 < david> caker: once someone (*looks at Jeff*) ports swsusp, then you can seemlessly switch UMLs between boxes 13:22 < david> caker: it's cheaper than fiber channel crud 13:23 * jdike whistles innocently 13:23 < caker> david: i dream of total redundancy on the host side (like this or NAS would allow) 13:23 < david> caker: right 13:26 -!- DnsInfector is now known as DnsInfectorEat 13:27 < caker> david: drdb is what I drempt about the other night, awesome that it actually exists! 13:28 < david> caker: heheh :-) 13:40 < smcavoy> david: Debian 3.0 supports both 2.2 and 2.4 and ext3 is basically ext2 (with journal support added), so the same tools should work with ext3 without a hitch... 13:41 < smcavoy> I just tried the same command on a ext2 fs, same problem 13:42 < david> smcavoy: have you tried resizing it within a UML? 13:42 < caker> david: drdb looks awesome - thanks for the info! 13:49 < david> caker: np 14:10 -!- silug [~steve@osiris.silug.org] has joined #uml 14:10 < david> silug: hello 14:11 < silug> hi 14:13 < smcavoy> david: just did, it's giving me the same error 14:17 -!- thy_ [koubt9am@u-pl5.ms.mff.cuni.cz] has quit [Quit: Client exiting] 14:36 < smcavoy> fdisk doesn't seem to recognize existing partitions on /dev/ubd/1, it does see it as being 2gb though 14:58 -!- DnsInfectorEat is now known as DnsInfectorNoz 15:26 < david> smcavoy: does it even have partitions? Probably not 15:32 -!- kara [~kara@64.240.156.226] has joined #uml 15:37 * jdike whacks together a standalone COW driver 15:38 < BB> to mount cow? 15:38 < jdike> BB: Yup, among other things 15:38 < BB> rar 15:39 < jdike> Also to clean up the ubd driver by getting the COW stuff out of it 15:39 < jdike> And to get McMechan to come up with a sane implementation of stackable COW files 15:39 < BB> heh 15:49 < smcavoy> david: it says there's an error in the partition table. 15:57 -!- C0nrad [~proto@finch.ems.mindspring.net] has joined #uml 16:07 -!- Lathiat [lathiat@130.95.13.25] has quit [Remote host closed the connection] 16:07 -!- Lathiat [lathiat@seven.sixlabs.org] has joined #uml 16:10 -!- C0nrad [~proto@finch.ems.mindspring.net] has quit [Quit: Client exiting] 17:21 -!- jdike [~jdike@129.170.249.105] has quit [Quit: Leaving] 18:03 -!- DnsInfector_ [~DnsInfect@ARennes-204-1-9-30.w81-53.abo.wanadoo.fr] has joined #uml 18:08 -!- DnsInfectorNoz [~DnsInfect@81.53.249.207] has quit [Ping timeout: 488 seconds] 18:20 -!- ansoft [~anelson@207.179.107.18] has joined #uml 18:21 -!- smcavoy [~smcavoy@207.61.160.163] has quit [Remote host closed the connection] 18:34 < ansoft> Has anyone had good luck building their own root filesystem? 18:38 < caker> ansoft: yes? 18:41 < ansoft> caker: Did you use any of the pre-made tools? 18:41 < caker> ansoft: tried, ended up just installing on the host onto a partition, then copying it off .. I've done this for about 5 distros 18:43 < ansoft> caker: Crud. Trying to make a bunch of them for my server. I want a clean interface I can use screen to manage. :( 19:31 < ansoft> Anyone had good luck with mkrootfs? 20:18 < mistik1> wow 20:19 < mistik1> caker: Have you ever come across a setup that was using eth0 for a DSL connection as well as for LAN 20:20 < caker> yeah? 20:20 < mistik1> really 20:20 < caker> I've seen friends of mine connect a DSL/cable modem to a hub, then the computer to that hub (along with others in the private subnet) 20:20 < caker> and it works 20:20 < caker> go figure 20:20 < mistik1> I had never seen that kind of setup before 30 minutes ago 20:20 < mistik1> that was it exactly 20:21 < mistik1> he was using pptp for DSL 20:21 < caker> i got it to work once, but I had to power-on the comps in a specific order 20:21 < mistik1> indeed 20:22 < mistik1> well i've managed to test this little script on about 10 machines with completely different LAN/WAN setups 20:22 < mistik1> david is my hero ;) 20:23 < caker> mine too 20:24 < mistik1> I'm quite good with networking but he is supreme with that bridge stuff 20:36 -!- DnsInfector_ [~DnsInfect@ARennes-204-1-9-30.w81-53.abo.wanadoo.fr] has left #uml [Client exiting] 21:10 -!- Koro [peterson@matrix.sedal.usyd.edu.au] has joined #uml 21:11 < Koro> have a quick UML qn ... anyone about? 21:11 < Koro> can't get TUN/TAP working 21:11 < caker> Koro: shoot :-) 21:11 < caker> Koro what error do you get? 21:11 < Koro> have followed rusty's guide, UML HOWTO section 6.7 21:11 < Koro> i set everything up 21:11 < Koro> and then I go into the UML and "ifconfig eth0 129.78.24.102" 21:12 < Koro> when I try to ping 129.78.24.102 from within the UML I get ping: sendto: Invalid argument 21:12 < Koro> ping: wrote 129.78.24.102 64 chars, ret=-1 21:13 < caker> what's the command line for your uml look like? 21:13 < Koro> linux ubd0=/home/peterson/uml/deb1 con=pty con0=fd:0,fd:1 eth0=tuntap,tap1 21:13 < caker> ok 21:13 < Koro> :) 21:13 < caker> are you bridging? 21:14 < Koro> sorry i don't know what you mean 21:14 < Koro> :) 21:14 < Koro> i am a bit of a UML n00b 21:14 < caker> let me go look at the howto 21:14 < Koro> k 21:14 < Koro> thanks 21:14 < caker> 6.7 you said? 21:14 < Koro> yep 21:14 < Koro> rusty's UML howto 21:14 < Koro> "TUN/TAP with a preconfigured device" 21:14 < caker> how are you configuring the tap device? 21:15 < caker> what's it's IP? 21:15 < Koro> $ /sbin/ifconfig tap1 192.168.0.101 up 21:15 < Koro> then I echo a 1 to ip_forward 21:15 < Koro> then /sbin/route add -host 192.168.0.101 dev tap1 21:15 < caker> 102 21:16 < Koro> sorry? 21:16 < caker> should be 102, methinks 21:16 < Koro> ahh 21:16 < Koro> heh i think i thought it was a typo :) 21:16 < Koro> but it's the other end of the link right? 21:16 < Koro> lol 21:16 < caker> its confusing 21:16 < Koro> hehll yeah 21:16 < Koro> ;) 21:16 < caker> you're adding a route to the UML's ip, but through the tap device 21:16 < Koro> roght 21:16 < Koro> ok 21:16 < Koro> so i need 2 ip's ? 21:17 < caker> yes, OR set the tap device to your hosts IP 21:17 < caker> works fine (I **think**) 21:17 < Koro> can i do this for multiple tap devices? 21:17 < caker> yes 21:17 < Koro> so ifconfig tapX 129.78.24.101 up 21:17 < Koro> (that's my host kernel box's IP) 21:17 < caker> right 21:18 < Koro> cool i will try it 21:18 < Koro> thanks very much for the help caker ;) 21:18 < caker> Koro: you're very welcome :-) 21:18 < Koro> heh btw UML rocks ;) 21:18 < Koro> looking forward to getting it networked ;) 21:19 < Koro> oh, caker: so if I set tap0 to be 129.78.24.101 (the host IP), do I still set the route add -host 192.168.0.101 (i.e. private IP) ? 21:19 < Koro> will it work across these network subnets? 21:20 < caker> route add -host **UMLIP** dev tapDevice 21:20 < caker> hmm 21:21 < caker> I'd bring up eth0:1 with a 192.168.0 address, and make that the gateway (masq) for the UMLs in the 192. subet 21:21 < caker> subnet even 21:21 < caker> that should work 21:21 < caker> Do you have real IPs for them? 21:22 < Koro> I want to run 2 UML's, one on 129.78.24.102, and the other on 129.78.24.103 21:22 < Koro> so, yes 21:22 < caker> Then forget what I just said 21:22 < Koro> okay 21:22 < Koro> lol 21:23 < caker> add the route with the real ips then, and make sure you proxy arp - that will make your switch recognize to send traffic for those IPs to your eth0 21:23 < Koro> okay, using the "echo 1 > .../proxy_arp" 21:23 < caker> that should be it, i think 21:23 < Koro> okay will try now 21:23 < caker> and the arp command 21:23 < mistik1> hmmm 21:23 < Koro> arp -Ds 129.78.24.102 eth0 pub 21:23 < Koro> (for the first uml) 21:23 < caker> y 21:23 < Koro> and .103 for the sec 21:23 < Koro> cool 21:23 < mistik1> my script could help here I think 21:24 * caker points to mistik1 21:24 -!- ansoft [~anelson@207.179.107.18] has quit [Quit: Client Exiting] 21:24 < mistik1> this I think would be done with the -e options 21:24 < caker> heh 21:25 < mistik1> no proxy arp needed at all 21:25 < mistik1> lets see 21:26 < mistik1> Koro: still with us? 21:26 < Koro> yep 21:26 < mistik1> ok 21:26 < Koro> now i can't ping the other umls, but i can ping the host kernel 21:26 < mistik1> lemme make sure I understand what you're after 21:26 < caker> you need to add a route inside the umls 21:26 < Koro> :) 21:27 < caker> Koro: bridge is a better solution 21:27 < mistik1> You have eth0 that is WAN 21:27 < Koro> have a host kernel on .101, want to add 2 UML's on .102 and .103 21:27 < Koro> all are public IP's 21:27 < Koro> 129.78.24.x 21:27 < mistik1> much better 21:27 < mistik1> ok 21:27 < mistik1> wget http://music.geeksinthehood.net:81/~mistik1/ifup-bridge-0.2.sh 21:27 < mistik1> get that 21:27 < mistik1> chmod +x it 21:27 < mistik1> kill your taps 21:28 < mistik1> stop the proxy arping 21:28 < mistik1> and let me know when the host is setup normally 21:28 < caker> **hope he has bridge support** 21:28 < mistik1> no taps, default route 21:28 < mistik1> set 21:28 < mistik1> caker: easy enough to find out 21:28 < mistik1> Koro: modprobe bridge 21:28 < Koro> k 21:29 < Koro> got the script 21:29 < mistik1> modprobe bridge, just so we know your host kernel has support 21:29 < Koro> bridge in 21:29 < mistik1> cool 21:29 < caker> that would only show it as a module, right? he also needs brctl tools, too? 21:29 < Koro> lsmod shows bridge 21:29 < mistik1> yes inded 21:29 < mistik1> ok 21:30 < mistik1> which brctl 21:30 < Koro> nope 21:30 < mistik1> distro? 21:30 < Koro> debian testing 21:30 < caker> muhaha 21:30 < mistik1> apt-get install bridge-utils 21:30 < Koro> kl 21:31 < Koro> k done 21:31 < Koro> got it 21:31 < mistik1> I also assume you have the uml_utilities installed 21:31 < Koro> yep 21:31 < mistik1> which tunctl 21:31 < Koro> tunctl etc 21:31 < mistik1> ok good 21:31 < Koro> that's how i set up tap1, tap2 21:31 < mistik1> print 21:31 < mistik1> route -n 21:31 < mistik1> route -n |grep ^ 21:32 < mistik1> route -n |grep ^0 21:32 < Koro> Kernel IP routing table 21:32 < Koro> Destination Gateway Genmask Flags Metric Ref Use Iface 21:32 < Koro> 129.78.24.103 0.0.0.0 255.255.255.255 UH 0 0 0 tap2 21:32 < Koro> 129.78.24.102 0.0.0.0 255.255.255.255 UH 0 0 0 tap1 21:32 < Koro> 192.168.0.102 0.0.0.0 255.255.255.255 UH 0 0 0 tap2 21:32 < Koro> 192.168.0.101 0.0.0.0 255.255.255.255 UH 0 0 0 tap1 21:32 < Koro> 129.78.24.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0 21:32 < Koro> 129.78.0.0 0.0.0.0 255.255.0.0 U 0 0 0 tap1 21:32 < Koro> 129.78.0.0 0.0.0.0 255.255.0.0 U 0 0 0 tap2 21:32 < Koro> 0.0.0.0 129.78.24.254 0.0.0.0 UG 0 0 0 eth0 21:32 < mistik1> caker: walk him through killing those other routes please 21:32 < caker> route del 192.168.0.101 (repeat 101-103) 21:33 < caker> ifconfig tapX down (all of em) 21:33 < caker> tunctl -d tapDevice (all of em) 21:33 < Koro> k 21:33 < caker> route del 129.78.0.0 x2 21:33 < Koro> Kernel IP routing table 21:33 < Koro> Destination Gateway Genmask Flags Metric Ref Use Iface 21:33 < Koro> 129.78.24.0 * 255.255.255.0 U 0 0 0 eth0 21:33 < Koro> default gateway24.sedal 0.0.0.0 UG 0 0 0 eth0 21:33 < Koro> much better 21:34 < caker> no not that last route sorry 21:34 < Koro> sorry which? 21:34 < caker> you're good 21:34 < caker> mistik1? 21:34 < Koro> sorry which route? 21:34 < caker> Koro: you're all set 21:35 < mistik1> okhe's ready 21:35 < Koro> hang on i didn't del the tap devices 21:35 < mistik1> do you run a firewall on that box? 21:35 < Koro> hmm yeah i think some rules 21:35 < Koro> iptables or chains 21:35 < Koro> not sure 21:36 < mistik1> If you do you will need to change the external device when we are done 21:36 < Koro> k taps fixed 21:36 < Koro> okay 21:36 < mistik1> you should do so now be do NOT restart the firewall scripts untill you run my script 21:36 < mistik1> the external device will become br0 21:36 < Koro> do i need to edit ur script? 21:36 < mistik1> no 21:37 < Koro> okay 21:37 < Koro> just run now? 21:37 < mistik1> no 21:37 < mistik1> this is how you will run the script... 21:37 < Koro> see i don't have an eth1 btw 21:37 < mistik1> that's ok 21:37 < Koro> i have an eth0 on the host box for external i/f and that's it 21:37 < Koro> k 21:38 < Koro> : 21:38 < Koro> ) 21:38 < mistik1> ./ifup-bridge-0.2.sh -e eth0 21:38 < mistik1> do not run it untill you understand what i'm saying 21:39 < mistik1> what user does UML run as? 21:39 < Koro> dsfds 21:40 < mistik1> ok 21:40 < Koro> peterson@matrix:~$ ss ./ifup-bridge-0.2.sh -e peterson eth0 21:40 < Koro> Set 'peterson-18026' persistent and owned by uid 1000 21:40 < Koro> Example UML startup line 21:40 < Koro> linux ubd0=root_fs eth0=tuntap,peterson-18026 21:40 < mistik1> ./ifup-bridge-0.2.sh -e dsfds eth0 21:40 < Koro> k? 21:40 < mistik1> yes 21:40 < Koro> shit it hung for about 5 secs 21:40 < mistik1> yes 21:40 < Koro> i am remote and i admit, i swore lol 21:40 < mistik1> It took the network down 21:40 < mistik1> then put it back up 21:41 * Koro breathes a sigh of relief 21:41 < mistik1> it does work pretty well ;-) 21:41 < mistik1> does uml run as peterson? 21:41 < Koro> yes i start them manually uysing "linux ..." 21:41 < mistik1> ok 21:42 < mistik1> you see that sample command line 21:42 < Koro> so now i just do a eth=tuntap,peterson-18026 21:42 < mistik1> eth0 yes 21:42 < mistik1> for that user 21:42 < Koro> and set uml eth0's using ifconfig ? 21:42 < mistik1> yup 21:42 < Koro> eth0 sorry 21:42 < Koro> k hang 1 21:42 < mistik1> or just setup the network settings on the uml distro 21:43 < Koro> same prob 21:43 < mistik1> caker: I really must get off my arse and write a README for this thing 21:43 < Koro> can only ping the host kernel from the uml 21:43 < caker> heh 21:43 < mistik1> hang on friend 21:43 < Koro> can't ping the uml IP itself from inside the uml 21:43 < Koro> :) 21:44 < mistik1> uml# route -n 21:44 < Koro> (none):~# route -n 21:44 < Koro> Kernel IP routing table 21:44 < Koro> Destination Gateway Genmask Flags Metric Ref Use Iface 21:44 < Koro> 129.78.24.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0 21:44 < Koro> 0.0.0.0 129.78.24.254 0.0.0.0 UG 0 0 0 eth0 21:44 < Koro> (none):~# 21:44 < Koro> ahh maybe it should have the HK box as default g/w? 21:45 < mistik1> that looks good 21:45 < caker> route add default gw 129.78.24.254 21:45 < Koro> .101 21:45 < caker> inside the uml 21:45 < Koro> .254 is the upstream g/w 21:45 < Koro> .101 is the HK 21:45 < mistik1> that corect 21:45 < Koro> k 21:45 < mistik1> You should be using your ISPs gateway 21:45 < Koro> (none):~# route add default gw 129.78.24.254 21:45 < Koro> SIOCADDRT: File exists 21:45 < mistik1> or whatever that is 21:45 < Koro> aldready set 21:45 < mistik1> its already there 21:46 < Koro> (none):~# ping 129.78.24.101 21:46 < Koro> PING 129.78.24.101 (129.78.24.101): 56 data bytes 21:46 < Koro> 64 bytes from 129.78.24.101: icmp_seq=0 ttl=64 time=0.2 ms 21:46 < Koro> can ping _host_ kernel 21:46 < Koro> (none):~# ping 129.78.24.102 21:46 < Koro> PING 129.78.24.102 (129.78.24.102): 56 data bytes 21:46 < Koro> ping: sendto: Invalid argument 21:46 < Koro> ping: wrote 129.78.24.102 64 chars, ret=-1 21:46 < Koro> cannot ping uml (self) 21:46 < Koro> :( 21:46 < mistik1> odd 21:46 < caker> print hosts's route -n please 21:46 < mistik1> yes 21:46 < Koro> u mean host kernel? 21:46 < caker> yup 21:47 < Koro> peterson@matrix:~$ ss route -n 21:47 < Koro> Kernel IP routing table 21:47 < Koro> Destination Gateway Genmask Flags Metric Ref Use Iface 21:47 < Koro> 129.78.24.0 0.0.0.0 255.255.255.0 U 0 0 0 br0 21:47 < Koro> 0.0.0.0 129.78.24.254 0.0.0.0 UG 0 0 0 br0 21:47 < Koro> peterson@matrix:~$ 21:47 < mistik1> looks fine 21:48 < caker> print matrix:$ ifconfig peterson-18026 21:49 < Koro> peterson- Link encap:Ethernet HWaddr 00:FF:E3:D8:DF:0C 21:49 < Koro> UP BROADCAST RUNNING PROMISC MULTICAST MTU:1500 Metric:1 21:49 < Koro> RX packets:18 errors:0 dropped:0 overruns:0 frame:0 21:49 < Koro> TX packets:2656 errors:0 dropped:1658 overruns:0 carrier:0 21:49 < Koro> collisions:0 txqueuelen:100 21:49 < Koro> RX bytes:1036 (1.0 KiB) TX bytes:434948 (424.7 KiB)ahh 21:49 < Koro> ahh 21:49 < Koro> no IP 21:49 < Koro> ? 21:49 < mistik1> no IP needed 21:49 < caker> do the same for eth0 please 21:49 < Koro> br0 Link encap:Ethernet HWaddr 00:10:5A:5B:D2:53 21:49 < Koro> inet addr:129.78.24.101 Bcast:129.78.255.255 Mask:255.255.255.0 21:49 < Koro> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 21:49 < Koro> RX packets:8278 errors:0 dropped:0 overruns:0 frame:0 21:49 < Koro> TX packets:4101 errors:0 dropped:0 overruns:0 carrier:0 21:49 < Koro> collisions:0 txqueuelen:0 21:49 < Koro> RX bytes:997084 (973.7 KiB) TX bytes:469910 (458.8 KiB) 21:49 < caker> ahh arp problem probably 21:49 < Koro> eth0 Link encap:Ethernet HWaddr 00:10:5A:5B:D2:53 21:49 < Koro> UP BROADCAST RUNNING PROMISC MULTICAST MTU:1500 Metric:1 21:49 < Koro> RX packets:16924011 errors:0 dropped:0 overruns:0 frame:0 21:49 < Koro> TX packets:203175 errors:0 dropped:0 overruns:0 carrier:0 21:49 < Koro> collisions:0 txqueuelen:100 21:49 < Koro> RX bytes:2696378406 (2.5 GiB) TX bytes:23366516 (22.2 MiB) 21:49 < Koro> Interrupt:16 Base address:0xc000 21:49 < Koro> k 21:49 < Koro> haven't done any aprs 21:49 < Koro> arps even :) 21:49 < Koro> we deleted them 21:49 < Koro> ? 21:50 < caker> not yet 21:50 * mistik1 wonders what arp -a in uml reports 21:50 < caker> arp -i eth0 pub 21:50 < Koro> (none):~# arp -a 21:50 < Koro> ? (129.78.24.103) at on eth0 21:50 < Koro> ? (129.78.24.101) at 00:10:5A:5B:D2:53 [ether] on eth0 21:50 < Koro> (none):~# 21:51 < caker> do 103 21:51 < Koro> sorry how? 21:51 < caker> arp -i eth0 pub 21:51 < Koro> in the HK? 21:51 < caker> y 21:51 < caker> or reboot and start fresh 21:51 < caker> or restart networking i guess 21:51 < Koro> arp -i eth0 129.78.24.102/103 pub 21:51 < Koro> 102 or 103 21:52 < caker> both/all 21:52 < Koro> they are my 2 uml's 21:52 < Koro> k 21:52 < Koro> hang 1 21:52 < Koro> peterson@matrix:~$ arp -i eth0 129.78.24.102 pub 21:52 < Koro> -bash: arp: command not found 21:52 < Koro> peterson@matrix:~$ ss arp -i eth0 129.78.24.102 pub 21:52 < Koro> 129.78.24.102 (129.78.24.102) -- no entry 21:52 < Koro> peterson@matrix:~$ ss arp -i eth0 129.78.24.103 pub 21:52 < mistik1> caker: why would the table not update by itself or with a ping? 21:52 < Koro> 129.78.24.103 (129.78.24.103) -- no entry 21:52 < Koro> peterson@matrix:~$ 21:53 < Koro> (none):~# arp -a 21:53 < Koro> ? (129.78.24.101) at 00:10:5A:5B:D2:53 [ether] on eth0 21:53 < Koro> (none):~# ping 129.78.24.102 21:53 < Koro> PING 129.78.24.102 (129.78.24.102): 56 data bytes 21:53 < Koro> ping: sendto: Invalid argument 21:53 < Koro> ping: wrote 129.78.24.102 64 chars, ret=-1 21:53 < Koro> --- 129.78.24.102 ping statistics --- 21:53 < Koro> 1 packets transmitted, 0 packets received, 100% packet loss 21:53 < Koro> (none):~# ping 129.78.24.103 21:53 < Koro> PING 129.78.24.103 (129.78.24.103): 56 data bytes 21:53 < Koro> --- 129.78.24.103 ping statistics --- 21:53 < Koro> 1 packets transmitted, 0 packets received, 100% packet loss 21:53 < Koro> (none):~# 21:53 < Koro> same prob 21:53 < Koro> but arp -a changed a bit (on the uml) 21:54 < caker> ok HK, chmod 666 /dev/net/tun ? 21:54 < caker> (long shot) 21:54 < Koro> i did that b4 21:54 < Koro> did again 21:54 < mistik1> this is really wierd 21:54 < caker> ping the gateway IP from inside the uml 21:55 < Koro> (none):~# ping 129.78.24.254 21:55 < Koro> PING 129.78.24.254 (129.78.24.254): 56 data bytes 21:55 < Koro> 64 bytes from 129.78.24.254: icmp_seq=0 ttl=255 time=4.7 ms 21:55 < Koro> yay! 21:55 < Koro> can ping: .101 (HK), .254 (g/w) 21:55 < Koro> cannot ping seld 21:55 < Koro> self 21:55 < Koro> .102 21:56 < mistik1> incredible 21:56 < Koro> ahh hang on 21:56 < mistik1> UML# irconfig eth0 21:56 < Koro> my linux line 21:56 < Koro> when I start the 2 instance 21:56 < Koro> ubd=/home/peterson/deb1 21:56 < Koro> but it should read /home/peterson/uml/deb1 21:57 < Koro> so i think i have 2 instances on the same root_fs 21:57 < Koro> could this be a prob>? 21:57 < Koro> lemme kill and restart 1 only 21:57 < mistik1> Koro: FYI you should make sure you use a different tap for each uml 21:58 < Koro> no 21:58 < Koro> yes i did 21:58 < Koro> tap1 for deb1, tap2 for deb2 21:58 < mistik1> to add a second tap just ./ifup-bridge-0.2.sh -a 21:58 < Koro> oh hang on 21:58 < mistik1> tap1? 21:58 < mistik1> You should have no such things anymore 21:58 < Koro> inux ubd0=/home/peterson/uml/deb1 con=pty con0=fd:0,fd:1 eth0=tuntap,peterson-18026 21:58 < Koro> yeah sorry confused 21:58 < Koro> that's my startup line 21:58 < mistik1> ok 21:59 < mistik1> I would stop both uml 21:59 < Koro> have done 21:59 < mistik1> and start one over 21:59 < Koro> just running 1 now (restarted) 21:59 < Koro> yep 21:59 < Koro> (none):~# ifconfig 21:59 < Koro> eth0 Link encap:Ethernet HWaddr FE:FD:81:4E:18:66 21:59 < Koro> inet addr:129.78.24.102 Bcast:129.78.24.255 Mask:255.255.255.0 21:59 < Koro> that's the uml 21:59 < mistik1> ok 21:59 < Koro> i can now ping everything except _myself_ from within the uml 22:00 < Koro> and i can ping the uml from outside 22:00 < Koro> so i'm _almost_ sweet ;) 22:00 < caker> check your iptables 22:00 < caker> but still, the error message you get is strange 22:01 < mistik1> caker: I was just going there 22:01 < caker> something isn't configured right 22:01 < mistik1> its the only thing I could think at this point 22:01 < caker> easier to reboot and try fresh 22:01 < mistik1> the IP/ARP portion is flawless now 22:01 < mistik1> wont solve it if its iptables though 22:01 < caker> i've had that problem before 22:01 < caker> mistik1: right 22:01 < Koro> okay 22:02 < Koro> i will reboot soon 22:02 < Koro> thanks guys for _all_ your help 22:02 < Koro> mostly wokring now 22:02 < Koro> ;) 22:02 < mistik1> Koro: 22:02 < caker> Koro: was fun 22:02 * caker highfives mistik1 22:02 < Koro> yeah, was fn ;) 22:02 * Koro hi-5's all round 22:02 < mistik1> just run the script with no args for usage ok 22:02 < mistik1> and do remember what you did 22:02 < Koro> heh "ping neo.sedal.usyd.edu.au" to hit the UML 22:02 < Koro> matrix.sedal.usyd.edu.au to hit the HK 22:02 < Koro> ;) 22:03 < mistik1> works fine from here 22:03 < caker> PING neo.sedal.usyd.edu.au (129.78.24.102) from 64.35.99.205 : 56(84) bytes of data. 22:03 < caker> 64 bytes from 129.78.24.102: icmp_seq=0 ttl=43 time=240.0 ms 22:03 < Koro> i hope u like the analogy ;) 22:03 < Koro> excellent 22:03 < Koro> gtg 22:03 < mistik1> later 22:03 < Koro> i have a mtg to attend 22:03 < Koro> l8rz guys 22:03 < mistik1> thanks caker 22:03 < Koro> thanks once again 22:03 < caker> np 22:03 < mistik1> np 22:03 < Koro> cyaz 22:03 -!- Koro [peterson@matrix.sedal.usyd.edu.au] has left #uml [] 22:03 < mistik1> that was indeed fun ;) 22:04 < caker> i'm starting to scare myself with the little networking knowledge i've picked up 22:04 < mistik1> he could have told me he was working remotely before running the script tho 22:04 < mistik1> hahaha 22:05 < caker> I was going to ask, but I figuered someone would know to ask himself such things when messing with some random guy on IRC's networking script 22:05 < mistik1> indeed 22:05 < caker> download this rootkit 22:05 < caker> i mean 22:05 < caker> network helper 22:05 < mistik1> lol 22:06 < mistik1> good thing i'm a real trust worthy charactor 22:06 < mistik1> :) 22:06 < caker> if you say so :-p 22:09 < mistik1> what I would like to do now is have the script create a subscript to store the exact settings for reboot 22:09 < caker> ./iWantToRunUML 22:09 < caker> What do you want to do: 22:09 < caker> 1. download a distro 22:09 < caker> 2. .... 22:10 < mistik1> no 22:10 < caker> uml needs a little helper app, i think 22:10 < mistik1> think so? 22:10 < caker> sets up all the default stuff 22:10 < caker> well, there are a few recurring questions 22:10 < mistik1> explain what this would do 22:10 < caker> like 22:10 < caker> console output 22:11 < caker> my filesystem won't bood (ubd issues) 22:11 < caker> big one: networking 22:11 < caker> where do i get X filesystem 22:11 < caker> etc 22:11 < mistik1> hmmm 22:11 < caker> I was thinking about a little command-line tool 22:11 < caker> like a commander type of thing 22:11 < caker> lets you setup configurations 22:11 < mistik1> like ./runuml deb3 22:11 < caker> save that config to a "profile" like you were saying 22:11 < caker> a lot of this I have for Linode, but it's all DB driven 22:12 < caker> DB and website obviously 22:12 < caker> "Where do you want console output to go" 22:12 < caker> and you have multiple-guess 22:12 < caker> when explainations 22:12 < caker> s/when/with 22:13 < mistik1> so basically a umlbuilder on steriods 22:13 < mistik1> non GUI that is 22:13 < caker> umlbuilder, umlconfig, umlrun all rolled into one 22:13 < mistik1> right 22:14 < caker> instead of umlbuilder, though - i've gone with pre-made filesystems 22:14 < mistik1> that would make a most interesting project 22:14 < mistik1> for something like this a premade fs would be ideal 22:14 < caker> I'm going to make all mine available soon 22:15 < mistik1> most distros have tools to extend them 22:15 < caker> they're pretty complete 22:15 < caker> yeah, also included in the commander app is a filesystem resize tool 22:15 < mistik1> mdk = urpmi, deb = apt, redaht = apt 22:15 < mistik1> yup 22:16 < mistik1> I'm not much of a coder but i'm always game 22:17 < mistik1> We must speak on this more 22:17 < mistik1> maybe get david to setup a forum page on uml.org to get an idea of interest 22:19 < caker> right - not sure I can provide much any more 22:19 < caker> but we'll see :) 22:19 < mistik1> :) 22:20 < mistik1> awe 22:20 < mistik1> David Brinkly has died 22:42 < david> mistik1: There are mailing lists - A forum would be pretty much useless, unless people suddenly hate e-mail 22:43 < mistik1> good point 22:43 < mistik1> I've been meaning to jump on the list 22:45 < david> I could setup phpbb, but no one has requested it 22:47 < caker> david: have you thought through much of the drbd stuff? .. I'd have to daisy-chain servers 22:48 < caker> and hope that A and B don't go down at the same time 22:48 < david> caker: I've no idea if daisy-chaining works 22:48 < david> caker: basically for each UML box I'd have a hot-space 22:48 < caker> I could see how this would be useful for two Storage Servers to mirror one-another, and then NFS mount everything to the hosts 22:49 < david> caker: the spare would not be 'active' unless the primary died 22:49 < david> caker: right, that's what I do with it atm 22:49 < caker> on the same host machine, you mean? 22:49 < david> caker: no 22:49 < david> I have one host machine running UMLs 22:49 < david> I have a hot space 22:49 < david> gahghas 22:49 < david> spare 22:49 < caker> spare machine 22:49 < david> if the host dies, all the UMLs come up on the spare 22:49 < david> caker: right 22:49 < caker> ok 22:49 < caker> wasteful 22:49 < david> caker: right, but it's unlikely that both boxes would die at the same time 22:50 < david> caker: and in theory, you could run some services/UMLs on the spare then just pull the rest over if the primary dies 22:50 < caker> right 22:51 < caker> A is mirrored on Ba B is mirrored on Cb, C is mirrored on Ac .. you'd have to leave room for the overhead of taking the down'ed machine's load across the other two servers 22:51 < caker> ack, couldn't do that 22:51 < caker> if A went down, B is the only choice 22:52 < david> it's a pain in the ass 22:52 < caker> sounds like it 22:53 < mistik1> maybe you guys can assist with an odity i've noticed 22:53 < david> mistik1: Jeff? 22:53 < mistik1> caker: did you notice how ifconfig did not print the full name of the device for Korn 22:54 < mistik1> the name was wuite long 22:54 < mistik1> quite 22:54 < mistik1> david: funny :) 22:54 < caker> you have to cat /proc/net/dev to get the full names 22:54 < caker> at least that is one way 22:54 < mistik1> yes I know 22:55 < mistik1> it just seemed odd to me that, in slackware and gentoo ifconfig list the full name as long as it is 22:55 < mistik1> rhat, mdk, debian ifconfig dont 22:55 < david> mistik1: different net-tools versions? 22:55 < mistik1> this is indeed a posability 22:56 < mistik1> It dont bother me really really as I know how to get it, was just curious 22:56 < mistik1> time for a nap, I cant type anymore --- Log closed Fri Jun 13 00:00:00 2003