errr | Using MAAS Version 2.1.1+bzr5544-0ubuntu1 (16.04.1) is there a limit to the number of machines returned by the API in a single request? | 06:41 |
---|---|---|
brendand | errr, not that i know of | 07:37 |
=== frankban|afk is now known as frankban | ||
errr | from what I could find in the source code thats what it was looking like to me but I wasnt 100% sure | 07:54 |
=== rvba` is now known as rvba | ||
aaran | Hi, tried to get support regarding an issue yesterday however there was no one around so I will ask again | 09:36 |
aaran | when i tell a client to network boot it gives the error Unable to locate configuration file | 09:36 |
aaran | After restarting the server the client now boots into a ubuntu instance however the server is showing commissioning until it fails | 10:00 |
g3 | hey yo! | 19:49 |
g3 | Anyone around? | 19:50 |
g3 | =( | 20:01 |
errr | is it possible to provision based on tags a machine has been given? | 20:08 |
errr | Using MAAS Version 2.1.1+bzr5544-0ubuntu1 (16.04.1) | 20:08 |
g3 | setting up maas for the first itme, currently stuck at One cluster is not connected to the region. | 20:12 |
g3 | Currently have a cluster controller setup at a network interface handling dhcp and dns | 20:13 |
g3 | and /etc/maas/clusterd.conf reflects that information | 20:14 |
g3 | Hmmm apt installed 1.9 | 20:21 |
pmatulis | errr, how are you provisioning? | 21:08 |
pmatulis | g3, i suggest using maas 2.1 if you're setting up for the first time | 21:09 |
g3 | Yeah I just realized I can't use Trusty for that | 21:10 |
g3 | So I'll have to upgrade another server first. | 21:10 |
roaksoax | errr: yes | 21:18 |
roaksoax | errr: you can 'allocate' a machine based on tags in order to deploy it | 21:19 |
roaksoax | errr: so you can allocate + deploy the machine | 21:19 |
=== menn0_ is now known as menn0 | ||
andrew-ii | What could cause a node to properly boot via PXE but not enlist? | 21:39 |
roaksoax | andrew-ii: lack of network access ? or not being able to contact the maas server ? | 22:03 |
andrew-ii | It seems to do ok for the most part | 22:04 |
roaksoax | andrew-ii: do you have a console log I can see ? | 22:04 |
andrew-ii | It boots under maas direction, and only seems to trip up at the end during the enlist stage | 22:04 |
andrew-ii | Lemme double check if rsyslog captured any of the end bits | 22:05 |
roaksoax | andrew-ii: if you could get me an error output or the whole output for the matter, I would be able to have an idea of what's wrong | 22:07 |
andrew-ii | Neat - is there a preferred place to put a log? (It's about 3.9MB at the moment) | 22:08 |
roaksoax | andrew-ii: not really, a pastebin would be good | 22:08 |
roaksoax | andrew-ii: if you have the snippet of the failure, that's better | 22:08 |
andrew-ii | Let me sed out the ureadahead bits and see if that makes it more obvious | 22:09 |
andrew-ii | Sorry for the delay - here's the last attempt: http://pastebin.com/FkMnpY8f | 22:13 |
andrew-ii | The strange bit is that I've got 4 other machines that commissioned happily. Just this one seems to get confounded. | 22:14 |
andrew-ii | Note that all the machines are different - it's a horrific Frankenstein's monster of boxes | 22:16 |
roaksoax | andrew-ii: | 22:20 |
roaksoax | Jan 25 21:56:24 maas-enlisting-node cloud-init[2325]: curl: (22) The requested URL returned error: 400 BAD REQUEST | 22:20 |
roaksoax | Jan 25 21:56:24 maas-enlisting-node cloud-init[2325]: + maas-enlist --serverurl http://10.222.222.10:5240/MAAS/api/2.0/machines/ | 22:20 |
roaksoax | andrew-ii: can the machines access 10.222.222.10 ? | 22:21 |
roaksoax | andrew-ii: or is that in a different subnet | 22:21 |
andrew-ii | They should - the maas region/rack controller is 10.222.222.10 and all the machines are on the same set of switches | 22:21 |
andrew-ii | That's the PXE address, at least | 22:22 |
roaksoax | andrew-ii: right, so tail -f /var/log/maas/regiond.log while that happens, is there any error ? | 22:22 |
roaksoax | andrew-ii: my guess, however, is that the machine itself cannot ping the above IP | 22:23 |
andrew-ii | Just did it and literally caught it | 22:23 |
andrew-ii | I think this is what you mean? 2017-01-25 16:22:17 -: [info] ::ffff:10.222.222.50 - - [25/Jan/2017:22:22:16 +0000] "POST /MAAS/api/2.0/machines/ HTTP/1.1" 400 119 "-" "curl/7.47.0" | 22:24 |
andrew-ii | Running another commission attempt to see if I can see it live | 22:25 |
andrew-ii | cat /tmp/enlist.out yields empty string, which is a bit disappointing | 22:33 |
andrew-ii | I think it was related to https://bugs.launchpad.net/maas/+bug/1520645 | 22:43 |
andrew-ii | Basically, I rebuilt the machine, this time adding *all* four NIC MACs instead of just the first. I guess it was attempting to phone home to MAAS via the second NIC, and that MAC wasn't included yet. | 22:45 |
andrew-ii | At any rate, thanks roaksoax! I was wondering how it could possibly boot but not be able to enlist. I guess the node just didn't reply with the first NIC =/ | 22:53 |
=== frankban is now known as frankban|afk | ||
g3 | Okay I'm now on 2.0 | 23:16 |
g3 | Should I upgrade to 2.1? | 23:16 |
g3 | So what does an untagged VLAN mean exactly? | 23:44 |
g3 | I setup a subnet with dhcp on a specific port | 23:44 |
andrew-ii | An untagged VLAN is basically stuff without VLAN tags/numbers | 23:46 |
andrew-ii | I'm not sure about dhcp ports (I'm currently trying to get a node to contact the outside world through the region controller) | 23:47 |
g3 | I see | 23:48 |
g3 | So now that I have MAAS setup with a subnet serving DHCP on a fabric. | 23:49 |
g3 | next setup would be to try to provision a node? | 23:49 |
andrew-ii | I'd give it a shot! I think PXE is untagged by default, hence why you can't provision on a VLAN. | 23:51 |
g3 | I don't think a VLAN is currently needed as I am using an entirely seperate network for this? | 23:52 |
andrew-ii | Right | 23:52 |
g3 | Can MAAS handle IPMI stuff? | 23:52 |
andrew-ii | At least, I think it's recommended to let maas be on its own hardware; that's what I'm trying | 23:52 |
g3 | LIke have MAAS provide DHCP and then use those IP's for IPMI management stuff? | 23:53 |
andrew-ii | Generally I think the IPMI stuff is pretty good | 23:53 |
andrew-ii | I have an HP iLo2 Gen 5 server off Craigslist that apparently has a firmware bug that just refuses to play ball with it, but otherwise every other machine seemed to connect well | 23:53 |
g3 | Oh and I see that it will handle APC pdu things interesting. | 23:54 |
andrew-ii | I set static IPs for the IPMI connections, but I think the node discovery may actually figure it out for you; I haven't tried that yet | 23:54 |
g3 | Yeah it looks like post-commission you can have it set up all of the network interfaces. | 23:55 |
g3 | Do you know the difference between commission, acquire and deploy? | 23:56 |
andrew-ii | Commission: bootstrap the machine and get it to report back settings; Acquire: plonk your name on it and reserve the node (I think); Deploy: Spin up the node with an image (full install) | 23:57 |
andrew-ii | Someone probably has a better answer about what Aquire does | 23:58 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!