--- Day changed --- Log opened Sat Jun 14 00:00:01 2003 01:06 < caker> Figured out the arp problem... ebtables + netfilter is evil 01:11 < david> caker: something was filtering? 01:12 < caker> i guess so ... I didn't need ebtables anyhow, so... bad caker 01:12 < david> heh 01:12 < caker> i should have known, since now I can see dhcp traffic on both hosts, from each other 01:16 < caker> has anyone ever seen "iptables: Resource temporarily unavailable" ? 01:16 < caker> some of my rules fail, seems totally random 01:17 < Lathiat> Heyas 01:17 < caker> Lathiat: hello 01:17 < Lathiat> david: mm ill do it tomorow, its late... *2 minutes later* :P - your as obsessive as i am*giggle* 03:00 < caker> hmm kernel bug in 2.4.21 .. 03:03 < Lathiat> is that one of those 03:03 < Lathiat> "dont run it" bugs? 03:03 < caker> i can make it crash by bringing up some taps and adding them to the bridge .. so I think there's an issue somewhere :( 03:05 < Lathiat> hrm 03:05 < Lathiat> ill try it on my box 03:10 < caker> seems like a race, because I have to add the tunS to the bridge at the same time to reproduce 03:31 < Lathiat> yay race conditions 03:31 < caker> yup 03:31 < caker> posting to lkml now 04:17 < caker> fixed it by not using Intel's e100 driver - using eepro100 now and it works 04:17 < Lathiat> heh 04:17 < Lathiat> :P 04:20 < pirlouit> i've had a problem with the e100 driver on a 2.4.20 series kernel 04:20 < pirlouit> s/problem/problems/ 04:21 < caker> the weird thing is my eth0 and eth1 reverse when I switch drivers 04:22 < caker> but, i'm running the e100 driver on another host and it seems fine 04:22 * caker recompiles to not take any chances 04:23 -!- gump [~gump@81.5.136.90] has quit [Read error: Connection reset by peer] 04:24 < pirlouit> they reverse ? does that mean that they don't use the PCI IDs to determine which one is what; weird 04:24 < caker> who knows 04:26 < caker> crap, with eepro my bridge iptable rules don't work 04:51 < caker> david: scratch that, i was wrong - MAC problem still exists.. time for bed, enough damage tonight 05:10 -!- DnsInfector [~DnsInfect@81.53.249.210] has joined #uml 05:23 -!- DnsInfector [~DnsInfect@81.53.249.210] has quit [Ping timeout: 488 seconds] 05:40 -!- kara [~kara@64.240.156.226] has quit [Ping timeout: 488 seconds] 05:40 -!- silug [~steve@osiris.silug.org] has quit [Ping timeout: 492 seconds] 05:41 -!- silug [~steve@osiris.silug.org] has joined #uml 05:41 -!- kara [~kara@varstation.lanscape.net] has joined #uml 06:38 -!- DnsInfector [~DnsInfect@ARennes-204-1-3-20.w193-252.abo.wanadoo.fr] has joined #uml 06:46 -!- DnsInfector_ [~DnsInfect@ARennes-204-1-1-187.w193-251.abo.wanadoo.fr] has joined #uml 06:51 -!- avoozzzl is now known as avoozl 06:53 -!- DnsInfector [~DnsInfect@ARennes-204-1-3-20.w193-252.abo.wanadoo.fr] has quit [Ping timeout: 496 seconds] 06:53 -!- DnsInfector [~DnsInfect@80.13.112.5] has joined #uml 06:57 -!- DnsInfector_ [~DnsInfect@ARennes-204-1-1-187.w193-251.abo.wanadoo.fr] has quit [Read error: Connection reset by peer] 07:05 -!- DnsInfector [~DnsInfect@80.13.112.5] has quit [Ping timeout: 488 seconds] 07:55 -!- DnsInfector [~DnsInfect@ARennes-204-1-12-199.w81-53.abo.wanadoo.fr] has joined #uml 08:49 -!- DnsInfector [~DnsInfect@ARennes-204-1-12-199.w81-53.abo.wanadoo.fr] has quit [Ping timeout: 492 seconds] 09:23 -!- gump [gump@81.5.136.90] has joined #uml 09:55 -!- gump [gump@81.5.136.90] has quit [Read error: Connection reset by peer] 12:12 -!- DnsInfector [~DnsInfect@ARennes-204-1-1-233.w193-251.abo.wanadoo.fr] has joined #uml 12:12 < DnsInfector> hi 12:12 < DnsInfector> too bad the precompiled kernel on http://user-mode-linux.sf.net/ doesn't support skas mode 12:13 < Lathiat> You need a host kernel modification too 12:13 < Lathiat> you got that? 12:14 < DnsInfector> yes 12:14 < DnsInfector> i've finally upgraded my old MDK kernel to a custom one 12:14 < DnsInfector> with skas patch 12:17 < green> so build custom uml kernel too, or get prebuilt one with skas from usermodelinux.org 12:17 < DnsInfector> And running uml doesn't show the requested "Checking for the skas3 patch in the host...found 12:17 < DnsInfector> Checking for /proc/mm...found" 12:18 < DnsInfector> i wanted to grabed a prcompiled kernel to save time, i'll have to recompile 12:19 * green suspects that precompiled kernels from +http://user-mode-linux.sf.net/ are so old that they are predate skas support at all ;) 12:19 < DnsInfector> :) 12:19 < DnsInfector> for sure 12:19 < DnsInfector> root fs are also quite old 12:19 < Lathiat> bah make your own 12:19 < Lathiat> easy for debian :P 12:21 < DnsInfector> easy but long 12:22 < Lathiat> takes me about 15 minutes 12:22 < Lathiat> tho, i do have a mirror on 10mbit at my disposal ;) 12:23 < DnsInfector> i imagine it helps 12:52 -!- Lathiat [lathiat@seven.sixlabs.org] has quit [Quit: Lost terminal] 12:52 -!- Lathiat [lathiat@seven.sixlabs.org] has joined #uml 12:52 < DnsInfector> re 12:53 < DnsInfector> hum 12:53 < DnsInfector> by default the rh image has many service running 12:53 < DnsInfector> it's very lonnng to start 12:54 -!- Lathiat [lathiat@seven.sixlabs.org] has quit [Client Quit] 12:55 -!- Lathiat [lathiat@130.95.13.25] has joined #uml 13:24 < avoozl> i just use rootstrap 13:25 < DnsInfector> in fact i hadn't reinstall uml_utilities 13:25 < DnsInfector> it starts much faster now 13:47 -!- gump [gump@81.5.136.90] has joined #uml 14:05 * gump is away -( out )- at 07:05p -( P:On / L:On )- 14:43 -!- DnsInfector [~DnsInfect@ARennes-204-1-1-233.w193-251.abo.wanadoo.fr] has quit [Quit: Client exiting] 14:52 -!- Gui [~Guilherme@200-103-237-103.ctame7041.dsl.brasiltelecom.net.br] has joined #uml 14:52 -!- DnsInfector [~DnsInfect@193.251.66.233] has joined #uml 14:52 -!- Gui [~Guilherme@200-103-237-103.ctame7041.dsl.brasiltelecom.net.br] has left #uml [] 14:52 -!- Gui [~Guilherme@200-103-237-103.ctame7041.dsl.brasiltelecom.net.br] has joined #uml 14:52 < DnsInfector> re 14:52 < DnsInfector> hum i had troubles with tuntap 14:53 -!- Gui [~Guilherme@200-103-237-103.ctame7041.dsl.brasiltelecom.net.br] has left #uml [] 14:53 < DnsInfector> it looks like there's a deadlock in the closing of tuntap devices 14:54 < DnsInfector> the kernel doesn't want to release the device until uml_net has no references on it 14:54 < DnsInfector> and uml_net don't wanna die until the device is released 14:54 < Lathiat> ohh nice 14:55 < DnsInfector> my conf is everything 2.4.21 14:55 < DnsInfector> and host has skas patch 14:58 * Lathiat waves 14:58 < Lathiat> night 14:58 < BB> night Lathiat 17:02 -!- alt [~alt@h24-85-231-239.wp.shawcable.net] has joined #uml 17:02 < alt> hi all 17:02 < alt> :) 17:04 < alt> anyone here run UML on Debian Woody? 17:04 < desaster> yep 17:04 < alt> desaster: have you come across a problem where you can't get any login screens for the UML session? It's woody as well. 17:05 < desaster> hmm, you probably don't have the proper tty entries in the /etc/inittab of your uml 17:05 < desaster> boot into it in single mode and fix /etc/inittab to point to tty1, tty2, ... 17:05 < desaster> instead of vc/0 or whatever the devfs uses 17:05 < desaster> or... you could run the uml with devfs, but i don't know much about it 17:07 < alt> hmm.. let me try that 17:09 < alt> okay.. that's different 17:09 < alt> it was ttys/0 which did work under mandrake 17:11 < alt> thx desaster 17:27 -!- DnsInfector_ [~DnsInfect@81.49.44.184] has joined #uml 17:35 -!- DnsInfector [~DnsInfect@193.251.66.233] has quit [Ping timeout: 488 seconds] 17:35 -!- _slh__ [~chatzilla@68.19.227.139] has quit [Read error: Connection reset by peer] 17:37 -!- rus [~admin@217.126.239.37] has left #uml [] 18:03 -!- avoozl is now known as avoozzzl 18:45 -!- Pahan [pahan@syr-24-59-98-110.twcny.rr.com] has quit [Quit: leaving] 18:45 -!- _slh__ [~chatzilla@68.154.228.11] has joined #uml 18:49 -!- Pahan [pahan@24.59.98.110] has joined #uml 19:24 < Pahan> Argh, I am getting bit load averages even with skas :( 19:24 < Pahan> >2 19:25 < david> Pahan: sure it's using skas? 19:26 < Pahan> model name : UML 19:26 < Pahan> mode : skas 19:26 < Pahan> I am installing some packages in two UMLs at once. 19:26 < Pahan> Host is Athlon Thunderbird 1.33@1.00 19:27 < Pahan> Err. Debian packages, that is. The thing is a CPU hog. 19:29 < david> Pahan: probably IO using CPU time, rather than much else 19:29 < david> Pahan: your /tmp is tmpfs, right? 19:30 < david> otherwise performance will go through the floor 19:30 < Pahan> No, my /tmp is on root filesystem, together with everything else. 19:30 < Pahan> Hrm. 19:30 < Pahan> How exactly does UML use host's /tmp/ 19:30 < Pahan> ? 19:32 < alt> http://user-mode-linux.sourceforge.net/UserModeLinux-HOWTO-7.html 19:32 < alt> sorry 19:32 < alt> wrong link 19:32 < alt> http://user-mode-linux.sourceforge.net/UserModeLinux-HOWTO-9.html 19:33 < Pahan> alt: That's grossly irrelevant to my question. 19:33 < alt> it is? 19:33 < Pahan> Yes. 19:33 < alt> f you want to access files on the host machine from inside UML, you can treat it as a separate machine and either nfs mount directories from the host or copy files into the virtual machine with scp or rcp. However, since UML is running on the the host, it can access those files just like any other process and make them available inside the virtual machine without needing to use the network. 19:34 < Pahan> You probably should read the conversation between me and david first. 19:34 < alt> you asked " How exactly does UML use host's /tmp/" 19:35 < alt> I'm giving you one possible solution 19:35 * Pahan shrugs. 19:35 < Pahan> You parsed that rather incorrectly. 19:36 < alt> okay, david asked you if your /tmp was on the host or on the VM 19:36 < alt> the way I read it, you indicated it was on the VM 19:36 < alt> david indicated that your performance would go down the crapper if that was the case 19:36 < Pahan> No, he asked whether my /tmp was using tmpfs or real disk space. 19:36 < Pahan> I asked him whether it matters. 19:36 < alt> okay.. I see it 19:36 < Pahan> I assumed he was talking about /tmp on host. 19:36 < alt> I apologize 19:39 < alt> so to correct my initial erroneous assuption... you are using /tmp as disk-based? 19:39 < Pahan> Yes. I always viewed tmpfs as an abstract exercise, not something I'd need to use. 19:39 < Pahan> I am going to try using it. 19:39 < alt> okay... yeah.. my bad 19:41 < alt> mount -t tmpfs none /tmp <--- dunno if that helps you 19:41 < alt> uml1:/# mount 19:41 < alt> /dev/ubd/0 on / type ext2 (rw) 19:41 < alt> proc on /proc type proc (rw) 19:41 < alt> none on /tmp type tmpfs (rw) 19:41 < alt> Filesystem Size Used Avail Use% Mounted on 19:41 < alt> /dev/ubd/0 485M 70M 390M 16% / 19:41 < alt> none 15M 0 14M 0% /tmp 19:41 < alt> that's what I get when I issue that mount command 19:41 < alt> don't know the syntax tho 19:42 < Pahan> "If you oversize your tmpfs instances the machine will deadlock since the OOM handler will not be able to free that memory" 19:42 < Pahan> Gross. 19:42 < alt> ugh 19:42 < Pahan> I hope that's not part of normal operation. 19:42 < alt> me too 19:42 < Pahan> That's from description to size= mount parameter. 19:43 < alt> I'm looking at the mount manpage.. obviously it doesn't talk about that problem 19:43 < Pahan> That's from kernel docs. 19:43 < Pahan> Documentation/filesystems/tmpfs.txt 19:43 < alt> size=nbytes 19:43 < alt> Override default size of the filesystem. The size is given in bytes, and 19:43 < alt> rounded down to entire pages. The default is half of the memory. 19:43 < alt> <--- that's what I'm looking at 19:51 < alt> bbl 19:57 -!- Pahan [pahan@24.59.98.110] has quit [Quit: leaving] 20:15 -!- _slh__ [~chatzilla@68.154.228.11] has quit [Ping timeout: 488 seconds] 20:24 -!- rawl [~rawl@200.75.95.21] has joined #uml 21:14 < alt> argh 21:15 < alt> I don't think I like UML on Debian :\ 21:16 < rawl> dont use debian ;) 21:20 < alt> heh 21:20 < alt> well, I like debian on the server, which is why I wanted to use it 21:21 < alt> I'm prototyping right now 21:21 < alt> I'm putting a server on the end of an E10 and want to set it up so the services run under UML 21:22 < alt> that way I can set up a few virtual servers that are independent of each other 21:22 < alt> I have the console working now 21:22 < alt> but I need to get networking working :\ 21:25 < alt> and then I;m done 21:51 -!- Pahan [pahan@syr-24-59-98-110.twcny.rr.com] has joined #uml 22:20 -!- rawl [~rawl@200.75.95.21] has quit [Quit: leaving] 22:36 < alt> note to self... check /usr/lib/uml/uml_net for SetUID and make sure self is in group uml-net when using Debian 22:57 < DnsInfector_> is there any ipv6 gurus here ? 22:57 < david> DnsInfector_: not really, but I'll try to be 22:57 < DnsInfector_> i got a problem with router advertisement 22:57 < DnsInfector_> here the arch: 22:58 < DnsInfector_> GW -eth0<--->eth0- djinn -tap0<--->eth0 -uml 22:58 < DnsInfector_> GW has the radvd daemon running 22:58 < DnsInfector_> so when djinn starts 22:59 < DnsInfector_> its eth0 NIC it receve an ipv6 address from GW 22:59 < DnsInfector_> the pb i have is that when uml starts djinn doesn't broadcast the "icmp router solicitation query" 23:00 < DnsInfector_> i was wondering if it's the normal behaviour or if i may tune soemthing 23:00 < DnsInfector_> GW is an USAGI host and 2 others are stock 2.4.21 23:01 < DnsInfector_> david: you're stuck ? 23:03 < david> DnsInfector_: your bridge is setup properly, right? 23:03 < david> does the UML's eth0 have a proper MAC address 23:03 < DnsInfector_> yes i can ping with ipv4 23:03 < DnsInfector_> yes 23:03 < david> odd 23:04 < DnsInfector_> looking with tcpdump 23:04 < DnsInfector_> i see the "router solicitation" on tap0, but nothing is sent on eth0(djinn) 23:05 < DnsInfector_> and when radvd sends router advertisement 23:05 < DnsInfector_> they aren't send on tap0 23:05 < DnsInfector_> i may have to USAGIfy djinn 23:13 < david> got no iea 23:13 < david> if they're not being forwarded by the bridge, I'd say there is a problem with the bridge 23:13 < DnsInfector_> i'm looking at tahi to see if this thing is tested 23:13 < DnsInfector_> this is not a bridge 23:14 < david> what is it? 23:14 < DnsInfector_> in ipv6 you can not call that a bridge 23:14 < david> eh? 23:14 < david> it's a layer 2 bridge 23:14 < david> it's nothing to do with ipv6 23:14 < david> it's below the ipv4/ipv6 layer 23:14 < DnsInfector_> yes 23:14 < DnsInfector_> you're right 23:15 < david> so it's a bridge with brctl? 23:15 < DnsInfector_> no 23:15 < david> what is it then? 23:15 < DnsInfector_> these packet are routed 23:15 < david> ok 23:15 < david> so, the IPv6 discovery packets are link-local 23:15 < david> so they're not going to be routed 23:15 < david> so you need to run radvd on the host running the UML 23:16 < DnsInfector_> not necessary 23:16 < david> or setup a bridge to forward the packets onto your LAN 23:16 < DnsInfector_> i think you're right 23:17 < DnsInfector_> but that not the idea i have of ipv6 address 23:17 < DnsInfector_> i think in theory you can have the same address space between 2 host separated by a router 23:18 < DnsInfector_> ipv6 address may be independant of underlying architecture 23:18 < DnsInfector_> but it 5AM here 23:18 < DnsInfector_> time to sleep 23:18 < DnsInfector_> good night 23:18 -!- DnsInfector_ [~DnsInfect@81.49.44.184] has quit [Quit: Client exiting] --- Log closed Sun Jun 15 00:00:00 2003