bootstrap without a jenv destroys an existing environment

[Jay R. Wren]

Easy to reproduce:

1. create a new environment
2. bootstrap it
3. remove the environments/name.jenv
4. bootstrap it again
5. instead of getting an "exists" message, you get a destroying message.

Use case for why this is bad:

Team Member A creates an environment for production and shares it with Team Member B (or all team for maintenance), by sharing the environment section of environments.yaml file, and keys and certificate files if needed.

Team Member B adds the environment to .juju/environments.yaml file and put the keys and certificates into files if needed.

Team Member B runs bootstrap, not realizing it already exists.

The production environment is destroyed.

Example output attached.

[Francesco Banconi (frankban)]

Using 1.21-alpha2 the environment is no longer destroyed, but a new one is still created and no "environment already exists" message is displayed. See below.

~ % juju version
1.21-alpha2-utopic-amd64
~ % juju bootstrap -e ec2 --upload-tools
Bootstrapping environment "ec2"
Starting new instance for initial state server
Launching instance
 - i-68b9b32a
Building tools to upload (1.21-alpha2.1-trusty-amd64)
Installing Juju agent on bootstrap instance
Waiting for address
Attempting to connect to ec2-54-73-27-206.eu-west-1.compute.amazonaws.com:22
Attempting to connect to ip-10-89-138-68.eu-west-1.compute.internal:22
Attempting to connect to 54.73.27.206:22
Attempting to connect to 10.89.138.68:22
Warning: Permanently added 'ec2-54-73-27-206.eu-west-1.compute.amazonaws.com,54.73.27.206' (ECDSA) to the list of known hosts.
Logging to /var/log/cloud-init-output.log on remote host
Running apt-get update
Running apt-get upgrade
Installing package: curl
Installing package: cpu-checker
Installing package: bridge-utils
Installing package: rsyslog-gnutls
Bootstrapping Juju machine agent
Starting Juju machine agent (jujud-machine-0)
Bootstrap complete
~ % juju status -e ec2
environment: ec2
machines:
  "0":
    agent-state: started
    agent-version: 1.21-alpha2.1
    dns-name: ec2-54-73-27-206.eu-west-1.compute.amazonaws.com
    instance-id: i-68b9b32a
    instance-state: running
    series: trusty
    hardware: arch=amd64 cpu-cores=1 cpu-power=100 mem=1740M root-disk=8192M
    state-server-member-status: has-vote
services: {}
~ % rm ~/.juju/environments/ec2.jenv
~ % juju bootstrap -e ec2 --upload-tools
Bootstrapping environment "ec2"
Starting new instance for initial state server
Launching instance
 - i-601c3e85
Building tools to upload (1.21-alpha2.1-trusty-amd64)
Installing Juju agent on bootstrap instance
Waiting for address
Attempting to connect to ec2-54-170-170-10.eu-west-1.compute.amazonaws.com:22
Attempting to connect to ip-10-72-159-254.eu-west-1.compute.internal:22
Attempting to connect to 54.170.170.10:22
Attempting to connect to 10.72.159.254:22
Warning: Permanently added 'ec2-54-170-170-10.eu-west-1.compute.amazonaws.com,54.170.170.10' (ECDSA) to the list of known hosts.
Logging to /var/log/cloud-init-output.log on remote host
Running apt-get update
Running apt-get upgrade
Installing package: curl
Installing package: cpu-checker
Installing package: bridge-utils
Installing package: rsyslog-gnutls
Bootstrapping Juju machine agent
Starting Juju machine agent (jujud-machine-0)
Bootstrap complete
WARNING expected one instance, got 2

At this point two EC2 instances are up and running.

[Richard Harding (rharding)]

Francesco, I'd suggest we raise this as a new bug on juju-core since this has landed and is due to be part of the release. I suggest we track this and we can raise the question on if this is a potential release blocker for the 1.21 release.

[Francesco Banconi (frankban)]

Food point Rick: filed bug 1386703.

[Francesco Banconi (frankban)]

Good point even...