Re: maas pxe rack / region controller

[Blake Rouse]

Erwan,

The diagram on https://maas.io/how-it-works shows which services run where.
The rack does not proxy any API calls, all API calls go directly to the
region.

Thanks,
Blake Rouse

On Tue, Apr 10, 2018 at 8:51 PM, Erwan MAS <er...@mas.nom.fr> wrote:

> Hi ,
>
> The rack seem to proxy some MAAS API call  but not all . correct ? bug ?
>
> Why only the region controller provide a proxy service ?
>
>
> --
> Erwan MAS
>
> --
> Maas-devel mailing list
> Maas-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/maas-devel
>
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel

Re: Network issues after upgrading to 2.3

[Blake Rouse]

Nick,

I would review your settings in both regiond.conf and rackd.conf to make
sure that they have the maas_url configured correctly. You should review
the interface configuration of your controller on the "Controllers" tab to
make sure the networking configuration looks correct.

If you need to make any changes to the regiond.conf and rackd.conf, be sure
to restart maas-regiond and maas-rackd.

Hopefully that helps!

Thanks,
Blake Rouse
Technical Lead Software Engineer
MAAS Team
Canonical Ltd.

On Mon, Mar 19, 2018 at 3:57 AM, Nick Tan <nick@spookfish.com> wrote:

> Hi again,
>
>
>
> I created a squashfs image with a root password so I could get in and look
> at the cloud-init logs.  This entry caught my eye:
>
>
>
> DataSourceMAAS.py[CRITICAL]: Giving up on md from [‘http://
> [::1]/MAAS/metadata/2012-03-01/meta-data/instance-id’] after 126 seconds
>
>
>
> Digging deeper, the cloud-init.log shows this at the start:
>
>
>
> 2018-03-19 07:00:11,115 - util.py[DEBUG]: Cloud-init v. 17.2 running
> 'init-local' at Mon, 19 Mar 2018 07:00:11 +. Up 12.77 seconds.
>
> 2018-03-19 07:00:11,116 - main.py[INFO]: wrote cloud-config data from
> cloud-config-url='http://10.18.39.2/MAAS/metadata/latest/
> by-id/8taaws/?op=get_preseed' to /etc/cloud/cloud.cfg.d/91_
> kernel_cmdline_url.cfg
>
>
>
> At this point it is using the correct IP address of the MAAS server
> (10.18.39.2).  Next, I had a look at the 
> /etc/cloud/cloud.cfg.d/91_kernel_cmdline_url.cfg
> file and it’s replaced the IP address with [::1]
>
>
>
> #cloud-config
>
> apt:
>
>   preserve_sources_list: false
>
>   primary:
>
>   - arches: [default]
>
> uri: http://archive.ubuntu.com/ubuntu
>
>   proxy: http://[::1]:8000/
>
>   security:
>
>   - arches: [default]
>
> uri: http://archive.ubuntu.com/ubuntu
>
> apt_proxy: http://[::1]:8000/
>
> datasource:
>
>   MAAS: {consumer_key: xx, metadata_url: 'http://
> [::1]/MAAS/metadata/',
>
> token_key: xx, token_secret: xx}
>
> manage_etc_hosts: true
>
> reporting:
>
>   maas: {consumer_key: xx, endpoint: 'http://[::1]/MAAS/metadata/
> status/8taaws',
>
> token_key: xx, token_secret: xx,
>
> type: webhook}
>
> rsyslog:
>
>   remotes: {maas: '::1:514'}
>
> system_info:
>
>   package_mirrors:
>
>   - arches: [i386, amd64]
>
> failsafe: {primary: 'http://archive.ubuntu.com/ubuntu', security: '
> http://security.ubuntu.com/ubuntu'}
>
> search:
>
>   primary: ['http://archive.ubuntu.com/ubuntu']
>
>   security: ['http://archive.ubuntu.com/ubuntu']
>
>   - arches: [default]
>
> failsafe: {primary: 'http://ports.ubuntu.com/ubuntu-ports', security:
> 'http://ports.ubuntu.com/ubuntu-ports'}
>
> search:
>
>   primary: ['http://ports.ubuntu.com/ubuntu-ports']
>
>   security: ['http://ports.ubuntu.com/ubuntu-ports']
>
>
>
>
>
> I’m thinking that somewhere in my MAAS server settings it’s configured to
> send a IPv6 address?  I can’t find anything in the web GUI, is there a CLI
> command I can run to find out and fix it?
>
>
>
> Thanks,
>
> Nick
>
>
>
> *Nick Tan,* Senior Systems Administrator
>
> *Spookfish Global Operations Pty Ltd*
>
> m: +61 413 134 846
>
> w: spookfish.com
>
> *From: *Maas-devel <maas-devel-boun...@lists.ubuntu.com> on behalf of
> Nick Tan <nick@spookfish.com>
> *Date: *Monday, 19 March 2018 at 12:19 pm
> *To: *"maas-devel@lists.ubuntu.com" <maas-devel@lists.ubuntu.com>
> *Subject: *Network issues after upgrading to 2.3
>
>
>
> This sender failed our fraud detection checks and may not
> be who they appear to be. Learn about spoofing
> <http://aka.ms/LearnAboutSpoofing>
>
> Feedback <http://aka.ms/SafetyTipsFeedback>
>
> Hi all,
>
>
>
> I had a functioning MAAS 2.0.0 install which I recently upgraded to
> 2.3.0.  Since then, I can’t commission or deploy machines.  It looks like
> I’m hitting the same problem as https://bugs.launchpad.net/
> maas/+bug/1738782 ie: the node boots, but stalls at “MAAS: searching for
> network data from DataSourceMAAS” before continuing to a login prompt and
> doing nothing.
>
>
>
> Nothing on the network side has changed - just the upgrade of MAAS.  All
> the MAAS components are installed on the same server running Ubuntu 16.04
> server.
>
>
>
> Has anyone else seen this issue and have any ideas on how to resolve it?
> Is it feasible to downgrade back to 2.0.0?
>
>
>
> Thanks,
>
> Nick
>
>
>
> *Nick Tan,* Senior Sys

Re: how to better debug

[Blake Rouse]

James,

I have asked for more information on [2] and [3]. If you could provide
answers to those questions we can work though the issues you are having and
resolve them.

Thanks,
Blake

On Tue, Oct 17, 2017 at 9:24 PM, James Beedy  wrote:

> Hello,
>
> I've been hitting some errors (seems to be on both 2.2 and 2.3 - I've
> updated to next and am still experiencing the issues, also updated curtin
> to daily per curtin bug below) that I have no idea how to debug or make
> progress on, or sense of. Hoping someone here has an idea.
>
> I have a few racks to enlist over the next few weeks. Just starting out
> with a few servers here, and having quite the difficulty getting things
> going. I'm hoping we can identify what the issue is, such that I can avoid
> it the rest of the journey.
>
> Here is my start -> https://imgur.com/a/Dremk
>
> Initially, all of the nodes commissioned and deployed. From what I can
> tell, I've hit [1], not sure how to recover from it, or if its even what is
> causing the issues I'm experiencing.
>
> I've created a bug here [3] for the "failure to exit rescue mode".
>
> The nodes that fail to commission also fail the 
> "00-maas-06-get-fruid-api-data"
> test, although there is no log attached, or output from the failed test,
> so I don't really know what I'm looking for here (although this is a huge
> improvement from previous versions of maas! - yea! - I still filed a bug on
> this [2]), I'm just a bit baffled about why this has happened to my nodes,
> and how I can recover them, and avoid this from happening in the future.
>
> Any insight here would be greatly appreciated
>
> thanks
>
>
> 1. https://bugs.launchpad.net/maas/+bug/1708052
> 2. https://bugs.launchpad.net/maas/+bug/1724402
> 3. https://bugs.launchpad.net/maas/+bug/1724401
>
>
>
>
> --
> Maas-devel mailing list
> Maas-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/maas-devel
>
>
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel

Re: maas installed from snap rack contoller issue

[Blake Rouse]

On Tue, Sep 12, 2017 at 8:54 PM, Daniel Bidwell  wrote:

> I installed it with:
>
> snap install --devmode --stable maas
>
> and then did
>
> maas init --mode all
>
> snap list shows
>
> maas  2.2.1-6078-g2a6d96e-snap  207   maas   devmode
>
> maas status returns
>
> bind9RUNNING   pid 2705, uptime 5:39:51
> dhcpdSTOPPED   Not started
> dhcpd6   STOPPED   Not started
> ntp  RUNNING   pid 2826, uptime 5:39:43
> postgresql   RUNNING   pid 2704, uptime 5:39:51
> proxyRUNNING   pid 2843, uptime 5:39:39
> rackdRUNNING   pid 2702, uptime 5:39:51
> regiond:regiond-0RUNNING   pid 2699, uptime 5:39:51
> regiond:regiond-1RUNNING   pid 2698, uptime 5:39:51
> regiond:regiond-2RUNNING   pid 2700, uptime 5:39:51
> regiond:regiond-3RUNNING   pid 2697, uptime 5:39:51
> tgt  RUNNING   pid 2736, uptime 5:39:49
>
> maas config returns
>
> Mode: all
> Settings:
> maas_url=http://maasmaster.andrews.edu/MAAS


This is the URL the rack controller and the deploying machines will use to
talk to the region controller. You should make sure that from the machine
the snap is running on that domain is resolvable to this machine.

Also unless you are running your own reverse proxy to point to the MAAS
running at 5240 this will not resolve. You should update this url to be:

sudo maas config --maas-url http://maasmaster.andrews.edu:5240/MAAS


If that doesn't work then I would try to reconfigure this with an IP
address at first to rule out that the DNS lookup is causing you the issue.

sudo maas config --maas-url http://{ip-address}:5240/MAAS

Hopefully that helps.

- Blake


>
> I have just removed all of the apt-get maas related packages and am
> rebooting.  I am still pretty new to snap packages.
>
> On Tue, 2017-09-12 at 16:35 -0500, Andres Rodriguez wrote:
> > Hi Daniel,
> >
> > If you are using MAAS from the snap, I'll make the assumption that
> > you are running MAAS 2.3 from the latest track and either, the beta
> > or the edge channel. As such, can you please share with us how have
> > you installed and configured your snap ?
> >
> > Also, you can check the status and config of the snap with the
> > following two commands:
> >
> > sudo maas status
> > sudo maas config
> >
> > Please note that if the MAAS snap is installed, you should not have
> > MAAS installed from the Ubuntu Archives (or PPA's).
> >
> > Also, for documentation you can also refer to [1].
> >
> > [1]: https://docs.ubuntu.com/maas/devel/en/installconfig-snap-install
> >
> > On Tue, Sep 12, 2017 at 4:18 PM, Daniel Bidwell 
> > wrote:
> > > I installed maas from snap.  I wish to provide dhcp to fabric-0.
> > > It
> > > wants me to select a "Rack Controller" but the list of available
> > > rack
> > > controllers is empty.  How do I go about getting access to the
> > > default
> > > rack controller for my main maas machine?
> > >
> > > I have tried doing "apt-get install maas-rack-controller" and maas-
> > > rack
> > > register ..., but no luck.
> > > --
> > > Daniel Bidwell 
> > >
> > >
> > > --
> > > Maas-devel mailing list
> > > Maas-devel@lists.ubuntu.com
> > > Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman
> > > /listinfo/maas-devel
> > >
> >
> >
> > --
> > Andres Rodriguez
> > Engineering Manager, MAAS
> > Canonical USA, Inc.
> --
> Daniel Bidwell 
>
>
> --
> Maas-devel mailing list
> Maas-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/maas-devel
>
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel

Re: MAAS 2.2 production ready?

[Blake Rouse]

Florian,

Many people run MAAS today in production including the 2.2.2 release of
MAAS. You are running into bug lp:1702703 [1] where the region controller
will not run correctly without the "/bin/maas-rack" present. This was
recently fixed in the MAAS 2.3 version of MAAS that we are currently
developing.

A quick fix for this bug is to install the rack controller on the region
controller as well. There is no issues with doing this even if you plan on
not using that rack controller.

sudo apt install maas-rack-controller

Installing that package will make the /bin/maas-rack present for the region
controller to do what it needs to do.

Hopefully this solves the issue for you. Let me know if you need anymore
help.

Thanks,
Blake Rouse
Lead Technical Software Engineer
MAAS Team - Canonical

[1] https://bugs.launchpad.net/maas/+bug/1702703

On Mon, Aug 28, 2017 at 8:44 AM, Flo <florian.engelm...@gmail.com> wrote:

> Hi,
>
> we are currently testing MAAS 2.2.2
> (2.2.2-6099-g8751f91-0ubuntu1~16.04.1) in a distrubuted setup (one
> region-controller and multiple rack-controllers). After two weeks of
> testing we are still not able to deploy hosts. Bugs we are facing:
>
> installing a region-controller without a "local" rack-controller causes:
> missing:
> /usr/sbin/maas-rack
>
> 2017-08-23 11:23:36 provisioningserver.utils.services: [info]
> observe-mdns: FileNotFoundError: [Errno 2] No such file or directory:
> b'/bin/maas-rack'
>
> The RPC communication fails very often because
> http://maasserver/MAAS/rpc/ is not responding
>
> It looks like the region-controller is not responding quite often -
> causing cloud-init to fail (timeouts)
>
> Anybody running MAAS 2.2.2 in production?
>
> All the best,
> Florian
>
> --
> Maas-devel mailing list
> Maas-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/maas-devel
>
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel

Re: Re: Regarding alarm and metrics in MAAS

[Blake Rouse]

On Wed, Aug 9, 2017 at 11:20 PM, sachidananda sahu <sachi...@gmail.com>
wrote:

> -- Forwarded message --
> From: "sachidananda sahu" <sachi...@gmail.com>
> Date: 10-Aug-2017 8:09 AM
> Subject: Re: Regarding alarm and metrics in MAAS
> To: "Lee Trager" <lee.tra...@canonical.com>
> Cc:
>
> Hello Lee, Thanks for your quick response .
> I am looking towards maintaining statistics at overall maas level(Total no
> of node deployed, total no of node failed in deployment , total no of node
> failed in specific event and their cause etc) so that the user can get a
> overall idea about maas region controller.
>

This is a feature request we have had before. Its not something currently
on the roadmap, but if you want to track the bug you will be alerted when
it is implemented.

https://bugs.launchpad.net/maas/+bug/1445574


>
>
>
> On 10-Aug-2017 5:59 AM, "Lee Trager" <lee.tra...@canonical.com> wrote:
>
>> Could you be more specific about the data you are looking for?
>>
>> `maas  node-script-result read  current-installation`
>> will give statistics about the installation. You can get a list of all
>> machine system_ids with `maas  machines read` To get event log
>> data you can use `maas  events query id=
>>
>>
>> On Wed, Aug 9, 2017 at 5:01 PM, sachidananda sahu <sachi...@gmail.com>
>> wrote:
>>
>>> Thanks for the detailed info, but all I can see which is only at node
>>> level, is there any way to get overall counts, failure counts and
>>>  statistics of all nodes ?
>>>
>>> On 10-Aug-2017 3:26 AM, "Lee Trager" <lee.tra...@canonical.com> wrote:
>>>
>>>> Anything you can do in the UI you can also do in the using the API/CLI.
>>>> To download the current installation log you can use the command
>>>>
>>>> maas  node-script-result download 
>>>> current-installation
>>>>
>>>> Similarly you can do the same for commissioning  or testing results
>>>> maas  node-script-result download 
>>>> current-commissioning
>>>> maas  node-script-result download  current-testing
>>>>
>>>> For commissioning and testing results you may want to download a tar
>>>> with all results instead of the text
>>>> maas  node-script-result download  current-testing
>>>> filetype=tar.xz
>>>>
>>>> MAAS stores metadata about the result status as status, runtime, and
>>>> when installation/commissioning/testing started
>>>> maas  node-script-result read  current-installation
>>>>
>>>> MAAS also stores historic results which can be viewed with
>>>> maas  node-script-results read 
>>>>
>>>> Hope that helps!
>>>>
>>>> Lee
>>>>
>>>> On Tue, Aug 8, 2017 at 7:29 PM, sachidananda sahu <sachi...@gmail.com>
>>>> wrote:
>>>>
>>>>> Yes Blake, the GUI provides only the node level events(Not at maas
>>>>> region level and not overall about all the nodes), but is there any
>>>>> directory where it is stored ?
>>>>>
>>>>> Jeff, you are right.This was my questions intention.
>>>>>
>>>>> It will be better to get the deployment log of each node in some
>>>>> specific directory(may be in /etc/maas/logs/node*
>>>>>
>>>>> In addition to that it will be nice to have the logs or mantaining
>>>>>  statistics at overall maas level(Total no of node deployed, total no of
>>>>> node failed in deployment , total no of node failed in specific event and
>>>>> their cause etc) so that the user can get a overall idea about maas region
>>>>> controller.
>>>>>
>>>>> On 09-Aug-2017 1:09 AM, "Jeffrey Lane" <jeffrey.l...@canonical.com>
>>>>> wrote:
>>>>>
>>>>> I think the question is something like "is there a directory on the
>>>>> MAAS Server Filesystem (like /var/log/maas/) where I can find text
>>>>> file logs for each node OUTSIDE maas.log and the web UI".
>>>>>
>>>>> I had a similar thought in that it would be nice at least if the UI
>>>>> would allow me to download a note's deployment logs, rather than
>>>>> forcing me to copy/paste several hundred lines of text to a file.  For
>>>>> example, click on the "Deployment" tab and have a simple

Re: Development Summary - June 8th

[Blake Rouse]

We are currently working on machine network configuration and overall
network configuration in the viscera part of the library that gets exposed
through the client Facade.

Example:

client = await login('http://localhost:5240/MAAS', 'admin', 'password')
fabric = await client.fabrics.get_default()
vlan = fabric.vlans.get_default()
vlan.mtu = 9000
await vlan.save()

machines = await client.machines.list()
for machine in machines:
boot_interface = machine.boot_interface
boot_interface.vlan = vlan
await boot_interface.save()

Creating bonds, vlan interfaces, and bridges will come as well.

- Blake Rouse


On Mon, Jun 12, 2017 at 3:32 AM, Ante Karamatić <
ante.karama...@canonical.com> wrote:

> On Thu, Jun 8, 2017 at 9:20 PM Christian Robottom Reis <k...@canonical.com>
> wrote:
>
>> > With the continuous improvement of the new MAAS Python Library
>> > (python-libmaas), we have focused our efforts on the following
>> improvements
>> > the past week:
>>
>> Ante, what are they key features missing from libmaas for the node
>> config step for the Foundation work (and is there more we'd like to
>> have)? I'd like to see if they might make 2.3.
>>
>
> libmaas has bones and viscera modules. Bones is low level lib, and it
> might support all that is needed (but it's also much easier to just use
> maas CLI then). Viscera module, which is a high level lib, should at least
> support enlisting machines and network/disk configuration. And this is
> because these are the most repetitive operations that, also take too long
> to execute using CLI. Fabrics, vlans, tags, spaces, etc are things one
> usually sets once, maybe iterate over, but in essence those are not time
> consuming operations.
>
> Andres, Blake and I have already touched on this few weeks back. I think
> we know how to approach the problem within MAAS itself that would solve all
> of the machine configuration in one go; using cloud-config syntax as input
> and output syntax for node configuration. API at that point wouldn't be a
> blocker and in fact wouldn't require a lot of work to support that feature.
> In future it could provide high level abstraction for configuration, but
> that just bonus.
>
> --
> Ante Karamatić
> ante.karama...@canonical.com
> Canonical
>
> --
> Maas-devel mailing list
> Maas-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/maas-devel
>
>
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel

Re: MAAS to ignore boot request from specific mac addresses

[Blake Rouse]

I just tested this with MAAS. I added a device with the MAC address of one
of machines that normally PXE boots from MAAS. I then powered it on, it PXE
booted from MAAS then loaded the pxelinux.0 and MAAS told it to boot from
its local disk.

Adding the ability to change this behavior would be a good improvement.
Possible options could be:

* Ignore
* Boot local
* Poweroff

I filed a bug about this here:
https://bugs.launchpad.net/maas/+bug/1686412

On Wed, Apr 26, 2017 at 9:09 AM, Mark Shuttleworth  wrote:

>
> Seems to me that it would make sense to be able to add the node as a
> device, in which case MAAS should avoid PXE responses altogether.
> Alternatively, add the ability to tell MAAS that this node should always
> just boot from its disk if it PXEs.
>
> Mark
>
>
> On 25/04/17 14:08, Mike Pontillo wrote:
>
> Hi Eswar,
>
>If the node is deployed by MAAS and PXE boots, it will be sent a PXE
> configuration that instructs it to boot from the local disk. If the PXE
> server is unavailable and the BIOS falls back to booting from the local
> drive, that should work, too. So you should be safe either way.
>
>If the node is NOT managed by MAAS but running on an
> otherwise-MAAS-managed subnet, I would consider setting its BIOS to NOT
> boot from the network. If you can't do that, I would consider adding an
> iptables rule that would match the machine's MAC address and filter PXE
> requests.
>
> Regards,
> Mike
>
> On Tue, Apr 25, 2017 at 10:14 AM, ESWAR RAO  wrote:
>
>> Hi All,
>>
>> Please let me know how can we disable a compute node (mac address) from
>> PXE booting by MAAS.
>> MAAS should ignore a boot request from specific mac addresses.
>>
>> I don't think this can be done by using MAAS CLI:
>> https://docs.ubuntu.com/maas/1.9/en/maascli
>>
>>
>> Please suggest me how can i achieve this :
>>
>> (1) disable the mac in dhcp conf of maas ??
>> (2) any filters like below ??
>> https://github.com/opencord/maas/blob/7b1f18c47329d08fb84d8a
>> 704d60a493b2b360be/automation/README.md
>>
>> Thanks
>> Eswar Rao
>>
>>
>> --
>> Maas-devel mailing list
>> Maas-devel@lists.ubuntu.com
>> Modify settings or unsubscribe at: https://lists.ubuntu.com/mailm
>> an/listinfo/maas-devel
>>
>>
>
>
>
>
> --
> Maas-devel mailing list
> Maas-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/maas-devel
>
>
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel

Re: MAAS 2.1.5 bug fix release now available!

[Blake Rouse]

Any time you make a change to the machine in the BIOS you need to
recommission so MAAS discovers this new information. Changing an IP address
is an Ubuntu software configuration and does not require re-commissioning.
Also you can just run commissioning again from ready state instead of
deleting and adding the machine again.

On Thu, Apr 20, 2017 at 3:51 AM, Patrizio Bassi <patrizio.ba...@gmail.com>
wrote:

> Dear Blake,
>
> i may have understood what it may have happened: when i started my machine
> (with 3 SAS disks) had storage configured as 1 RAID1 and 1 Jbod. The jbod
> disk was shown as sda first device (BIOS related). I discovered,
> commissioned and deployed the machine with the issues i spotted.
>
> Than i went back to the System BIOS, i changed the disk config to RAID1
> and RAID0 (instead of Jbod) so the storage bacame sda (for raid1) and sdb
> for single disk.
>
> But when deploying it looks like MAAS kept the old config "in cache" or
> alive. To fix this i had to delete the machine from inventory, rediscover
> from scratch, commission and redeploy and everything was fine.
>
> Is this the designed behaviour? In general even when i need to do small
> changes, such as configure a network interface of a deployed machine
> (deployed->mark broken->change ip->mark fixed->ready->deploy), most of time
> not passing through the commissioning phase makes the deploy fails.
>
> Thanks,
> Patrizio
>
>
> 2017-04-18 15:19 GMT+02:00 Blake Rouse <blake.ro...@canonical.com>:
>
>> Patrizo,
>>
>> Are you sure that when the ephemeral environment is booted that vda and
>> vdb are not switched by the kernel? It could be possible that vda showed as
>> vdb and vdb showed as vda on the booted kernel. That will cause MAAS to
>> pick the 25GB as the boot disk during commissioning.
>>
>> Best way to check is the block device output in the commissioning output
>> section of the Web UI.
>>
>> Thanks,
>> Blake
>>
>> On Tue, Apr 18, 2017 at 4:10 AM, Patrizio Bassi <patrizio.ba...@gmail.com
>> > wrote:
>>
>>> Hi Andres,
>>>
>>> i'm not sure it is an EFI issue as i could reproduce on my virtual
>>> machines provided by a standalone kvm too.
>>>
>>> How to reproduce:
>>> create a VM with 2 disks, 25G (vda) and 10G (vdb).
>>>
>>>
>>>   
>>>   
>>>   
>>>   
>>> 
>>> 
>>>   
>>>   
>>>   
>>>   
>>> 
>>>
>>>
>>> in the Machine summary i correctly see:
>>> Storage 37.6GB over 2 disks
>>>
>>> but it detects as vda the 10G disk, and installs there. On boot,
>>> instead, kvm tries to start from first, failing because maas installed on
>>> secondary hd.
>>>
>>> If not in maas (i checked the fix and and you just sort by name, which
>>> should be safe), can be daily ephemeral images recently changed the
>>> something?
>>>
>>> Patrizio
>>>
>>>
>>> 2017-04-14 13:52 GMT+02:00 Andres Rodriguez <
>>> andres.rodrig...@canonical.com>:
>>>
>>>> Hi Patrizio,
>>>>
>>>> It sounds like you are deploying a machine with BIOS on Legacy mode.
>>>>
>>>> Such being the case, are you deploying to the correct disk? The BIOS
>>>> will, by default, boot for the configured device. So, you either to
>>>> configure your MAAS machine to install into such disk, or you need to
>>>> change the priority of disk to the one you are installing in the BIOS.
>>>>
>>>> Note that this should be a problem with EFI provided that it is smart
>>>> enough to figure you which disk was installed.
>>>>
>>>>
>>>> On Fri, Apr 14, 2017 at 7:06 AM Patrizio Bassi <
>>>> patrizio.ba...@gmail.com> wrote:
>>>>
>>>>> Andres,
>>>>>
>>>>> it looks like i'm now affected by https://bugs.launchpad.net/
>>>>> maas/+bug/1671605 which should be addressed by this release (and i
>>>>> didn't have before).
>>>>>
>>>>> After deploy (succesfully completed) on reboot all machines with more
>>>>> than 1 disk fail to boot.
>>>>>
>>>>>
>>>>> 2017-03-29 15:38 GMT+02:00 Patrizio Bassi <patrizio.ba...@gmail.com>:
>>>>>
>>>>>> It works now, i'm upgrading.
>>>>>>
&g

Re: MAAS 2.1.5 bug fix release now available!

[Blake Rouse]

Patrizo,

Are you sure that when the ephemeral environment is booted that vda and vdb
are not switched by the kernel? It could be possible that vda showed as vdb
and vdb showed as vda on the booted kernel. That will cause MAAS to pick
the 25GB as the boot disk during commissioning.

Best way to check is the block device output in the commissioning output
section of the Web UI.

Thanks,
Blake

On Tue, Apr 18, 2017 at 4:10 AM, Patrizio Bassi 
wrote:

> Hi Andres,
>
> i'm not sure it is an EFI issue as i could reproduce on my virtual
> machines provided by a standalone kvm too.
>
> How to reproduce:
> create a VM with 2 disks, 25G (vda) and 10G (vdb).
>
>
>   
>   
>   
>   
> 
> 
>   
>   
>   
>   
> 
>
>
> in the Machine summary i correctly see:
> Storage 37.6GB over 2 disks
>
> but it detects as vda the 10G disk, and installs there. On boot, instead,
> kvm tries to start from first, failing because maas installed on secondary
> hd.
>
> If not in maas (i checked the fix and and you just sort by name, which
> should be safe), can be daily ephemeral images recently changed the
> something?
>
> Patrizio
>
>
> 2017-04-14 13:52 GMT+02:00 Andres Rodriguez  com>:
>
>> Hi Patrizio,
>>
>> It sounds like you are deploying a machine with BIOS on Legacy mode.
>>
>> Such being the case, are you deploying to the correct disk? The BIOS
>> will, by default, boot for the configured device. So, you either to
>> configure your MAAS machine to install into such disk, or you need to
>> change the priority of disk to the one you are installing in the BIOS.
>>
>> Note that this should be a problem with EFI provided that it is smart
>> enough to figure you which disk was installed.
>>
>>
>> On Fri, Apr 14, 2017 at 7:06 AM Patrizio Bassi 
>> wrote:
>>
>>> Andres,
>>>
>>> it looks like i'm now affected by https://bugs.launchpad.net/
>>> maas/+bug/1671605 which should be addressed by this release (and i
>>> didn't have before).
>>>
>>> After deploy (succesfully completed) on reboot all machines with more
>>> than 1 disk fail to boot.
>>>
>>>
>>> 2017-03-29 15:38 GMT+02:00 Patrizio Bassi :
>>>
 It works now, i'm upgrading.

 Thank you

 Patrizio


 2017-03-29 14:53 GMT+02:00 Andres Rodriguez <
 andres.rodrig...@canonical.com>:

> Hi Patrizio,
>
> I've triggered a rebuild of the packages in PPA to force the
> re-publishing of the deb files. It should be available now. Otherwise,
> please do let us know to further investigate.
>
> Thanks.
>
> On Wed, Mar 29, 2017 at 2:11 AM, Patrizio Bassi <
> patrizio.ba...@gmail.com> wrote:
>
>> Hi Andres,
>>
>> i can't find in the stable ppa, i still get 2.1.4 version.
>> Am i missing something?
>>
>> Regards
>>
>> Patrizio
>>
>> 2017-03-28 0:28 GMT+02:00 Andres Rodriguez <
>> andres.rodrig...@canonical.com>:
>>
>>> The MAAS team is happy to announce that  MAAS 2.1.5 has been
>>> released. This is a new upstream bug fix release that addresses the
>>> following issues:
>>>
>>>- LP: #1671891Fix deployment with AUTO IPs set as the
>>>gateway link.
>>>- LP: #1671048Fix crash when getting the default gateways
>>>for a machine.
>>>- LP: #1673916Fix regression that causes traceback in
>>>rack-based mDNS monitoring service.
>>>- LP: #1673724Due to a potential regression in lsblk, try to
>>>ensure list of block devices is ordered correctly.
>>>
>>> *Availability for Xenial Users*
>>> MAAS 2.1.5 has been made available in:
>>>
>>>- *ppa:*maas*/stable*
>>>
>>> NOTE: MAAS 2.1.5 will soon be SRU'd into Ubuntu Xenial.
>>>
>>> *Filing bugs*
>>> Users can file bugs in [1], please use a prefix for the subject,
>>> such as [2.1].
>>>
>>> [1] https://bugs.launchpad.net/maas/+filebug
>>>
>>> --
>>> Andres Rodriguez
>>> Engineering Manager, MAAS
>>> Canonical USA, Inc.
>>>
>>> --
>>> Maas-devel mailing list
>>> Maas-devel@lists.ubuntu.com
>>> Modify settings or unsubscribe at: https://lists.ubuntu.com/mailm
>>> an/listinfo/maas-devel
>>>
>>>
>>
>
>
> --
> Andres Rodriguez
> Engineering Manager, MAAS
> Canonical USA, Inc.
>



>>>
>>>
>>> --
>>>
>>> Patrizio Bassi
>>> www.patriziobassi.it
>>> http://piazzadelpopolo.patriziobassi.it
>>>
>> --
>> Andres Rodriguez
>> Engineering Manager, MAAS
>> Canonical USA, Inc.
>>
>
>
>
> --
>
> Patrizio Bassi
> www.patriziobassi.it
> http://piazzadelpopolo.patriziobassi.it
>
> --
> Maas-devel mailing list
> Maas-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/maas-devel
>
>
-- 
Maas-devel mailing list

MAAS 2.2.0 RC 2 Released!

[Blake Rouse]

The MAAS team is happy to announce that the new upstream (development)
release has now been released.

*Availability*
MAAS 2.2.0 RC2  is currently available in:

*ppa:maas/next*
Bugs fixed in this release

   - LP: #1376483   AttributeError: 'Port' object has no attribute 'socket'
   - LP: #1381129   AttributeError: 'NoneType' object has no attribute
   'stopListening'
   - LP: #1672735   [2.2] TimeoutError resolving DNS
   - LP: #1674148   hook for storage: storage already attached - Juju using
   already mounted disk
   - LP: #1676167   [2.2b1] Device BMC passwords seem to have been lost in
   upgrade
   - LP: #1677336   [2.2b5] ntp test failed: can't find ntpq
   - LP: #1677936   [2.2] Failure while handling notification to 'event'
   - LP: #1678323   [2.1.5] Power parameters stopped working for
   UCSC-C220-M4L Cisco system
   - LP: #1680277   [2.2rc1] rackd — 75% connected to region controllers.
   - LP: #1680278   [2.2rc1] Failed to process message for node
   - LP: #1680876   [2.2, UI] Cannot update storage tags over the UI
   - LP: #1681383   [2.2 UI] If I cancel editing a package repository, the
   header bar is no longer accesible
   - LP: #1681757   [2.2.0rc2, UI, Machine actions] All the actions'
   confirmation buttons (Cancel/[Action] in the machine details page are
   placed on the left
   - LP: #1681856   [2.2.0rc2, UI, Machine details] The Edit button is
   placed on the bottom right instead of the top right in the summary page
   - LP: #1682255   [2.2] Removing a custom image yields misplaced
   confirmation message
   - LP: #1682290   [2.2] Cannot delete custom image
   - LP: #1433012   TestAcquireNodeForm.test_storage_with_named_constraints
   is flaky
   - LP: #1616417   [2.0] The IP of a Machine's BMC is not listed on the
   subnet details page
   - LP: #1657471   [2.1] 400 BAD REQUEST: Unrecognised signature:
   method=GET op=is_registered
   - LP: #1661203   [2.1+] Cannot create a bridge on an interface that also
   has VLAN interfaces
   - LP: #1672220   [2.x] User gets no control over @ RRset in DNS domains
   - LP: #1675915   [2.2, UI] Script selection doesn't show scroll bar
   - LP: #1677933   [2.2] Adding a pod should add machines with their given
   names
   - LP: #1679431   [2.2] If a single commissioning script times out, all
   scripts will enter "Timed out" state.
   - LP: #1681379   [UI, 2.2.0rc2] Global: actions dropdown is overlapped
   by secondary navigation
   - LP: #1681389   [UI, 2.2.0rc2] Add node listing zone / DNS form,
   controls are next to form elements
   - LP: #1681399   [UI, 2.2.0rc2] Adding fabric / VLAN / Space
   - LP: #1681407   [UI, 2.2.0rc2] Footer link 'View release notes' is
   broken and points to 404 page
   - LP: #1682139   AttributeError: 'ACKDatagram' object has no attribute
   'mode'
   - LP: #1682152   [2.2, UI] Deploy action options are misaligned
   - LP: #1533822   should pass overlayroot_cfgdisk=disabled when booting
   enlistment, commissioning, install
   - LP: #1635061   [2.1] Event log "load 30 more days" button is for
   loading 1 additional day
   - LP: #1671517   [UI, 2.1] Wrong ordering of ranges
   - LP: #1675919   [2.2] Commissioning and Testing tabs should have icon
   while running or when done to show something happened.
   - LP: #1676882   [2.2b4] RSD Password is shown in clear text under the
   'Power' tab
   - LP: #1676962   [2.2] Commissioning script output should preserve
   whitespace
   - LP: #1681378   [UI, 2.2.0rc2] Node details tags alignment issue
   - LP: #1681386   [UI, 2.2.0rc2] Pod overview-Add Pod - Empty state is
   not displayed all users see is an empty table
   - LP: #1681387   [UI, 2.2.0rc2] Node details events - view full history
   button should be a link
   - LP: #1682099   [2.2.0rc2, UI, Images] In custom source the "Advanced
   options button changes position when I click Show/Hide
   - LP: #1555759   [2.0a1] If no VLAN has DHCP enabled, MAAS should notify
   the user.

-- 
Blake Rouse
Technical Lead Software Engineer, MAAS
Canonical USA, Inc.
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel

Re: Nodes could not be commissioned

[Blake Rouse]

MAAS installs the SSH key that you provided MAAS. There is no way to login
to the VM directly, you must use SSH. If you add the correct SSH key to
MAAS for you user account, the following command should work.

ssh ubuntu@{ip-of-vm}

Hope that helps,
Blake

On Thu, Apr 6, 2017 at 11:30 AM, Amira Hamila  wrote:

> Thank you for your replies  !
> I switched to kvm and everything works fine. I deployed some nodes and
> theyre working right except I can't get through them (needs a login and
> pass ?) I tried with ssh but the "permission denied" message keeps popping
> out ...
> What's the default login and password ?
>
> 2017-04-05 19:55 GMT+01:00 Peter Matulis :
>
>>
>> On Wed, Apr 5, 2017 at 2:07 PM, Brendan Donegan <
>> brendan.done...@canonical.com> wrote:
>>
>>>
>>> Yes, there is a 'virsh' power type, but the instructions are a bit hard
>>> to find - I think they have gone missing from the current docs. They are
>>> here: https://docs.ubuntu.com/maas/1.9/en/nodes#virtual-machine-nodes
>>>
>>>
>> For adding a KVM-backed node, I went to the 'Nodes' section in the menu
>> and chose 'Add nodes'. On the resulting page [1] you will find:
>>
>> _KVM guest nodes_
>> KVM-backed nodes are common and so a little extra guidance is provided
>> here...
>>
>> You can also use the search facility, albeit a rough prototype
>> implementation at this time.
>>
>> [1]: https://docs.ubuntu.com/maas/2.1/en/installconfig-add-nodes
>>
>> Peter.
>>
>>
>>
>>
>
> --
> Maas-devel mailing list
> Maas-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/maas-devel
>
>
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel

Re: Nodes could not be commissioned

[Blake Rouse]

MAAS currently doesn't have a driver for VirtualBox. I recommend using KVM
(libvirt) for a better experience with MAAS. To use VirtualBox you should
set the power type of the machine to "manual". The once you start
commissioning turn the VM's power on manually (MAAS will turn it off).
Deployment will work the same way, start deployment and then manually turn
the VM on.

Hope that helps.

- Blake

On Wed, Apr 5, 2017 at 1:12 PM, Amira Hamila  wrote:

> Hello,
>
> I decided to test MaaS with some vm's using virtualbox, the thing is, the
> nodes are "ready" but could not be commissioned. I'm guessing that it's
> because I put the wrong power type which is manual, but I don't know how to
> do otherwise.
>
> Can anyone help me with this?
>
>
> My MaaS version is 2.1
>
> --
> Maas-devel mailing list
> Maas-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/maas-devel
>
>
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel

Re: maas custom images with juju

[Blake Rouse]

Matt,

At the moment Juju does not support deploying a charm with a custom image
in MAAS. It is required to use the provided Ubuntu images with Juju. It is
possible to modify the preseed of the deploying machine which will allow
you to make some changes to the system before Juju takes over [1]. Again
this is also not recommended with Juju, Juju wants to be running on a clean
Ubuntu image.

[1]  https://docs.ubuntu.com/maas/1.9/en/development/preseeds

Note: This is documentation for MAAS 1.9, but the same applies for 2.0+. I
have filed a bug for the missing documentation after 1.9.
https://bugs.launchpad.net/maas/+bug/1668315

Thanks,
Blake

On Mon, Feb 27, 2017 at 4:53 AM, Matt Rae  wrote:

> Hi, is there is a suggested way to utilize custom MAAS images when
> deploying services with juju?
>
> Realizing in most cases it is suggested to use a charm to install packages
> rather than updating the images.
>
> Thanks!
>
> Matt
>
> --
> Maas-devel mailing list
> Maas-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/maas-devel
>
>
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel

Re: dhcp on a rack controller

[Blake Rouse]

On the rack controller details page it does expose the managed VLAN for the
rack controller.

On Mon, Dec 5, 2016 at 6:10 AM, Christian Robottom Reis <k...@canonical.com>
wrote:

> On Fri, Dec 02, 2016 at 09:30:27AM -0500, Blake Rouse wrote:
> > Identify the VLAN that MAAS discovered on the rack controller in your
> test
> > rack. Visit that VLAN's details page on the "Subnets" tab. Use the action
> > button at the top right header of the page to select "Provide dhcp".
> > Configure your DHCP options and click "Provide DHCP". You now have a
> > running DHCP server in your test rack.
>
> Blake, are we making UI changes to make this process more obvious, in
> particular for users that migrated from 1.x? Perhaps highlighting in the
> rack
> controller view a link into the manage VLAN page?
> --
> Christian Robottom Reis [+1] 612 888 4935 | [+55] 16 99112
> 6430
> Canonical VP Storage & Automation http://launchpad.net/~kiko
>
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel

Re: dhcp on a rack controller

[Blake Rouse]

Identify the VLAN that MAAS discovered on the rack controller in your test
rack. Visit that VLAN's details page on the "Subnets" tab. Use the action
button at the top right header of the page to select "Provide dhcp".
Configure your DHCP options and click "Provide DHCP". You now have a
running DHCP server in your test rack.

On Fri, Dec 2, 2016 at 9:06 AM, Daniel Bidwell  wrote:

> I have two clusters of servers in two racks.  My test cluster runs on a
> rack controller, but the rack controller is not running the dhcp
> server.  The machines on the test rack are on a different vlan and a
> different fabric with a different network address space.  How do I
> trigger the rack controller to run the dhcp server?
>
> The subnet on this fabric has a dynamic address range.
> --
> Daniel Bidwell 
>
>
> --
> Maas-devel mailing list
> Maas-devel@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/maas-devel
>
-- 
Maas-devel mailing list
Maas-devel@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/maas-devel