=== ogra_ is now known as ogra [10:35] Hello guys, are many people up or are you all mostly US based? [10:39] lazzurs: middle of the atlantic, in Pico island from Azores Portugal [10:43] wow that is kinda cool :) [10:44] why? [10:46] just being on a small island :) [10:47] it's not that small it has 447sqkm [10:47] just 15000 habitants [10:48] where do you live? [10:51] in Scotland [11:14] it's also an island a little bit bigger but it's an island [11:14] indeed :) [16:12] hi all, I'm trying to install clustering. Having trouble starting tftpd-ha. it doesn't show as listening in netstat or running in ps. [16:14] cyclops: not that I know anything about the edubuntu setup but are you sure it is not running from inet/xinetd ? [16:16] lazzurs, it's actually ubuntu 8.04x ,install docs from edubuntu easyclustering. I did set up as service in inetd. per wiki-docs [16:16] cyclops: cool well if it is running from inetd then you will only see a process running when it is being used [16:17] when I issue command: sudo /etc/init.d/tftpd-hpa restart , nothing happens & nodes do not boot. [16:17] that is what inetd does, it listens on the port for the application and when a connection comes in it starts the application [16:17] if you are going to start it with the init.d script then you will have to reconfigure it to run as a daemon rather than through inetd [16:18] I believe that it is set that way . [16:19] cyclops: can you paste in a link to the wiki page you are working from? [16:19] stndby [16:20] lazzurs, https://wiki.edubuntu.org/EasyUbuntuClustering/UbuntuKerrighedClusterGuide [16:22] cyclops: ok so if you followed that guide then it is running from inetd and is most likely working fine [16:22] lazzurs, dhcp3 starts, nfs-kernel starts. testing with default ubuntu image 1st , not kerrighed- though I did patch a kernel. [16:23] yet I can't get a node to boot. [16:23] what happens when it boots? [16:23] nothing , pxe is looking for dhcp server [16:24] thats not a tftp error [16:24] I am on the server machine now. 2 x nics [16:24] are you sure there is no second dhcpd running anywhere in your network ? [16:25] ok so the nic eth0 that dhcp will be listening on, is that the same network that the clients are running on [16:26] eth1 (100x) dhcp client to internet eth0 (1000x) dhcp3-server ,just 1 node connected right now on a switch. [16:27] and that node is connected to eth1? [16:28] no to eth0 [16:28] eth1 is traight out to firewall & internet [16:29] what address do you have on eth0 and did you copy the dhcp configuration exactly from the wiki? [16:30] netstat -lu does not show tftp listening on udp. [16:30] cyclops: it might show inetd listening however...which might need restarted after the reconfiguration [16:30] that document is broken TBH [16:31] cyclops: to restart the tftp server, you do: sudo invoke-rc.d openbsd-inetd restart [16:31] To test the tftp server, just install a tftp client and try to connect and download a file from the server [16:32] To see if the dhcp server is running, you do: sudo invoke-rc.d dhcp3-server status [16:32] sudo invoke-rc.d openbsd-inetd restart [16:32] * Restarting internet superserver inetd [ OK ] [16:33] Status of DHCP server: dhcpd3 is running. [16:34] sudo apt-get install tftp-hpa (that's the client) [16:34] tftp localhost [16:34] get /ltsp/i386/pxelinux.cfg [16:34] still don't see any change in netstat -lu for tftp , not listed [16:35] I am not using ltsp. [16:35] OK, use a different test file [16:35] if your PXE client looks for an address you are not even near the tftp stage [16:35] You don't see this in netstat -lu? udp 0 0 *:tftp *:* [16:35] no [16:36] am starting tftp as per the docs - easyubuntuclustering. [16:37] " nothing , pxe is looking for dhcp server" [16:38] thats what you said above [16:38] ok so then i should look 1st to dhcp3-server error, as node is still looking for dhcp. [16:38] ltsp or not, if your pxe client is looking for a dhcpd that should be the first issue you solve .... [16:40] I have the dhcp-conf set to hand out fixed, no range. i will try to change to a range. [16:40] woah, thats a broken guide [16:41] there seems to be not a single default he doesnt set differently for no purpose at all [16:41] ogra, thanks for the info. so what scrap the installation? [16:41] well, it seems to have worked for him [16:42] just debugging something that was set up after suchg a guide isnt easy, since all defaults are different [16:43] # /etc/default/dhcp3-server # [16:43] interfaces="eth0" [16:43] thats just mean [16:43] #Defaults for tftp-hpa [16:43] RUN_DAEMON="YES" [16:43] OPTIONS="-l -s /srv/tftp" [16:43] that completely contradicts with the part below [16:43] your tftp cant start if its run by two daemons ... it will just clash [16:43] I realize that, it seems the only guide around dealing with kerrighed. [16:50] start with reverting /etc/default/dhcp3-server and /etc/default/tftpd-hpa [16:50] (this netsplitting of freenode is annoying today) === ogra_ is now known as ogra [16:52] oh, the inetd setup surely breaks if you dont use xinetd [16:53] (which isnt the default inetd in ubuntu) [16:53] trying to apply xinetd configs to openbsd-inetd isnt going to work at all [16:54] ogra, so apt-get install xinetd? [16:55] no [16:55] use the instealled inetd [16:55] just remove the cruft you added per that howto [16:55] then comment in the old tftp line [16:56] ok :)...nice vocabulary. [16:56] and if you really want to go with /srv/tftp change the old line from /var/lib/tftpboot to /srv/tftp [16:57] yes, doing that now. [16:58] then restart openbsd-inetd [16:58] i dont get why he makes you add xinetd config to /etc/inetd.conf [16:58] that doesnt make any sense at all [16:59] ok restrarted inetd [16:59] now try what alkisg told you [17:00] i see tftp in netstat. [17:00] ah, good [17:01] now to your dhcpd [17:01] ok [17:01] did you revert /etc/default/dhcp3-server ? [17:01] interfaces="" is the correct setting [17:01] u mean revert back to default orig? [17:01] right [17:02] stndby. [17:02] setting the inetrfaces line enforces dhcpd to start even if it cant serve anything ... thats a very bad thing [17:02] interfaces="" makes it run a selftest so it tells you if there is actually an interface it can work on [17:03] ok have the orig back [17:03] good [17:03] try to restart your dhcp3-server [17:03] and see if it starts up [17:04] restart - fail. [17:04] right, thats what i expected [17:04] paste the output of "ifconfig -a" to paste.ubuntu.com [17:06] ogra, done: http://paste.ubuntu.com/138395/plain/ [17:07] looks ok so far, do you use exactly the dhcpd.conf file from the howto ? [17:08] highvoltage: ping [17:09] I did original, but now i deleted everything below for nodes, & left a range. I could revert-though. [17:10] try that [17:10] also you dont have ltsp installed, right ? [17:10] sory it is already orig per ur suggestion. [17:11] else the files needs to be in /etc/ltsp/ [17:11] stndby. [17:14] ogra, /etc/default/dhcp3-server & /etc/dhcp3/dhcpd.conf are orig! [17:15] and you dont have ltsp-server or ltsp-server-standalone installed, right ? [17:16] correct [17:18] ok, so now copy the exact file from the howto in place for /etc/dhcp3/dhcpd.conf [17:19] ok [17:21] ogra, there is a mistake with the howto conf, he is missing declaring subnet-mask. [17:22] subnet 192.168.1.0 netmask 255.255.255.0 { [17:22] looks ok to me [17:22] at the dns section ? [17:22] # Information about the network setup [17:23] shall I try dhcp3 restart? [17:23] yep [17:23] it shouldnt fail [17:24] * Starting DHCP server dhcpd3 [ OK ] [17:24] good [17:24] you need to make sure the MAC addresses match for your nodes though [17:25] else your clients wont get IPs [17:26] aw..not an easy thing with built in pxe boot on each. [17:26] what if i use etherboot? [17:26] do i still need the mac address? [17:27] if you want fixed ip adresses for your nodes you always need the mac address [17:29] oh, i can be without fixed, don't think it's necessary. [17:30] well, then just drop all the group stuff [17:30] and add: range 192.168.1.20 192.168.1.250; [17:30] ...and also drop: deny unknown-clients; [17:30] right in the subnet declaration [17:31] that will serve dynamic IPs between 20 and 250 [17:31] ok, for now i added the mac, & node 1 is trying to get tftp. [17:31] it didn't get tftp. [17:32] whats the error ? [17:33] a timeout. [17:33] do you have all files in place in /srv/tftp [17:34] pxelinux.0, pxelinux.cfg, vmlinuz and initrd.img should be in there [17:35] pxelinux.cfg being a dir with a single file in it called "default" [17:35] am defaulting to 1st test scenario, with initrd.img-2.6.24-23-generic & vmlinuz-2.6.24-23-generic & pxelinux.0 [17:35] pxelinux.cfg/default ? [17:35] yes [17:36] and the file is configured right i assume ? [17:37] as per the howto 1st test setup [17:38] try what alkisg said before [17:38] tftp localhost [17:38] u mean local tftpclient? [17:38] get pxelinux.cfg/default [17:39] or get pxelinux.0 [17:40] from where? [17:40] tftp localhost [17:40] get pxelinux.0 [17:40] try that [17:40] after you installed tftp-pha [17:40] *hpa [17:42] no get command [17:44] can you elaborate ? [17:45] sorry, installed tftp-hpa, then tftp localhost, then get pxelinux.0 & transfer timed out. [17:46] aha [17:46] grep tftp /etc/inetd.conf ... [17:46] paste the output here [17:47] tftp dgram udp wait root /usr/sbin/in.tftpd /usr/sbin/in.tftpd -s /srv/tftpboot [17:48] aw, i see the mistake. [17:49] :) [17:51] ok node 1 loading . [17:52] ok got a login prompt. [17:52] good :) [17:52] congrats [17:52] this is with default image, i will try with the kerrighed image-already made. [17:53] thank - mucho. [17:55] enjoy [17:56] wait, i am going to reboot node to see if it works.stndby. [17:58] it didn't boot , bu i think it's a name problem , I will work on it. Thanks for all the help. the howto should be amended [17:59] or fixed :) [18:01] might be the patched kernel wasn't built properly , I will redo it. [18:05] ogra, I have the kerrighed image booting up now on node1 [18:05] good [18:06] or bad, whatever kerrighed is :) [18:07] is supposed to be a replacement for openmosix using 2.6 kernels, or clustering as SSI. [18:10] right, and i think there were reasons both was abandoned upstream [18:10] but i dont have enough insight to judge :) [18:12] I understand. I have some apps for a friend that needs massive computing & rendering, so I was trying to help him. === Eeyore-Jr is now known as Ahmuck