maas-devel team mailing list archive

[Andres Rodriguez <andres.rodriguez@xxxxxxxxxxxxx>]

On Wednesday, October 28, 2015, John Meinel <john@xxxxxxxxxxxxxxxxx> wrote:

> Oh, and the fact that you address spaces in the CLI via their internal ID
> (1,2,3) rather than via their name (space-foo, my-magical-space) means I
> generally have to keep an internal mapping and refresh it with "maas $P
> spaces read" when I forget.
>
>
> It also runs into problems with stuff like:
>
> maas m1 subnet read 2
> Tells me that the space is "space-0" but if I want to update it:
> maas m1 subnet update 2 space=ceph_cluster
> It tells me:
>  {"space": ["Select a valid choice. That choice is not one of the
> available choices."]}
> But it doesn't tell me what the available choices are. I figured it was
> the integer thing so doing:
> maas m1 subnet update 2 space=2
> Worked, and then "subnet read 2" tells me that it is in "space":
> "ceph_cluster". But it is quite confusing that you don't set a name even
> though "read" gives you a name.
>

  This is a known usability issue on the CLI and hoping to get it fixed
after we address other higher priority and critical issues.

>
> Also, Releasing virtual machines in 1.9 seems to break. The nodes are off,
> and if you do "check now" for the power status it says it is off. But I
> get: "Failed to power off node - Node could not be powered off: Failed
> talking to node's BMC for an unknown reason." and the node gets stuck in
> "Releasing Failed".
>

Please file a bug for this!

Thanks!

>
> John
> =:->
>
>
.


>
> On Wed, Oct 28, 2015 at 3:25 PM, John Meinel <john@xxxxxxxxxxxxxxxxx
> <javascript:_e(%7B%7D,'cvml','john@xxxxxxxxxxxxxxxxx');>> wrote:
>
>> Another niggle that I've run into repeatedly is that if you decomission a
>> node and have "disk erasing" on, it has to reboot another PXE boot to
>> trigger the disc erasing. The problem is that it doesn't always cause the
>> machine to go into shutdown/reboot properly, so it just hangs waiting for
>> it to reboot so that it can be erased. And nothing on the console or logs
>> indicates that it is hung, or how to resolve the problem. The first time I
>> thought it was just taking forever to erase the disk. Some indication of
>> "waiting for machine restart" would be quite helpful.
>>
>> Also, is there any way to configure network spaces via the Web UI? 1.7.6
>> let you declare networks and their subnets, and you could put Mac Addresses
>> into one of them. But after upgrading to 1.9.0 beta1+bzr4417 it shows me
>> fabric-0 and space-0 but my second subnet is just dangling and there is
>> nothing I can edit or add from the Web UI (that I've found at least).
>>
>> John
>> =:->
>>
>>
>> On Wed, Oct 28, 2015 at 3:20 PM, John Meinel <john@xxxxxxxxxxxxxxxxx
>> <javascript:_e(%7B%7D,'cvml','john@xxxxxxxxxxxxxxxxx');>> wrote:
>>
>>> Thanks. There's no description on that PPA when adding it, and it would
>>> be nice if there were links from other ppa's (like dailybuilds) to indicate
>>> that it would be the preferred PPA to use.
>>>
>>> John
>>> =:->
>>>
>>>
>>> On Wed, Oct 28, 2015 at 3:02 PM, Dean Henrichsmeyer <dean@xxxxxxxxxxxxx
>>> <javascript:_e(%7B%7D,'cvml','dean@xxxxxxxxxxxxx');>> wrote:
>>>
>>>> You can get the latest 1.9 release (beta1) by doing add-apt-repository
>>>> ppa:maas/next
>>>>
>>>> -D
>>>>
>>>>
>>>>
>>>> On Wed, Oct 28, 2015 at 7:33 PM, John Meinel <john@xxxxxxxxxxxxxxxxx
>>>> <javascript:_e(%7B%7D,'cvml','john@xxxxxxxxxxxxxxxxx');>> wrote:
>>>>
>>>>> Looks like I'm using 1.7.6 as it was what was on the 14.04.3 ISO I
>>>>> downloaded and used "Install MAAS cluster". Is there a good way to upgrade
>>>>> that? Ideally I want to be using 1.9 for testing.
>>>>>
>>>>> Can I just add ppa:maas-maintainers/dailybuilds (it seems the default
>>>>> install of MAAS this way doesn't install software-properties-common to give
>>>>> add-apt-repository either.)
>>>>>
>>>>> John
>>>>> =:->
>>>>>
>>>>> On Wed, Oct 28, 2015 at 10:54 AM, Mike Pontillo <
>>>>> mike.pontillo@xxxxxxxxxxxxx
>>>>> <javascript:_e(%7B%7D,'cvml','mike.pontillo@xxxxxxxxxxxxx');>> wrote:
>>>>>
>>>>>> On Tue, Oct 27, 2015 at 11:27 PM, John A Meinel <
>>>>>> john.meinel@xxxxxxxxxxxxx
>>>>>> <javascript:_e(%7B%7D,'cvml','john.meinel@xxxxxxxxxxxxx');>> wrote:
>>>>>>
>>>>>>> Its a little clumsy to start a node, then click through to get to
>>>>>>> set the "power settings" to VIRSH and copy the same virsh string and set
>>>>>>> the new name of the machine. It feels like something where MAAS could be
>>>>>>> aware of a VIRSH system, rather than treating each machine that is started
>>>>>>> independently.
>>>>>>
>>>>>>
>>>>>> Which version of MAAS are you using?
>>>>>>
>>>>>> You can use the "Add Hardware > Chassis" command (in 1.8+) to add a
>>>>>> virsh system to MAAS. It will automatically discover and add to MAAS all
>>>>>> the VMs it can see (and set the boot order properly as well - in addition
>>>>>> to using the hostnames defined on the hypervisor).
>>>>>>
>>>>>> But you're correct that it woudl be better if MAAS was "more aware";
>>>>>> that is, if MAAS remembered that there was a virsh system, and could later
>>>>>> check for new machines. (This is something we'd like to take on in the
>>>>>> future.)
>>>>>>
>>>>>> I agree about the ease (or lack thereof) of setting up SSH access for
>>>>>> virsh. In MAAS 1.9, you'll notice that we prompt the user with
>>>>>> notifications indicating what they should install, if we can detect that
>>>>>> dependencies are unmet.
>>>>>>
>>>>>> Regards,
>>>>>> Mike
>>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> Mailing list: https://launchpad.net/~maas-devel
>>>>> Post to     : maas-devel@xxxxxxxxxxxxxxxxxxx
>>>>> <javascript:_e(%7B%7D,'cvml','maas-devel@xxxxxxxxxxxxxxxxxxx');>
>>>>> Unsubscribe : https://launchpad.net/~maas-devel
>>>>> More help   : https://help.launchpad.net/ListHelp
>>>>>
>>>>>
>>>>
>>>
>>
>

-- 
Andres Rodriguez
Engineering Manager, MAAS & OIL DevOps
Canonical USA, Inc.