[05:22] hello..does anyone know how to configure nbdswapd.conf [05:22] i have a 30 clients that hav about 383mb a piece.. [05:22] they are all p3s [05:22] it's an lts.conf parameter [05:23] oh..no need to create the nbdwapd.conf file anymore? [05:23] in 8.04? [05:23] or is that appended to lts.conf now. [05:23] ?? [05:24] specifically i need to know what to set the value to...i.e...if each system has 383mb what to set it to. [05:24] SIZE=?? [07:58] ogra: btw, just reminding you to ask about my CMPC for iTalc testing :) (I also poked RichEd) [07:59] stgraber: i asked in a mail last night :) ... we meet with them in 2 min [07:59] no promise of rapid response ... but the topic has been raised [07:59] RichEd: cool, thanks [09:08] Hi I am trying to setup faxing for the clients, what is the best way to do this? GFax isn't working fully :(, efax-gtk dosen't put its phone number on the correct screen :( [09:14] you should probably file a bug [09:14] for efax-gtk [09:14] and also find out their mailing lists or something === gnomefre1k is now known as gnomefreak [10:33] I have installed the ltsp-server package, but the tftp server isn't working :( I tested with another computer with the 'tftp' program, what could be stoping it ? [12:47] Can anyone suggest a more appropriate channel to discuss serious LTSP issues with Hardy? [12:47] or did I manage to land in the correct place :) [12:49] astinus, here or #ltsp is fine [12:50] most ppl here are in both channels anyway :) [12:51] ogra: I'm getting a nasty /dev/nbd0 on /rofs issue [12:51] what kind ? [12:52] ogra: we just regenerated SSH keys for our LTSP server, it had been dist-upgraded to Hardy previously but we'd kept the LTSP NBD chroot as Gutsy for some decent enough reasons [12:52] thats likely to break [12:52] anyway... that went boom when the SSH key change happened, and squashfs-tools appears to have changed in Hardy such that it no longer builds viable stuff [12:52] *so* I rebuilt a brand new chroot using ltsp-build-blah which also fails [12:53] its being exported via NBD just fine, the DHCP servers and TFTP is fine too :( [12:53] so whats the exact error message [12:53] Mounting /dev/nbd0 on /rofs failed: No such device [12:53] and how exactly did you rebuild ? did you properly clean out /opt/ltsp = [12:54] are you sure your client is running Hardy's kernel ? (2.6.24-16) [12:54] " No such device" sounds like a kernel issue [12:54] yeah.. we have a somewhat hacked environment because we cannot run DHCP on our LTSP boxes, but /opt/ltsp on the servers got wiped, and its definitely running a Hardy kernel [12:54] can you try to : modprobe nbd [12:54] on a client [12:54] in busybox [12:54] sec [12:55] modprobe nbd works [12:55] that looks like you are not using an ltsp initramfs [12:56] (which forcefully lodas nbd) [12:56] astinus: try, nbdclient /dev/nbd0 [12:56] stgraber: I can't do that, busybox has no concept of nbdclient [12:57] nbd-client [12:57] (missing a dash) [12:57] hmm, right nbd-client :) [12:57] generates a large stack trace [12:57] with end error message being "File exists." [13:00] we use DHCP to bounce the request (via next-server) to our LTSP box, which runs TFTP and the pxelinux.cfg/default is set to: [13:00] DEFAULT vmlinuz ro initrd=initrd.img nosplash nbdport=2003 [13:00] inside that directory is all the usual stuff, vmlinuz + initrd + nbi [13:01] so the kernel seems to load fine... but something is causing /dev/nbd0 to not work :( [13:01] grep nbdroot /etc/inetd.conf ? [13:01] exists [13:01] and netstat shows them running fine [13:01] i want to see the lines [13:02] 2003 stream tcp nowait nobody /usr/sbin/tcpd /usr/sbin/nbdrootd /opt/ltsp/images/bytemark-hardy.img [13:02] bytemark-hardy.img is what you built as new chroot/image ? [13:02] correct. [13:02] ok [13:02] 289M -rwxr--r-- 1 root root 289M 2008-05-14 12:48 bytemark-hardy.img [13:03] des it work to run nbd-client as above from another machine [13:03] *does [13:03] sudo modprobe nbd && sudo nbd-client 2003 /dev/nbd0 [13:04] [alex #8] images > sudo nbd-client 89.16.168.152 2003 /dev/nbd0 [13:04] Negotiation: ..size = 295532KB [13:04] bs=1024, sz=295532 [13:04] hmm, looks fine as well [13:05] okay, after the DHCP ip config on boot [13:05] I do actually see the NBD negotiation happen and succeed [13:05] aha [13:05] but then I get the message above... [13:05] can't get it onto /rofs [13:06] which is why its so puzzling.... [13:06] and you surely rebuilt the chroot and image and didnt just run apt-get dist-upgrade in there ? [13:06] yeah [13:07] it was totally rebuilt, after my Gutsy one stopped working [13:07] it still smells like a wrongly set up initramfs to me [13:07] right.. so lets manually ensure thats correct, lemme chroot in, one sec [13:07] should /etc/modules in a newly created chroot contain nbd + other stuff? [13:08] no [13:08] then how is the LTSP stuff building the initramfs? [13:09] its set in /usr/share/initramfs-tools/conf.d/ using teh script /usr/share/initramfs-tools/scripts/ltsp-nbd during boot [13:09] ogra: when is today's meeting slot ? [13:10] RichEd, i whish the fridge could tell [13:10] supposed to be a late one [13:10] stgraber, did we do an early one last week ? [13:10] /usr/share/initramfs-tools/scripts << does not contain a ltsp-nbd script [13:10] right [13:10] astinus, in the chroot [13:10] ogra: kind of :) [13:10] correct. [13:10] cool stgraber is as good as a fridge [13:10] ogra: you forgot about it and we talked for a sec in #edubuntu instead [13:11] oh wait, I'm being a muppet [13:11] * astinus checks again [13:12] stgraber, right, i justs couldnt remember the time, my wednesdays are overcaset by total tiredness nowadays, having my first meeting at 7am and intel calls afterwards isnt really relaxing :/ [13:14] ogra: Alright, I think it was missing NBD, maybe... [13:14] ogra: I've built the initrd again with NBD explicitly enabled, and I'm now getting a nastier 'Negotiation: Error: Server closed connection.' message after the DHCP IP config stuff [13:15] anything in the server logs ? [13:17] where does NBD log by default? [13:17] finding it on the loghost via grep would take all day :( [13:17] syslog [13:18] Hrmph [13:18] its looking for /opt/ltsp/images/amd64.img [13:18] we're trying to supply it bytemark-hardy.img [13:19] what is "it" ? [13:20] nbdrootd or do you additionally have an nbd-server configured (dont do that) [13:20] okay, thats fixed [13:20] nbd_server[5836]: connect from 89.16.168.253, assigned file is /opt/ltsp/images/amd64.img [13:21] nbd_server[6199]: connect from 89.16.168.253, assigned file is /opt/ltsp/images/amd64.img [13:21] nbd_server[6199]: Size of exported file/device is 302624768 [13:21] so that's now working, but we're back to the /rofs error [13:21] right [13:21] are your clients amd64 machines ? [13:21] the chroot is i386 [13:21] the fact its looking for amd64.img is an anomaly [13:21] the clients are a mixture [13:21] welll, you apparently boot amd6.img [13:22] which indeed cant work at all [13:22] where is that configured though? [13:22] inetd.conf [13:22] my fix was simple, hardlink amd64.img -> bytemark-hardy.img [13:22] depending on eh port your kernel tries to connect to [13:22] 2000 stream tcp nowait nobody /usr/sbin/tcpd /usr/sbin/nbdrootd /opt/ltsp/images/bytemark-hardy.img [13:22] well, your pxe config showed 2003 before for that image [13:22] I just altered that. [13:22] its expected. [13:23] nbdport=2000 is defined in the pxelinux.cfg file [13:23] it *is* mounting the image, as evidenced by nbd_server logs + the client showing the right sorta strings... [13:23] its just not able to do stuff *after* that for some reason [13:23] it logs the connect attempt ... [13:24] *attempt* :) [13:24] hmm [13:24] that server is amd64 ? [13:24] correct. [13:24] hmm [13:24] clients are a mixture, so we do lowest common denominator and run i386 [13:24] but there's no reason interoperability between the two should be broken [13:25] did you run the rebuild like that: ltsp-build-client --arch i386 --chroot bytemark-hardy [13:25] ? [13:25] when you upgraded [13:26] wyeah [13:26] so you are 100% positive you dont have an amd64 image now [13:26] ? [13:26] yes [13:27] (which will be the default if you didnt additionally give the --arch option) [13:27] I definitely passed --arch and --chroot [13:36] * astinus sighs [13:36] i'm adding tons of debug stuff to the script which is failing and we'll go from there I guess [13:37] * ogra would try to mv /opt/ltsp /opt/ltsp.bak and re-run ltsp-build-client for a test [13:38] * astinus is doing that too [13:38] and also with a clean inetd.conf and tftp dir [14:15] We have a clean install of 7.04 Edubuntu where adding -r to /etc/default/syslogd breaks TC booting. The server logs a DHCPOFFER, but the TC never gets it. [14:17] moquist, is there an /etc/ltsp/syslog file ? [14:17] hmm, 7.04 ... ancient .... [14:19] heh, 12 months old != ancient [14:19] in the ubuntu world .... [14:19] I really disagree with the notion a large organization should need to update every 6-12 months. It's just not feasible with 2500 servers. Heh. [14:19] thats why we have LTS ;) [14:20] yeah, that was a fucking disaster with Dapper [14:20] ogra: nope, doesn't look like default installs of Edubuntu 7.04 have /etc/ltsp/syslog*, though I don't have direct access to the box in question ATM. [14:20] (which i have to admit was not really great wrt ltsp for dapper) [14:20] moquist, i know the default hasnt :) [14:22] Well, everything was fine, and my colleague added the -r and everything broke, and then he took it back out and everything got better again. Weird. [14:22] hmm [14:22] DNS issue ? [14:23] * ogra wonders if syslog tries host lookups [14:23] how would that make DHCPOFFERs logged by the server never reach the TCs? [14:23] good question [14:24] did feisty have a syslog user ? [14:24] * ogra cant remember [14:24] if so, try -r -u syslog [14:24] instead of plain -r [14:24] that way it should at least not be able to interfer with dhcpd [14:25] no matter what it does [14:27] ogra: you're right; feisty's default "-u syslog" gets tromped by putting "-r" in /etc/default/syslogd [14:27] we'll try that out [14:27] ogra: thanks [14:27] :) [17:23] ogra: ping [17:23] komputes, ? [17:23] ogra: how are you? [17:24] busy packing my bags fr tomorrow :) [17:24] gunna be a fun trip? [17:24] meeting all ubuntu devs ? sure :) [17:25] ogra: that awesome. just a quick question about the LTSPQuickInstall instructions from yesterday, I seem to be having trouble with DHCP since the server it is not listening on any interface. [17:26] did you configure the interfacs with a static IP as described ? [17:26] *interface [17:27] yes /etc/network/interfaces has been setup for static with address, netmask, network, broadcast, gateway, dns [17:27] for which IP ? [17:27] 192.168.0.140 [17:27] well, then it should work, any messages in syslog ? [17:28] from dhcpd [17:28] i canb pastebin it if it would be more helpful [17:28] let me check [17:29] ogra: pretty much the same dhcp error "Not configured to listen on any interface" [17:29] that and no subnet declaration for (restart 0.0.0.0) [17:29] but the 192.168.0.140 interface is up ? [17:29] yes [17:29] up with the ip i gave it and everything [17:30] i changed /etc/defaults/dhcp3-server to listen on eth0, still nada [17:30] dont do that ! [17:30] ok will remore and retry [17:30] revert that change [17:31] remove* [17:31] dhcpd checks all interfaces for a matching ip on startup, thats disabled if you specify one in that file [17:31] make sense [17:31] and interface names are not static nowadays [17:32] your eth0 might become eth1 or eth2 there is no guarantee it stays eth0 forever [17:32] depending on which order the kernel loads them in, i.e. usb eth adapter can be placed at high priority and get eth0 before the intrgrated controller, yes [17:33] ogra: ok so i will try restarting the dhcpd to see if it spits back an error [17:34] same thing, 1) No subnet declaration for restart (0.0.0.0) 2) Not configured to listen on any interfaces [17:35] did you make any changes to /etc/ltsp/dhcpd.conf ? [17:35] ogra: how about the change i made to /etc/ltsp/dhcpd.conf [17:35] great minds think alike ha [17:36] well, fix the file :) by default it should just work [17:36] what did you change ? [17:36] http://ubuntuforums.org/showthread.php?t=599166 [17:36] just like on there [17:36] copy paste [17:36] first post [17:37] but why ? [17:37] because i tried just booting a thinclient off the base ltsp server install and it did not work, so I tried configuring it [17:37] indeed thats broken with the LTSPQuickinstall notes as well as the defaults [17:37] :-? [17:38] * ogra shrieaks [17:38] sudo sh -c 'echo 1 > /proc/sys/net/ipv4/ip_forward' [17:38] *SIGH* [17:38] its fun if people dont know what they are doing and write documentation about it [17:39] why the heck does the poster think he needs nat ? [17:39] ;) welcome to Ubuntu! [17:39] heh [17:39] the networks are separated on purpose by default :) [17:39] its a feature that there is no nat (needed) [17:41] yay for mixinag all available info for a topic across the board [17:41] I don't know that I made a point of NAT, i just copied the /etc/ltsp/dhcp and the /etc/network/interfaces and /etc/default/dhcp which i later reverted [17:42] actually the /etc/default change was recommended by a collegue, so all this being said and done, i'm still confused on how to setup an LTSP server and client. [17:42] well, for that setup you need a 192.168.1.XX network set up first place [17:42] ogra: no need, it's just 2 pc's with a crossover [17:43] well, the instructions on LTSPQuickInstall should have given you a working ltsp without more configuration [17:43] didn't [17:43] shoulda but didna [17:43] before or after you fiddels with the files ? [17:43] before [17:43] well, would have been nice to ave a look at your setup back then [17:43] i can reset them all back to defaults and the dhcp server still won't start [17:44] * ogra doubts that [17:44] ogra: oh i'll just time machine system restore mya........oh [17:44] * komputes sticks his head in the dirt [17:44] the dhcp server wil start as soon as you define an 192.168.1.X network or change the dhcpd.conf file to match the 192.168.0.x net you have now [17:45] either will make it start [17:46] as i said before, the dhsp server looks for static interfaces matching the network setup in dhcpd.conf [17:46] let me take a look [17:46] if there is a declaration for 10.0.1.X you need an interface with a static ip like 10.0.1.10 [17:46] ok [17:47] the forums post simply breaks that default without noting it anywhere [17:48] so /etc/network/interfaces and /etc/ltsp/dhcpd.conf need to match at which line? [17:48] your interface needs to match the network you declared in dhcpd.conf [17:49] look at the subnet and range declarations [17:49] make sure the IP in your interfaces file has the same first three bytes identical [17:50] ok interfaces was on 0.x and dhcpd.conf was on 1.x [17:50] right [17:50] make them match [17:50] then start dhcpd [17:50] (and indeed make sure the interface actually uses the ip) [17:50] ifdown/ifup [17:52] ogra: made them match, same error [17:52] ifconfig reports the interface runs with that ip ? [17:53] 192.168.0.140 [17:53] and you changed the dhcpd.conf ? [17:53] to match the 192.168.0. addresspace ? [17:53] must it say 0.140 in dhcpd.conf? [17:53] no [17:53] because i have that nowhere [17:53] pastebin interfaces and dhcpd.conf [17:54] ok [17:54] thanks for your hgelp btw [17:56] ogra: pastebin.com/m1512ed78 [17:58] these are the exact files ? [17:59] and where is your loopback inerface in the ifconfig output ? [17:59] *interface [18:01] ogra: i excluded lo and wifi0 [18:01] and whats the output of: dpkg -l dhcp3-server [18:01] wifi0 ? [18:01] what ip does that have ? [18:01] sorry wlan0, all disconnected no ip [18:02] k [18:02] you want the version? [18:02] yes [18:02] dhcp3-server 3.0.6.dfsg-1ubuntu9 [18:03] hmm, thas right as well [18:03] mmmhmmm [18:03] that's why i'm kicking myself [18:03] i see no reason why dhcpd wouldnt start in a default ubuntu hardy with ths setup [18:04] its definately used in the default config by many many people [18:04] * komputes sighs [18:05] well thanks for the effort [18:05] hmm [18:05] the error message you get is quite weird though [18:05] tell me aboot it [18:05] No subnet declaration for restart (0.0.0.0) [18:06] and i did declare the subnet, but i think it may be lkooking in another file perhaps? maybe? [18:06] not if /etc/ltsp/dhcpd.conf exists [18:06] it does [18:06] that overrides everything [18:07] grep ltsp /etc/init.d/dhcp3-server [18:07] just to be sure [18:08] WOHOO [18:08] ? [18:08] restarted the computer and the thin client and I'm at the Usplash [18:08] ah, great [18:08] i wonder why you couldnt restart dhcpd then [18:08] hmm, well [18:09] i'll try it again now see if i get the same error [18:09] (that ios if i get a login prompt any time soon) ;) [18:09] you wont ... if it started properly [18:10] blinking cursor in the corner (oh blinky what does thou want?) [18:11] did you add an lts.conf file (you shouldnt) [18:12] no [18:12] what graphics card does the client have ? [18:13] amd [18:13] geode ? [18:13] geode is broken sadly, fix is on the way but might take some time still [18:14] no, not this one [18:14] https://bugs.launchpad.net/bugs/214119 [18:14] i'm working with the developer for geode on another box which is what i will be testing on this LTSP [18:15] all that amd, nsc and geode stuff is a bit confused atm [18:15] This one is just AMD not geode, but shouldn't that be automatically configured with the all wonderful new X in 8.04 [18:15] well, it has issues there were naming and driver changes [18:15] * komputes is getting anopther computer with intel video to see if he can network boot [18:16] you might need to create an lts.conf file and force XSERVER=amd or XSERVER=nsc (not sure which one your card needs) [18:16] the pci codes for amd/geode and nsc are messed up as i understand the bug [18:17] what if the network is made up of a slew of different computers (like many schools) with multiple random video cards? [18:17] well, apart from the geode mess it should be fine [18:18] i have several via and trident clients here that all work without a hitch [18:18] ok, i got to gdm for the ltsp with intel video, now i need to add a user [18:18] s/gdm/ldm/ :) [18:18] no xdmcp in ltsp5 [18:18] ah [18:18] i just got used to calling login prompt gdm [18:19] it does look diff now that you mention it [18:19] (you can force it though, but nobody made the features work for it) [18:19] ldm has an unencrypted mode you can switch on in lts.conf with the LDM_DIRECTX variable [18:20] in case you find the default encrypted ssh tunnel to slow [18:20] well I think I can manage from here, but I do hope that this becomes simpler for educational purposes [18:20] well, if X isnt broken it usually works fine :) [18:21] * ogra cant do much about upstream breaking the drivers [18:21] true. true. [18:21] i'm working on it! ;) [18:22] why are you not at UDS btw ? [18:22] alas I have no idea, i thought i was going [18:22] Q-funk the debian maintainer of the geode/amd/nsc driver will be there [18:22] it's prague right? [18:22] yep [18:22] i'm not much of a dev yet though [18:23] i'd love to meet everyone though [18:23] yeah [18:24] when you see cr3's shirt, tell him komputes did a good job! [18:24] will do :) [18:24] i'll surely spend one or the other evening with him :) [18:25] even though i dont see him listed on https://launchpad.net/sprints/uds-intrepid === juliux_ is now known as juliux [18:26] hehe, you have to light a fire under his .... maybe then he'll sprint [18:26] lol [18:26] well, i see him on the room list, all fine :) [18:27] btw, just a follow up, i created a few users, no user can log into the LTSP server [18:27] howdy. is there a meeting tonight? [18:29] highvoltage, RichEd wanted to make one, i'm not sure i can really pay attention though (packing here) [18:29] ogra: and they all have permissions for FuseFS/LTSP [18:29] komputes, what does the login maager tell you ? [18:30] ogra & highvoltage : I have a phone interview at 9:00-10:00 pm ... I will pop into the channel after it ends. [18:30] ogra: user settings or login window? [18:30] komputes, login window [18:30] We can see who is around for a meeting then, [18:30] RichEd: ok [18:31] ogra: i love the names of the utilities changing every distro, it's so....um...consistant [18:32] well, whats the error you get trying to log in ? [18:32] i bet its an ssh key issue because you used the wrong ip in the beginning [18:33] ogra: ah, do i need to keygen on the server? [18:33] (so the workstation key isnt matching) [18:33] komputes, depends what error message you get [18:33] it says that the users is not allowed to login via LTSP or something.brb [18:33] the exact message would be most helpful :) [18:34] rebooting because login window kept crashing [18:35] if its "This *Workstation* isnt authorized ..." you need to run: sudo ltsp-update-sshkeys && sudo ltsp-update-image .... get a coffee, reboot the client and log in [18:36] ok i can't get into Admin>Login Window at all, but i'll try that thx [18:36] ???? [18:36] what are you talking about ? [18:36] i'm taking about the login window on a thin client [18:37] the screen that comes up asking for userbame and password if you boot a client [18:37] komputes, what does the login maager tell you ? komputes, login window - ahhhh never mind [18:38] ok the error on the client is: The workstation isn't authorized [18:38] right [18:38] run the command from above [18:38] so let me try your recommendationsd [18:38] sudo ltsp-update-sshkeys && sudo ltsp-update-image [18:38] copy paste that ;) [18:39] (and i'm serious about the coffee ... it takes a moment) [18:39] lunch time! yabadabadooo [18:39] thx ogra, lifesaver, truely [18:39] (all that doesnt happen if you configure the IP right in the beginning btw) [18:40] in the begining meaning debian installer blue screen? [18:40] i guess all we deal with atm is based on that one issue [18:40] no, before installing ltsp-server-standalone and running ltsp-build-client [18:40] if the IP is set the keys are generated properly [18:40]