Upgrade RabbitMQ instance type from mq.t3.micro to mq.m7g.medium

[jim-counter]

Summary

• Upgrades the Auto Drive production RabbitMQ broker from mq.t3.micro to mq.m7g.medium in response to an AWS retirement notice
• mq.m7g.medium is a Graviton3-based instance type and AWS's recommended replacement for RabbitMQ brokers

Why this change is safe to apply

Data preservation

Amazon MQ for RabbitMQ uses EBS-backed storage. During an instance type change, AWS detaches the EBS volume from the old instance and reattaches it to the new one — durable queue definitions and persistent messages are preserved across the change. See AWS: Editing a broker and Amazon MQ storage.

Expected behaviour during apply

• The broker will restart as part of the instance replacement — a brief period of unavailability is expected
• This is a SINGLE_INSTANCE deployment with apply_immediately = true, so the change takes effect immediately (no maintenance window delay)
• Active client connections will drop and need to reconnect

Application reconnection

The Auto Drive backend has a keepalive/reconnect mechanism (keepAliveInterval) that will detect the dropped connection and re-establish it automatically, typically within 60 seconds.

Queue declarations

Queues in the Auto Drive app are declared with assertQueue() on each publish, so they are automatically re-created after the broker restarts. No manual intervention is required.

References

• Amazon MQ instance types for RabbitMQ
• Editing a broker
• RabbitMQ storage on Amazon MQ

Test plan

• Run ./resources/terraform/manage.sh auto-drive-production plan and confirm only the broker instance type changes (1 resource, 0 destroy)
• Apply during a low-traffic window
• Confirm broker comes back healthy in the AWS Console (Amazon MQ → Brokers → status: RUNNING)
• Confirm Auto Drive backend reconnects and resumes processing

🤖 Generated with Claude Code

Closes #531.

[EmilFattakhov]

Makes sense 👍