public service announcement to digital ocean users

This is not explicitly Ansible related (we’ll be making a code change because of it) but is worth sharing as I know there are a lot of Digital Ocean users out there who may have not seen this report:

https://digitalocean.com/blog_posts/transparency-regarding-data-security

It turns out digital ocean has not been erasing past customer disk space before giving customers new virtual instances, which makes passing an optional “scrub” flag on deletion important – something we weren’t aware Digital Ocean was not doing on machine creation and did not have as part of our community module (like fog).

Here they are wrongly placing blame on various client libraries for not passing a flag that should have a sensible default in the API, but that’s beside the point – scratch space has not been cleaned up on machine creation in the past, and regardless of provider, your data with this particular cloud provider is not safe.

Note: we do not host any of our infrastructure on Digital Ocean and I’m seriously considering whether we keep modules in core for this. Here was a past issue: https://www.digitalocean.com/blog_posts/avoid-duplicate-ssh-host-keys

I’ve applied a change in Ansible at 99616d0c80b0b11211a1ed8fba160af7f79e9409 and hope to see a change in dopy as well.

Meanwhile, I’d welcome feedback about whether this warrants continued inclusion in Ansible, as I do view inclusion on the “cloud” sidebar as something close to advocacy, and simply cannot fathom the conscious design decisions made in the above blog post and especially the denial of the breach.

Thoughts welcome.