[14:02] rvba, blake_r, mpontillo, roaksoax, maas+juju net call? [14:05] dimitern, they are sprinting this week fwiw [14:05] Hi dimitern. [14:07] kiko, rvba, hi [14:07] ok, nothing new [14:07] happy sprinting ;) === roadmr is now known as roadmr_afk === roadmr_afk is now known as roadmr [21:03] hello all, got an enlistment question [21:03] the first node that i'm trying to add to my maas cluster boots from pxe, ultimately ends up at an ubuntu login prompt, but my cluster controller has no knowledge of the node3 [21:03] of the node* [21:05] http://pastebin.com/xQedSKqV [21:26] http://pastebin.com/gXrMwwUF [21:34] travnewmatic: is the node in MAAS marked New ? [21:34] the node does not show up anywhere in my maas interface [21:34] but it did boot into an os over the network [21:34] so the dhcp stuff is working [21:34] and i can ping it [21:35] but its like the region controller or the cluster controller have no knowledge of it [21:38] travnewmatic: so it needs to dhcpd from maas, then pxe, then you will see a prompt like maas-enlist [21:38] travnewmatic: and the node needs to do stuff inseide of itself to register itself in maas [21:38] hmm [21:39] well iknow dhcp is working [21:39] and the pxe part is pushing an image to the server [21:41] i see a table [21:42] net device info [21:42] route info [21:42] and then it goes into some url_helper.py[WARNING] stuff === menn0_ is now known as menn0 [21:42] errno 113 no route to host [21:45] http://i.imgur.com/A5D5hXR.jpg [21:46] then this http://i.imgur.com/RE4cHtm.jpg [21:48] aaaaand then this http://i.imgur.com/6SqfzvD.jpg [21:48] and if i hit enter the usual ubuntu login: pops up [21:48] travnewmatic: can your nodes contact 192.168.100.2 ? [21:48] travnewmatic: 192.168.100.2 is the address of the MAAS region server? [21:49] travnewmatic: it seems the nodes can't access it [21:49] yes! [21:49] hmm :| [21:49] alright [21:49] subnet issue? [21:50] travnewmatic: that might be it. can you ensure your nodes would be able to ping the maas server? [21:50] lemme check [21:50] travnewmatic: is the DHCP range configured in maas in the same network, and can actually ping? [21:50] well [21:50] as it is now [21:50] i only have one node [21:50] and that node is presenting me with a login prompt [21:50] and i'm not sure what the un/pw is [21:51] also [21:51] i first setup the region controller [21:51] then i setup the cluster controller [21:51] travnewmatic: that's fin [21:51] travnewmatic: let the user auto register itself in maas, and you will see it appear there [21:52] hmm [21:52] well i ran the dkpg configure cluster controller command thing [21:52] put in the api address of the region controller [21:52] and added the second maas controller to the first [21:53] so the first maas controller, the 192.168.0.2 machine, lists 2 clusters [21:53] on the region controller, i went into that added cluster controller and added a managed interface, with dhcp and dns [21:54] the cluster controllers address is 192.168.0.3 [21:54] when i go to 192.168.0.3/MAAS [21:54] its like everything is fresh [21:54] i dont see that managed interface that i created [21:55] although it does exist on the 192.168.0.2 interface for the 192.168.0.3 cluster controller [21:55] travnewmatic: that's because you don't need to install the Region controller on the Cluster controller [21:55] aaaaaah okay [21:55] travnewmatic: when you install sudo apt-get install maas , it will install both Region (maas-reiogn-controller) and Cluster (maas-cluster-controller) [21:55] uhuh [21:56] travnewmatic: if you want to add *another* cluster controller, which is not running on the same machine of the region, you only have to install maas-cluster-controller, and then dpkg-reconfigure maas-cluster-controller to point it to the *region* controller [21:56] mmm i see [21:57] when i did the cluster controller, i did it from the ubuntu server install media maas option [21:57] travnewmatic: so the base installation (sudo apt-get install maas) should allow you to control your maas and configure the *default* cluster (the one runnin with the region) [21:57] mhm [21:57] travnewmatic: and if you have nodes that can talk ot the maas server, you just need to allow them to PXE boot and that's it [21:58] so the cluster controllers act as intermediaries between nodes and the region controller [21:59] travnewmatic: correct [21:59] hmm [22:00] https://www.dropbox.com/s/bjhn5qrs3ma02rf/Screenshot%202015-02-17%2015.59.38.png?dl=0 [22:00] so if i wanted to put everything on the same subnet [22:01] i think right now i'm using 192.168.0.1 through 192.168.0.4 [22:02] what is the router ip? [22:03] travnewmatic: depends whether you hvae a default gw, or maas is your default gw [22:03] i've got a box doing nat [22:04] so 192.168.0.1 is what i've got everything point to for their gateway [22:06] travnewmatic: say, your router is 192.168.0.1, your maas server 192.168.0.2. When you configure your dynamic range, it will use 0.4 to 0.14, and static range from 0.15 to 0.25 [22:06] travnewmatic: so router address is 192.168.0.1 [22:06] mhm [22:06] looks like he has 1.5 there. not 1.7. [22:07] maas I mean. [22:07] travnewmatic: ah, so if there's just 1 range, then you only need 1 range [22:07] i'm putting everything on 192.168.0.0 [22:10] http://pastebin.com/0nRnmUpH [22:11] https://www.dropbox.com/s/duoi6f19nt8vlw5/Screenshot%202015-02-17%2016.10.59.png?dl=0 [22:12] aaand now dhcp and pxe is borked :D [22:16] restarting my cluster controller [22:16] the cluster controller has the only managed interface [22:19] i forgot to change the network in the network tab in the maas interface [22:19] might that have had something to do with it? [22:21] dpkg-reconfigure maas-cluster-controller didn't help to get dhcp back? [22:23] yaaay pxe works again [22:23] i think [22:23] hmm [22:23] now its just sitting there [22:23] TFTP. [22:23] gateway ip is 192.168.0.1 [22:24] http://imgur.com/2vQfpf7 [22:25] hanging on the tftp thing [22:26] HNNNG it should be .2 [22:26] hold on [22:26] should it? [22:27] my gateway to my nat box is .1 [22:27] region controller is .2 [22:27] cluster controller is .3 [22:27] and the interface with dhcp is .4 [22:28] You set the router IP to 192.168.0.1. It should be .1. [22:28] so thats right? [22:29] yeah its still hangin on the tftp thing [22:29] That's right, it's getting IP address from the dhcp server which you set to .4. But I *think* the node needs to reach cluster controller to get the image files. [22:29] but it is getting an address from dhcp [22:29] hmm [22:30] could you set eth1 internface as unmanaged on maas web ui, and set eth0 interface to manage dhcp and dns with the same settings, except IP would be 192.168.0.3 not 192.168.0.4. [22:30] this is odd i was getting farther with having stuff on a separate subnet [22:30] hmm alright i can try that [22:31] so i'll ditch eth1 [22:32] https://www.dropbox.com/s/026yyfjix0h5zpr/Screenshot%202015-02-17%2016.32.13.png?dl=0 [22:32] didnt ditch it just set it to unmanaged [22:32] that looks good. [22:32] and moved the config stuff to eth0 [22:33] config stuff? [22:33] i mean [22:33] the dhcp and dns form stuff [22:33] router ip , high, low, etc [22:33] right. [22:33] aaaaaand we're on our way [22:33] pxe pushing stuff over as it should [22:35] maas-enlisting-node login! [22:35] wow thanks guys this is really flipping sweet [22:35] great. does the node show up on the nodes page on maas web ui now? as New. [22:36] mmm [22:36] does not :( [22:36] or Declared [22:36] the node should power off, after that it will show up on the web UI. [22:37] erhmahgerd thats exactly what it did [22:38] https://www.dropbox.com/s/8v58n0ppiswxhoa/Screenshot%202015-02-17%2016.37.28.png?dl=0 [22:38] https://www.dropbox.com/s/q33bvk4nz7s5wgx/Screenshot%202015-02-17%2016.38.00.png?dl=0 [22:38] great. now you can commission the node, so MAAS knows how many CPU cores and RAM it has. [22:39] i'm not sure it knows how to power it back on [22:39] Click Edit node on the node page. [22:40] MAAS should have created IPMI credentials to power control the node. [22:42] did not :( [22:42] manually powering it back on [22:43] could you show the screen shot of the node details page (click Edit node). [22:43] unmomento [22:45] https://www.dropbox.com/s/10b8tv3izrs9jmp/Screenshot%202015-02-17%2016.44.55.png?dl=0 [22:46] It says the cluster controller for this node is not responding, power type validation is not available. [22:47] :( [22:47] do you mind doing dpkg-reconfigure maas-cluster-controller again to make sure you tell the cluster controller where region controller is? [22:48] after the restart https://www.dropbox.com/s/ids7cnfcm1jntu7/Screenshot%202015-02-17%2016.47.40.png?dl=0 [22:49] done [22:49] If it's all working as it should, you should not need to power on/off the node manually, MAAS should take care of that for you. [22:50] MAAS region controller API URL should be http://192.168.0.2/MAAS for the cluster controller configuration. [22:50] yep thats exactly what i put [22:51] you mentioned you had two cluster controllers. [22:51] no, a region and a cluster [22:51] ok. [22:52] i don't remember exactly how maas 1.5 should show the power control info on the node page. [22:53] well [22:54] thres a dropdown in the edit page "power type" [22:54] MAAS should have it filled up automatically after the enlistment. [22:55] i don't know what log files to look at in /var/log/maas/ to find out what's going on. [22:56] ERROR 2015-02-17 16:54:46,136 maasserver Unable to get RPC connection for cluster 'maas' [22:56] might this be it? [22:57] https://bugs.launchpad.net/maas/+bug/1350925 [22:57] Launchpad bug 1350925 in MAAS "Unable to get RPC connection for cluster 'maas'" [Critical,Fix released] [22:58] poop [23:00] could be DNS hostname resolve issue [23:01] hmmmm [23:01] or an ssh key issue [23:07] does putting hostname in the /etc/hosts file help? [23:07] if anyone uses chef here, i just pushed up: https://rubygems.org/gems/knife-maas [23:08] it would be awesome if yall could give it a shot and report any bugs to the issues page [23:08] catbus1, on the region controller? [23:14] yeah i tried it on another 1950 and it doesnt like the ipmi [23:15] i see that thread though [23:15] https://bugs.launchpad.net/ubuntu/+source/maas/+bug/1321885 [23:15] Launchpad bug 1321885 in openipmi (Ubuntu) "IPMI detection and automatic setting fail in Ubuntu 14.04 maas" [Undecided,Confirmed] [23:17] travnewmatic: I am not sure actually. I have never deployed region and cluster on different nodes. Region couldn't contact cluster, and it could be name resolving issue, cluster is providing DNS service, so region should reach to cluster for DNS service. I would try this, edit /etc/resolv.conf and have cluster controller IP as the first nameserver entry, and see if it works. [23:17] well [23:17] i could try cutting out the cluster controller [23:18] does this not appear to be an issue with the ipmi in the 1950? [23:18] my error does seem to match what is described in that bug [23:18] it could be related to hardware. yes. [23:21] travnewmatic: you can find your maas version with apt-cache policy maas [23:21] 1.5 [23:22] fix verified with 1.7 [23:23] so i should upgrade [23:24] travnewmatic: you can get 1.7.1 from maas-maintainers/stable ppa, sudo add-apt-repository ppa:maas-maintainers/stable [23:25] gotcha [23:25] sudo apt update and sudo apt install maas [23:27] choose to use the new configuration template during the install maas process. after it's upgraded to 1.7, you need to re-import the image files and there is a new IP range (Dynamic and Static) in the network interface you will need to configure. [23:28] hmm i see [23:30] both region and cluster are upgrading [23:30] Y or I : install the package maintainer's version [23:31] this is what you're referring to? [23:31] yes. [23:31] gotcha [23:32] i really appreciate all the help this afternoon [23:40] shit you guys are awesome https://www.dropbox.com/s/0rq88uhv6872tcb/Screenshot%202015-02-17%2017.39.57.png?dl=0 [23:41] i was wondering why some of the screenshots in some of the guides looks a tad different than what i was seeing on my screen [23:42] wow it seems like this update adds a lot [23:49] also it appears that my cluster controller has been disconnected from my region controller after the upgrade [23:49] travnewmatic: yeah 1.7.1+ is the way to go [23:50] most of the docs are 1.5 [23:50] ive noticed [23:52] hopefully the documentation will catch up soon [23:56] where does juju get installed? on the region controller?