Block Storage Improvements

[Oogy]

• Attach on create if attached_id parameter set. Previously it would have required a second terraform apply run.
• Sleep 5 seconds prior to attaching on create. This allows new servers sufficient time for last mile bootstrap scripts to complete prior to attaching block storage.
• Cleaned up resourceVultrBlockStorageRead() to use Get()(returns the specified subscription) vs. List()(returns list of all subscriptions, requires searching).
• Handle a previously attached server(attached_id) having been destroyed in resourceVultrBlockStorageUpdate().
• Detach on delete.

[badgerwithagun]

@Oogy - with this fix, if both the block storage and vultr server are managed by terraform, if I taint the server and run an apply, will the block storage be modified in-place with the new attached_id?

[Oogy]

@badgerwithagun Yes that is correct.

[badgerwithagun]

Fantastic. This is a bit concerning:

Sleep 5 seconds prior to attaching on create. This allows new servers sufficient time for last mile bootstrap scripts to complete prior to attaching block storage.

Is it possible that this will end up being unstable in practice?

[Oogy]

@badgerwithagun This is necessary since it is not possible to live attach block storage. Without this the reboot during the attachment process will interrupt the server bootstrap scripts leaving the server in a non-working state. So with the sleep stability is improved.

[ddymko]

@badgerwithagun as @Oogy stated this due to a constraint as to how we handle deployment and attaching of block storage to instances. This shouldn't cause any instability issues.

@Oogy Tested and verified that everything is working with the ability to taint servers and have the block storage get reattached. The acceptance tests also pass locally with no issues!