Might be leaving orphaned EBS volumes

[coderanger]

I ended up with 51 unattached volumes after a day of testing.

[sethvargo]

@coderanger I'm not super familiar with the EC2 API/gem. Do you know if there's something special we need to do to also destroy the EBS volumes?

[JeanMertz]

@sethvargo @coderanger I am not exactly familiar with the correct syntaxes, but this is what I define in my vagrant-aws config to make sure EBS's are destroyed after termination:

block_device_mapping  = [
  {
    'DeviceName' => '/dev/sda',
    'Ebs.VolumeSize' => 100,
    'Ebs.DeleteOnTermination' => true
  }
]

Hope this helps.

[fnichol]

@JeanMertz would that config go into the server create call in fog? Thanks!

[JeanMertz]

@fnichol see: https://groups.google.com/forum/#!topic/ruby-fog/EhAPWxsXsCY

In short:

@compute.servers.create(
  flavor_id: 'm1.small',
  image_id: 'ami-83e4bcea',
  block_device_mapping: [{
    'DeviceName' => '/dev/sda1',
    'Ebs.VolumeSize' => 20,
    'Ebs.DeleteOnTermination' => true
  }]
)

[fnichol]

@JeanMertz I think that's what I needed, thank you :)

[JigarJoshi]

@fnichol ran into exact same issue with at little higher scale, this is quick fix #45 as @JeanMertz provided

[JigarJoshi]

can we merge it ?

[sethvargo]

Fixed in a18c88a

[akuzminsky]

The bug still exists

$ chef gem list | grep -e ec2 -e kitchen
...
kitchen-ec2 (3.2.0)
...
test-kitchen (2.3.3)

---
driver:
  name: ec2
  iam_profile_name: kitchen-test

After kitchen test the volume didn't get destroyed.