EzDevInfo.com

maas interview questions

Top maas frequently asked interview questions

Why is it that aa full deployment will require a minimum of 10 servers?

I wish to try MaaS, and there's only two machine, could I install one with MaaS service, and use MaaS and juju to deploy the other bare machine with openstack nova-compute?

Thank you so much for answering.


Source: (StackOverflow)

Howto configure MAAS to separate PXE traffic and IPMI traffic on different subnets?

I am trying to set up a part of our network as a linux cluster. Since its a little educational for me, I choose using MAAS with JuJu. However there are some questions that boggle my mind and I was hoping that someone could clarify that for me.

The linux cluster I'm about to set up consists of 10 machines. Half of it Dell and the other HP. Both types of machines have a lights-out module (HP=>iLO2, Dell=>DRAC) that support IPMI on a seperate 100Mb NIC. They both support PXE on the first onboard gigabit NIC. I configured the lights out module with a static IP matching the physical layout of the racks and position height. Installing MAAS however didn't ask me on what subnet and vlan the IPMI protocol should be configured. How do I do this?

Also I want only the region controller to be able to contact the internet for package management. The other provisioned nodes should only be allowed to connect to the internet via a proxy on the region controller. So the region controller in my case should be configured with 3 subnets; 1 for internet, 1 for client protocol connectivity and 1 for cluster traffic. The region controller itself should also be a node for JuJu.

Then at last there is the node configuration that should have a sort of basic layout that can be used within JuJu. As far as I could see there is no possibility to set up cluster subnet configuration. Each machine has at least 4 NIC's that I like to assign the different subnets to; 1 for the IPMI traffic, 1 for the PXE boot traffic, 1 for the cluster traffic and 1 for the storage/client network. What I like to do is to bond all these interfaces together as one big trunk and then use VLAN's to separate the traffic before provisioning. Then when provisioning a node, MAAS should automagically configure the network interfaces as the layout suggests above.

Maybe what I'm looking for is a advanced configuration tutorial/guide for MAAS and JuJu.

Regards, Joham


Source: (StackOverflow)

Advertisements

How to configure MAAS to be able to boot virtual machines

I am running a virtual (kvm) MAAS/juju setup where most of the MAAS nodes (including MAAS master) are virtual, but some are also physical nodes. The physical Dell 1950 nodes configures automatically for boot in MAAS, so when i deploy via Juju they power on automatically. My problem lies in trying to set up boot for virtual systems. I see the posibility in the MAAS for power type. I can choose virsh. But i need to fill in information i do not know. What should i fill in for Driver and Power ID?

Anyone has experience with that?


Source: (StackOverflow)

How do I use MAAS to prepare to install OpenStack?

I have a set of servers and I would like to install OpenStack, how can I install OpenStack with Metal-as-a-Service (MAAS)?


Source: (StackOverflow)

What is the MAAS node login?

I have a problem, when I want to login in nodes. My nodes are created via MAAS dashboard, where I'm adding nodes. But when installation completes I'm asking to login and password to get into the node. But they wasn't shown or I missing something. I need to login to nodes to check clock and solve other issues.


Source: (StackOverflow)

Architecture diagram explaining the Ubuntu Cloud Infrastructure with MAAS / Openstack architecture?

I'm interested in Ubuntu Cloud Infrastructure with MAAS (based on Openstack from what I understand) but cannot find any architecture diagram which gives an overview of how MAAS works.

The case is Ubuntu Cloud Infrastructure with MAAS.

It requires at least 10 nodes (as described in Installing Ubuntu Cloud Infrastructure) and this "10 MAAS nodes architecture" would typically be a perfect candidate for an architecture diagram!

Wouldn't it be an easy/reality-based way to understand how MAAS works and what it offers?

Looking for docs, I stumbled upon a technical and up-to-date documentation with architecture diagrams(!): Revisiting OpenStack Architecture: Essex Edition


Could we imagine that an architecture diagram and/or description could be provided to easily understand how a minimal 10 nodes Ubuntu Cloud Infrastructure with MAAS installation works?


Source: (StackOverflow)

Is Ubuntu MAAS free?

Ubuntu MAAS, very cool, awesome in fact, looks like a unique tool for several jobs.

It looks free, but part of its documentation starts already with clauses that would scare anyone with interest in it:

  • Documentation is copy righted by Canonical;
  • Documentation must be used only for non-commercial purposes;
  • If documentation is distributed within the non-commercial clause you must retain copyright;

It just sounds a lot for a guide on how to install MAAS + Juju + Openstack and that scares me a bit. Why does documentation built from the community prevent me from doing anything except look at it? We produce a guide, but I cannot take it to work and look at it or show it as a possible tool that my company would be interested in using or contributing to?

My question is then: Under what license is Ubuntu MAAS distributed and what would be the reasoning for being so worried about copyrighting a simple guide like that so heavily? Is Ubuntu MAAS free?


Source: (StackOverflow)

deploying openstack on 6 servers

I'm new to openstack and i'm still learning. In 2 month i'll get 6 servers like:

  • 4 servers with dual xeon e5-2630 , 96gb ram, 2 1gb ethernet and 2 500gb hdd in raid
  • 1 DAS with 40tb of storage (10x4tb HDD)
  • 1 server with single xeon e5-2630 with 16-32 gb of ram.

I has a few questions for this setup and with openstack in general:

  • i practically has 3 dedicated compute nodes, 1 compute+storage node(1 server+DAS) and 1 node/network controller. As i read the best practice for networking server is to have this service run on all nodes for redundancy as opposed to running on a single controller where is it fails then all the ecosystem fails too.What is your oppinion? Another thing is the colocation of services as i don't have the necessary number of hardware to deploy every service on it's private hardware. I was thinking that the controller should run the controlling part service of nova, the object storage server and block storage server and the compute+storage server should run nova+object storage+block storage. What do you think of this setup?

  • for the DAS should i use raid or not because i read somewhere that swift and opesntack in general doesn't get along very well with raid?

  • should i use openstack-networking or the newer quantum?
  • i'll use this servers for hosting, storage like dropbox and virtualization. Except the latter, the first 2 will run in VMs . The thing is that i don't understand how openstack works with available resources. Does it make a pool of the available ram, cpu, storage etc? The DAS will be connected to only 1 server. Reading and writing to disk uses ram memory for cache etc;my question is : will openstack fill the ram proportionaly on all nodes when reading or writing or will it use only the ram from where DAS is attached to?This is an important problem for me. Also, only having one machine with actual storage capability will it help adding 2 more 1gb ethernet to cope with intense requests on the storage server? Or should i distribute the HDD's to all servers and install the storage services on them?
  • i didn't find a good tutorial on how to deploy openstack using maas and juju, because it seems to be an easier setup then manual install.
  • how does openstack get along with amd opteron 6300 series? I'm asking this because there is a possibility that i could change from intel to amd, the latter being much cheaper so i can get for the same ammount of money twice and something the number of cores (e5-2630 being a 6 core with ht, and from what i understand i should disable ht)

    To be continued when i will be confronted by other problems.

Thank you for your help!


Source: (StackOverflow)

Step by Step Install of MAAS and JUJU

I am working on understanding the pieces that I am missing in being able to deploy Juju across the other MAAS nodes. I don't know If I have a step out of place, or missing a few.

The MaaS server handles the DHCP and DNS. The router (which is at 10.0.0.0) has DHCP turned off. Any assistance is greatly appreciated. When I am at the end I will either get a 409 error, or arbitrary pick tools 1.16.0 error. It is worth mentioning that local, and aws works fine. Hopefully with all of these steps spelled out it will help someone else along the way too.

Updated Nov 3 2013 In this scenario for the POC, there are three physical servers. One is a HP DL 140 with 2 cpu 8gb ram, the second server is a DELL Power Edge 860 Server 1 cpu w 4 cores and 8 gb ram, and the third is a rackable systems sp5000psl with two cpus and 8gb of ram. They are all able to be on their own seperate network with access to the internet or can be on the first network. (second network is 10.0.0.0, and the first is 192.168.1.0)

Steps

Setting Up MAAS and JUJU - 12.04 LTS Clean install SSH only from the package selection during install

sudo apt-get install software-properties-common python-software-properties
sudo add-apt-repository ppa:maas-maintainers/stable      
sudo add-apt-repository ppa:juju/stable   
sudo apt-get update     
sudo apt-get dist-upgrade
sudo reboot     
sudo apt-get install maas maas-dns maas-dhcp    
sudo ufw disable    
sudo reboot

- edit /etc/dhcp/dhcpd.conf

authoritive
subnet 10.0.0.0 netmask 255.255.255.0 {
next-server 10.0.0.2;
filename "pxelinux.0";
}

sudo maas createsuperuser
sudo maas-import-pxe-files

Login to MAAS http://10.x.x.x/MAAS


  • cluster controller configuration for eth0
  • manage dhcp and dns
  • IP 10.0.0.2
  • subnet 255.255.255.0
  • broadcast 10.0.0.0
  • routerip 10.0.0.1
  • ip low 10.0.0.5
  • ip high 10.0.0.180

Commissioning default and distro is set at 12.04

default domain is at local

sudo maas-cli login maas http://10.x.x.x/MAAS/api/1.0 api-key
ssh-keygen -t rsa -b 2048 - enter - no password  - cat id_rsa.pub and enter key into MAAS ssh 
sudo maas-cli maas nodes accept-all (interestingly enough I only get back [] when executing this )

PXE one machine, accept and commision, start and deploy.

sudo apt-get install juju-core juju-local

MAAS config:

maas:
    type: maas
    maas-server: '://10.x.x.x:80/MAAS'
    maas-oauth: 'MAAS_API_KEY'
    admin-secret: 'nothing'
    default-series: 'precise'

juju switch maas
sudo juju bootstrap --show-log

Source: (StackOverflow)

Is there a way for MaaS to Commission a Raspberry Pi 3B?

I have a small lab (literally) of 3 RBP-3B units with SATA and SD drives attached running Ubuntu 16.04 LTS. I have installed software on them to create a Ceph cluster but its tricky and pretty fragile.

RP3B Ceph Cluster

Ultimately I would like to administer this "Worlds Smallest Cloud Storage Cluster" concept with Juju and make it the storage backend for an Intel NUC vBlock I won from the OSIC contest at the Openstack summit. Unfortunately I cant find a way to PXE boot the Raspberry Pi 3B, and chip support for this feature is on the horizon but currently not present.

I want to use MaaS to bootstrap the OS and feed these little guys into Juju deployments since its so easy from there to manage everything, but I am having trouble finding a working solution. I know that Juju can add machines after the OS is installed but its not apparent how to do that either, and if I set Juju up to use existing machines I don't know if I can still point it to MaaS controlled systems as well to create a heterogenous mix of gear without breaking functionality.

Any thoughts?


Source: (StackOverflow)

MaaS minimum requirements with juju-jitsu?

I've browsed through so many different sites and found so much contradictory information. As I am getting tired of this and do belive this question affects many other users, so I would like to collect the "once and for all times" answer. Unfortunately, the documentation on MaaS and Juju is ... well, not the best, sorry to say that.

What are the minimum system requirements for setting up a MaaS cluster, which is going to be orchestrated with juju-jitsu?

Do they need to have the exact system specifications or can I just combine different hardware?

What are the minimum requirements for the master machine?

E.g. "You need at least 8GB of RAM, a dual core CPU with at least 3.0 GHz."

How many machines to I need to deploy MaaS on?

I've read six machines, nine machines, and so on. I clearly want to know: "You need one for the Master and e.g. five nodes."

Do I need to attach as many NICs (network interface cards) to my master machine as there are nodes, or can I simply attach two NICs and a switch?

One NIC for connecting to the internet, one for handling the MaaS tasks, connected to a switch, which connects my nodes to the master?

Is Juju now ready for local deployment?

The last time I experimented with juju and had to reboot my machine, the services orchestrated by juju were gone. This was an issue I also found on the official juju site. Unfortunately, as mentioned above, the documentation is not the best, so I could not find the necessary info on that again. So:

Can I use juju on a local environment or will a reboot break my setup?


Source: (StackOverflow)

Can I manage ZFS with Landscape and MAAS?

I have 5 different servers. I am very interested in installing ZFS across all of them. I also am interested in using JuJu, MAAS, and Landscape to automate the installation of ZFS, and the management of each server. For instance I do not want to manually install ZFS on each machine.

I have 4 questions:

  1. Can ZFS be used smoothly with Landscape, JuJu, and MAAS?

  2. Is it possible to have Landscape manage the ZFS installation process?

  3. Landscape supports two types of storage. One is iSCSI, and the other is Ceph storage. Can any of these be configured to install ZFS during the storage allocation phase?

  4. How is raid/mirroring set up using JuJu, Landscape, and MAAS?

  5. Can I setup ZFS mirroring from the Landscape console? .

Referenced Programs/Utilities: (ZFS (Z File System), JuJu (Ubuntu Cloud), MAAS (Metal as a Service), and Landscape (See Page)).


Source: (StackOverflow)

Juju not seeing the MaaS slaves... at least not after some time?

At our company are doing a project where we want to start doing Juju, MaaS and OpenStack. All in all we are satisfied, but we have one major problem that is setting us back. We haven't seemed to find out where the problem lies - with juju or MaaS.

We have set up a MaaS master with ubuntu 12.10 - and we use 12.04 on the MaaS slaves. We have one bootstrap slave and 6 slaves for OpenStack...

In the beginning everything seems to work fine - getting all slaves in ready state in MaaS and deploying with juju also seems to work as expected. We deploy a whole openstack environment. But at some point (this has been experienced 3-4 times after redoing it all over again) Juju will not see new MaaS slaves and when we want to add units or other juju charms on new nodes those will not be seen by Juju. They are stated as Ready in MaaS, but Juju does not see them, and when we use juju deploy the charm just will say pending and machine will say pending, not picking the ready nodes from maas. I can see that juju version in the slaves is 0.6 but i can see there is a newer version. Will it work with the newer version?


Source: (StackOverflow)

Allow loop mounting files inside LXC containers?

I'm trying to set up a MaaS server inside an LXC container.

When I import the PXE files, it needs to be able to mount loop devices.

I have set the following options in the container config file to allow loop mounting, but I'm missing something.

lxc.cgroup.devices.allow = b 7:* rwm
lxc.cgroup.devices.allow = c 10:237 rwm

I get the following error because the script cannot loop mount a file:

mount: cannot mount block device /dev/loop0 read-only
Wed, 13 Nov 2013 07:26:41 +0000: failed to mount /var/lib/maas/ephemeral/precise/ephemeral/i386/20131010/disk.img
Traceback (most recent call last):
  File "/usr/sbin/maas-import-ephemerals", line 26, in <module>
    main(args)
  File "/usr/lib/python2.7/dist-packages/provisioningserver/import_images/ephemerals_script.py", line 428, in main
    target.sync(source, args.path)
  File "/usr/lib/python2.7/dist-packages/simplestreams/mirrors/__init__.py", line 85, in sync
    return self.sync_index(reader, path, data, content)
  File "/usr/lib/python2.7/dist-packages/simplestreams/mirrors/__init__.py", line 237, in sync_index
    self.sync(reader, path=epath)
  File "/usr/lib/python2.7/dist-packages/simplestreams/mirrors/__init__.py", line 83, in sync
    return self.sync_products(reader, path, data, content)
  File "/usr/lib/python2.7/dist-packages/simplestreams/mirrors/__init__.py", line 315, in sync_products
    self.insert_item(item, src, target, pgree, ipath_cs)
  File "/usr/lib/python2.7/dist-packages/provisioningserver/import_images/ephemerals_script.py", line 251, in insert_item
    self.extract_item(path, flat)
  File "/usr/lib/python2.7/dist-packages/provisioningserver/import_images/ephemerals_script.py", line 295, in extract_item
    tarball, target_dir, temp_location=self._simplestreams_path())
  File "/usr/lib/python2.7/dist-packages/provisioningserver/import_images/ephemerals_script.py", line 124, in extract_image_tarball
    call_uec2roottar(image, os.path.join(target_dir, 'dist-root.tar.gz'))
  File "/usr/lib/python2.7/dist-packages/provisioningserver/import_images/ephemerals_script.py", line 97, in call_uec2roottar
    subprocess.check_call(["uec2roottar"] + list(args))
  File "/usr/lib/python2.7/subprocess.py", line 540, in check_call
    raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '[u'uec2roottar', u'/var/lib/maas/ephemeral/precise/ephemeral/i386/20131010/disk.img', u'/var/lib/maas/ephemeral/precise/ephemeral/i386/20131010/dist-root.tar.gz']' returned non-zero exit status 1
root@maaslxc2:~# mount /dev/loop0 /mnt
mount: block device /dev/loop0 is write-protected, mounting read-only
mount: cannot mount block device /dev/loop0 read-only

So, what do I need to change in the container's configuration to allow it to mount loop devices? It appears that this is not just a MaaS problem, but a restriction that would cause trouble for anything (not just MaaS) that needs to loop mount a file in a LXC container.


Source: (StackOverflow)

MaaS Enlistment through 12.04 installation CD fails

I am trying to enroll my nodes in my MaaS cluster. I am doing this by booting them from the 12.04 installation CD, and choosing the MAAS installation option.

I'm only trying to set up one node right now. When i do the enrollment process, the node shutsdown, and I see 0 nodes on the MAAS homepage of my master node.

I tried adding the node from the MAAS web gui, when I do this, I get an error on the page that says:

Unable to create Node: Unknown problem encountered with the Cobbler server.

In the cobbler.log is the following:

Tue May  8 20:43:47 2012 - DEBUG | REMOTE maas authorization result: True; user(?)
Tue May  8 20:43:47 2012 - INFO | find_items; ['system']
Tue May  8 20:43:47 2012 - INFO | Exception occured: <class 'cobbler.cexceptions.CX'>
Tue May  8 20:43:47 2012 - INFO | Exception value: 'dns-name duplicated: node1'
Tue May  8 20:43:47 2012 - INFO | Exception Info:
File "/usr/lib/python2.7/dist-packages/cobbler/remote.py", line 1944, in _dispatch
return method_handle(*params)
File "/usr/lib/python2.7/dist-packages/cobbler/remote.py", line 910, in xapi_object_edit
self.modify_system(handle, 'modify_interface', imods, token)
File "/usr/lib/python2.7/dist-packages/cobbler/remote.py", line 827, in modify_system
return self.modify_item("system",object_id,attribute,arg,token)
File "/usr/lib/python2.7/dist-packages/cobbler/remote.py", line 820, in modify_item
return method(arg)
File "/usr/lib/python2.7/dist-packages/cobbler/item_system.py", line 644, in modify_interface
if field == "dnsname"             : self.set_dns_name(value, interface)
File "/usr/lib/python2.7/dist-packages/cobbler/item_system.py", line 288, in set_dns_name
raise CX("dns-name duplicated: %s" % dns_name)

It seems like there is a chicken and egg game being played between dnsmasq's DNS server, and the cobbler's hostname database -- this is just a guess on my part.

Here is the contents of dnsmasq.leases: the MAC addr of node1 is 00:03:47:22:1c:e2

dbuttric@sid:/var/lib/misc$ more dnsmasq.leases 
1336531172 00:03:47:22:1c:e2 192.168.5.119 * *
1336530695 00:90:27:32:3b:53 192.168.5.148 node3 *
1336530676 00:90:27:18:f2:0b 192.168.5.242 node2 *

How can I delete hostnames from cobbler, and retry my node adding activities?

Here is more: I can successfully register a node and the MAAS server shows that it has 1 node registered in it.

5 minutes later, I go back to the MAAS web gui, and it reports 0 nodes. I have not looked at error logs yet, but what can cause this to happen?

Should I just try re-installing the master node?


Source: (StackOverflow)