[01:28] <pepe233> Hi
[12:28] <CyborgCygnus> Are maas & juju two completely different things, are they competitors or do you use them together/same time?
[12:32] <rbasak> CyborgCygnus: they perform at two different layers in the stack.
[12:32] <rbasak> CyborgCygnus: Juju works at "I need an instance" level. It can use EC2, Openstack, Azure and others to fulfil that need. Or it can use MAAS.
[12:33] <rbasak> CyborgCygnus: MAAS is a "cloud provider" in the same sense as EC2's "just give me a working instance" by API model. Except that it uses bare metal to get those instances.
[12:34] <rbasak> CyborgCygnus: so, MAAS to provision bare metal, and Juju to make use of that in a smart way. Or just use MAAS for bare metal and have something else call the MAAS API directly. Or just use Juju with a different provider.
[12:42] <CyborgCygnus> rbasak, okay still confusing but good information, I've been eyeing off all of these ubuntu server, openstack cloud & other affiliated those tools.
[12:43] <rbasak> CyborgCygnus: still confusing> maybe I can try again? MAAS is relatively simple to explain. Get freshly installed bare metal machines from a pool of bare metal machines by API. MAAS takes care of managing the machines and provides the API. Does that make sense?
[12:45] <CyborgCygnus> rbasak, yeah get that pretty much, does juju use maas to do its thing? I was confused as to whether maas is the essential thing to have an juju just an accessory for web browser based controller for ease
[12:47] <rbasak> CyborgCygnus: let me try and explain Juju first. Juju lets you supply a model of the deployment you want in terms of a set of services and how those services are connected to each other. Then it turns that model into reality by deploying instances are required.
[12:47] <rbasak> by deploying instances as required.
[12:47] <rbasak> Does that make sense?
[12:47] <rbasak> That means that Juju needs to somehow deploy instances.
[12:47] <rbasak> To do that Juju uses a pluggable architecture of "providers".
[12:48] <rbasak> There is a Juju MAAS provider, which you can use to turn your Juju model into reality on bare metal.
[12:48] <CyborgCygnus> rbasak, okay yeah a bit more understandable for noob slow me
[12:48] <rbasak> There is also a Juju EC2 provider, which you can use to turn the same Juju model into reality on EC2 instead.
[12:48] <CyborgCygnus> rbasak, which one do you prefer, EC2 or MAAS?
[12:49] <rbasak> CyborgCygnus: depends on where you want your deployment.
[12:49] <rbasak> CyborgCygnus: want it on your own physical machines? You'll spend money on those machines, electricity and hardware maintenance but nothing after that.
[12:50] <rbasak> CyborgCygnus: or, instead, want Amazon to take care of capital expenditure, electricity and hardware maintenance for you? Then use EC2 instead.
[12:50] <CyborgCygnus> rbasak, Oh okay :S , so say at home for some reason I wanted 6 machines in the house to install ubuntu through the one server & have their accounts stored on the server so they can log on from any computer & log into their account
[12:51] <rbasak> 6 machines to do what, and what do you mean by account exactly?
[12:54] <CyborgCygnus> rbasak, well I remember way back when I was in high school every student had a login that they can put into any school pc & they get access to their files aswell as the operating system. The 6 machines just ubuntu operating system
[12:56] <rbasak> CyborgCygnus: so you want 6 machines running Ubuntu desktop with central authentication and a central file server? You can do that with Ubuntu, all the tools are there. But there isn't a turnkey system to provide that - most sysadmins put together a customised deployment manually.
[12:57] <CyborgCygnus> rbasak, so what would be an example use of the ubuntu server with openstack, maas & juju?
[12:59] <rbasak> CyborgCygnus: a simple example? A web app.
[13:00] <rbasak> CyborgCygnus: a complex example is an Openstack deployment itself, though that doesn't generally use Openstack obviously.
[13:04] <CyborgCygnus> rbasak, alright thanks, So deploying a web app to where? Or where do you deploy openstack to? As in installing it to another server or computer over the network to ?
[13:05] <rbasak> CyborgCygnus: deploying a web app to anywhere that can provide instances. EC2, Azure, an Openstack cloud, or directly on bare metal with no virtualisation with MAAS (though for a basic web app that'd be quite wasteful).
[13:05] <rbasak> CyborgCygnus: deploying Openstack is normally to bare metal in a datacentre somewhere in order to run a private cloud - like EC2 but a private one.
[13:13] <CyborgCygnus> rbasak, thanks for your help
[13:35] <smoser> garibaldi, sorry, i went afk
[13:35] <smoser> the 'maas list' shows the accounts that you have set up.
[13:36] <smoser> so you did: maas login some-name-here some-url-here some-creds-here
[13:36] <smoser> then, from then on out, MYMAASNAME is 'some-name-here'
[13:36] <smoser> ie, you have to tell the maas cli which maas end point you're talking to
[14:10] <garibaldi> smoser: I commissioned a new node and then after was able to successfully start it with "maas local node start <system_id> distro_series=trusty user_data=$(base64 --wrap=0 /tmp/script.sh)"
[14:11] <garibaldi> smoser: /tmp/script.sh contains the bash shebang followed by "echo "test $(whoami)" > /home/ubuntu/test.txt" on a new line; the node successfully installed the OS, but once it was done /home/ubuntu/test.txt did not exist so it seems it didn't run my user_data script
[14:11] <garibaldi> smoser: what could be wrong?
[14:12] <smoser> htat should work
[14:12] <smoser> can you pastebin /var/log/cloud-init.log and /var/log/cloud-init-output.log ?
[14:20] <garibaldi> smoser: yes, I'm re-installing the node now but when it finishes I will post them
[14:25] <smoser> k. thanks.
[15:24] <mup> Bug #1461135 was opened: Can't assign the same IP address to two different MAC addresses <networking> <MAAS:Triaged> <https://launchpad.net/bugs/1461135>
[16:04] <drhalan> hey all. is there a way to run a custom script after startup of a node? basically I want to deploy a ssh key so i have root access to the machine. our hardware only has WOL and we need a way to shut them off...
[16:45] <smoser> drhalan, you can use user-data
[16:46] <drhalan> smoser: what's that exactly?
[16:46] <smoser> user-data is the same concept in maas that it is in ec2.
[16:47] <smoser> cloud-init runs in maas or ec2 instances and consumes user-data that the user fed it.
[16:47] <smoser> what you can do with user-data is described about:blank
[16:47] <smoser> err..
[16:47] <smoser> https://help.ubuntu.com/community/CloudInit
[16:48] <smoser> and you can deploy a node with user-data by passing user_data key to maas node start
[17:45] <mup> Bug #1461181 was opened: [1.8rc1] Too many open files, after upgrade to rc1 <MAAS:New> <https://launchpad.net/bugs/1461181>
[17:56] <kiko> hah!
[18:25] <garibaldi> smoser: here are my cloud-init.log (http://pastebin.com/vvbbDNDm) and cloud-init-output.log (http://pastebin.com/1h1snXQM) files
[18:25] <garibaldi> smoser: the command I used to start the VM (after commissioning it) was maas local node start <system_id> distro_series=trusty user_data=$(base64 --wrap=0 /tmp/script.sh)
[18:26] <garibaldi> smoser: where /tmp/script.sh contains the bash shebang followed by "echo "test $(whoami)" > /home/ubuntu/test.txt" on a new line
[18:33] <smoser> garibaldi, just for your own sanity, there is a tool called 'pastebinit' that allows use of sane paste bins via:
[18:33] <smoser> pastebinit /var/log/cloud-init.log
[18:33] <smoser> or pastebinit < /var/log/cloud-init.log
[19:08] <drhalan> is there a way to set a comment or something for each machine that for example indicates the location of the machine?
[19:08] <smoser> garibaldi, it does seem like you didn't get user_data through though
[19:08] <drhalan> we have a tag on each machine that i would like maas to know about
[19:18] <smoser> drhalan, you may be able to abuse tags for that.
[19:21] <mup> Bug #1461226 was opened: Clicking "Nodes" in 1.8 takes several seconds to load - seems much slower than 1.7.5 <MAAS:Incomplete> <https://launchpad.net/bugs/1461226>
[19:21] <mup> Bug #1461233 was opened: Node power summary less informative in 1.8 <MAAS:New> <https://launchpad.net/bugs/1461233>
[19:25] <smoser> drhalan, maas admin tags new name=row35
[19:25] <smoser> maas admin tag update-nodes row35 add=node-260cbb52-094d-11e5-a379-00163ee0d91e
[19:25] <smoser> its not a perfect fit.
[19:25] <drhalan> maybe i can just maintain a list outside of maas :)
[19:25] <drhalan> thanks!
[19:48] <mup> Bug #1461236 was opened: Clicking Nodes makes three webfont requests that return 404 in 1.8 UI <MAAS:New> <https://launchpad.net/bugs/1461236>
[19:59] <drhalan> i am using the fast installer and get the following error during install. any idea? http://pastebin.com/m5jmmxdG
[20:32] <garibaldi> smoser: thanks for the tip about pastebinit; any ideas on why the user_data did not get through for this node?
[20:36] <smoser> garibaldi, well...
[20:36] <smoser> could you verify that it didn't ?
[20:37] <smoser>  can you verify taht /var/lib/cloud/instance/user-data.txt has nothing in it ?
[20:37] <smoser> and also.. what version of maas are you using ?
[20:37] <smoser> if user-data were completely busted, then we'd know. as juju uses it when pointed at maas.
[20:37] <smoser> and that gets tested heavily
[20:38] <smoser> but if the maas cli was busted, probably would'nt know so fast, but would still seem odd.
[20:39] <drhalan> can somebody explain me what maas actually does when acquiring a node? it seems to reboot during the process
[20:40] <drhalan> i could ssh into my machine and my updated curtin script seemed to have deployed the ssh keys and packages I needed. suddenly it rebooted and everything was gone
[20:42] <garibaldi> smoser: confirmed, the test file I tried to create as part of the script is not there; and /var/lib/cloud/instance/user-data.txt is empty
[20:42] <garibaldi> is there a debug mode I can enable?
[20:42] <mup> Bug #1461256 was opened: 1.8rc1: Filter by node broken in Chromium - angular errors in java script console <oil> <MAAS:New> <https://launchpad.net/bugs/1461256>
[20:42] <mup> Bug #1461258 was opened: 1.8rc1: Deploy/Release in single node view keeps adding blank pages at the top <oil> <MAAS:New> <https://launchpad.net/bugs/1461258>
[20:43] <smoser> drhalan, yeah, thats sort of a bug
[20:43] <drhalan> what do you mean ?
[20:43] <smoser> the ephemeral environment that maas uses to run curtin in also gets your user's keys
[20:43] <smoser> so you can ssh into the install enviornment as 'ubuntu@'
[20:43] <drhalan> oh yeah
[20:43] <smoser> and not realize that you're in iscsi root
[20:44] <smoser> the sanest fix seems to be to just have the user in the ephemeral root environment be 'ephemeral' or 'admin' or something.
[20:44] <smoser> then you wouldnt think you were in
[20:45] <drhalan> that makes sense. but i sitll dont understand which file i have to modify to change the final install. i tried /etc/maas/preseed/curtin and /etc/maas/preseed/curtin_userdata
[20:45] <smoser> garibaldi, what versoin of maas ?
[20:46] <drhalan> the first one only seems to affect the install and the second one has no effect for me
[20:46] <drhalan> there is also curtin_userdata_custom which i am not sure what it is for
[20:46] <smoser>  /etc/maas/preseed/curtin_userdata is the right file.
[20:47] <smoser> what i'd suggest is running 'curtin in-target' in a late_command
[20:47] <smoser> as the easiest way to affect the install
[21:02] <garibaldi> smoser: MAAS 1.6 (the version available in trusty); I didn't know that I needed to modify curtin_userdata, I thought I just needed to pass the user_data bundle along with the "maas" cli command?
[21:03] <garibaldi> I tried putting this under late_commands in curtin_userdata but it also did not work: "  enable_test: ['sh', '-c', 'echo "this is a test" > /home/ubuntu/test.txt']"
[21:08] <smoser> garibaldi, you're right
[21:08] <smoser> the user_data comment was for drhalan
[21:11] <garibaldi> smoser: ah, okay
[21:12] <drhalan> smoser: thanks! is there a document that explains the syntax that curtain expects?
[21:13] <smoser> not really. drhalan :-(
[21:13] <drhalan> :/
[21:13] <drhalan> i just want to define packages to be installed
[21:13] <smoser> drhalan, well, personally i'd do that in user-data
[21:14] <drhalan> oh that is something else?
[21:14] <smoser> when you deploy a node, you can say "use this user-data".
[21:14] <smoser> which can be anything... and then cloud-init interprets it on boot
[21:14] <smoser> same way it does on EC2 or on azure
[21:14] <smoser> except for garibaldi is having trouble getting user-data to work :)
[21:15] <smoser> https://help.ubuntu.com/community/CloudInit <-- that describes cloud-init user-data
[21:15] <drhalan> oh but i need it for every node
[21:15] <smoser> well, you can do it every time you deploy a node.
[21:15] <smoser> i do understand the desire for central place.
[21:15] <smoser> i have to run.
[22:00] <mup> Bug #1461295 was opened: Maas is not picking up new release root-tgz automatically <oil> <MAAS:New> <https://launchpad.net/bugs/1461295>
[22:00] <mup> Bug #1461298 was opened: maas documentation should document how to configure manual DHCP for UEFI <doc> <MAAS:New> <https://launchpad.net/bugs/1461298>
[22:40] <catbus1> hi, the maas was working fine, until now it says the only cluster I have is disconnected. I don't see anything suggesting errors from /var/log/maas/maas.log to check, I have dhcp-reconfigured maas-region and cluster-controller, but that didn't help.
[22:40] <catbus1> any ideas where I can look at to find out why it's disconnected?
[22:41] <catbus1> both region and cluster are on the same server.
[22:43] <catbus1> an error keeps repeating: twisted.internet.error.CannotlistenError
[22:43] <catbus1> on pserv.log
[22:47] <catbus1> https://bugs.launchpad.net/maas/+bug/1374233
[23:10] <mup> Bug #1374233 was opened: pserv continually failing: address already in use <packaging> <MAAS:Confirmed> <https://launchpad.net/bugs/1374233>