bug: StreamAlert SQS Classified Logs Missing Prefix

[Ryxias]

Background

Similar to #886

During deployment on test account we noticed that the streamalert_classified_logs SQS queue is missing a prefix.

Description

All SA resources require a prefix. Missing the prefix causes trouble when there are multiple SA deployments on a single AWS Account. Terraform will detect that a duplicate SQS resource already exists during deployment and error.

Steps to Reproduce

Deploy SA to the staging account.

Desired Change

Add the prefix to the SQS queue name.

[Ryxias]

Leaving this for myself so I don't forget:
If you use python manage.py deploy --function classifier rule you have to deploy twice. The reason for this is the SQS queue is created on the first run, but its URL is generated during creation, so Terraform doesn't know that the URL is changed and fails to update the classifier's SQS_QUEUE_URL environment variable.

[Ryxias]

Confirmed on stage:
There will be data loss if the SQS queue is not replaced quickly. This is because the Classifier does not actually crash when it runs into the QueueDoesNotExist/NonExistentQueue error. It actually performs 5 retries using backoff before giving up. When it gives up, it exits gracefully, resulting in no Lambda error, but results in the Kinesis iterator popping records off nonetheless.

[Ryxias]

Changing this to a 3.0.0 release. There is no way to do this gracefully without introducing a breaking change.