Adding multiple machines with the same instance-ID does not fail

[Andrew Wilkins]

See lp:1266765

If one does "juju add-machine ssh:<address>" multiple times with the same address, then multiple machines are added with the same instance ID. There should be a one-to-one mapping between machines and instance IDs.

[John A Meinel (jameinel)]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 2014-01-09 5:55, Andrew Wilkins wrote:
> Public bug reported:
>
> See lp:1266765
>
> If one does "juju add-machine ssh:<address>" multiple times with
> the same address, then multiple machines are added with the same
> instance ID. There should be a one-to-one mapping between machines
> and instance IDs.
>
> ** Affects: juju-core Importance: High Status: Triaged
>

I just tried that with juju trunk and it gave me "this instance has
already been provisioned". I was trying it with locally created LXC
machines, (which can't actually be manually provisioned yet because it
tries to install LXC in an LXC, which we know fails.)

So *maybe* it works if the instance finishes coming up, but when I was
trying to fix juju-1.17 add-machine in a 1.16 environment, it was
refusing to let me add the same machine twice.

So I think we need a reproducible test case/script so that we can be
sure this really is an active bug.

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.13 (Cygwin)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iEYEARECAAYFAlLOQSIACgkQJdeBCYSNAANm2wCg0cwaN7Q31WQYMvxCdkq8oU86
L9gAn37/u2OWIFwIUr+wPN8e4PzBpfHT
=1Vnd
-----END PGP SIGNATURE-----

[Andrew Wilkins (axwalk)]

I'll clean up my LXC environment and try to reproduce.

> which can't actually be manually provisioned yet because it tries to install LXC in an LXC, which we know fails

lxc isn't installed upfront anymore, you can manually provision LXC now

[Andrew Wilkins (axwalk)]

Confirmed. I did the following:
 1. bootstrapped a manual environment
 2. created two containers (manually, with lxc-create)
 3. copied `which juju` and ~/.juju to each container
 4. in each container, issued "juju add-machine ssh:127.0.0.1"

The status is now (ignore gaps in machines; I reused an env):

machines:
  "0":
    agent-state: started
    agent-version: 1.17.1.1
    dns-name: 10.0.3.1
    instance-id: 'manual:'
    series: saucy
    hardware: arch=amd64 cpu-cores=4 mem=15952M
  "5":
    agent-state: pending
    instance-id: manual:127.0.0.1
    instance-state: missing
    series: precise
    hardware: arch=amd64 cpu-cores=4 mem=15952M
  "6":
    agent-state: pending
    instance-id: manual:127.0.0.1
    instance-state: missing
    series: precise
    hardware: arch=amd64 cpu-cores=4 mem=15952M

[Anastasia (anastasia-macmood)]

Juju 2 does not use LXC but LXD. If this behavior exists with LXD, please file a bug against "juju".

For "juju-core", we will not address this as it is not a Critical bug.