AWS EC2 lifecycle diagram shared

A diagram posted on X on May 24 by user Tayaan laid out the Amazon EC2 instance lifecycle as a single flow: pending, running, stopping, stopped, reboot, shutting-down and terminated. The graphic circulated as a quick reference for engineers who need to reason about what an instance can do at each point in its life. AWS documents the core EC2 state model as pending, running, stopping, stopped, shutting-down and terminated, with reboot treated as an operation on a running instance rather than a separate billable state. (docs.aws.amazon.com) The chart matters because EC2 state changes are not only a console detail. (docs.amazonaws.cn) AWS uses those transitions in billing, health checks, automation and event notifications, which means deployment scripts, troubleshooting runbooks and Auto Scaling policies all have to account for them. ### Which EC2 states actually exist in AWS? Amazon EC2 documents six instance states: pending, running, stopping, stopped, shutting-down and terminated. AWS says an instance enters pending when it is launched or started, moves to running when ready for use, enters stopping and then stopped when it is shut down, and enters shutting-down and then terminated when it is deleted. (docs.aws.amazon.com) Boto3 and AWS API references also expose numeric codes for those states — 0 for pending, 16 for running, 32 for shutting-down, 48 for terminated, 64 for stopping and 80 for stopped. (docs.aws.amazon.com) Those codes show up in filters and waiters used by automation. ### Why does “reboot” show up on diagrams if it is not a state? AWS treats reboot as an instance action rather than a standalone lifecycle state. A reboot applies to a running instance and returns it to running after the guest operating system restarts, which is why engineers often draw it on lifecycle diagrams even though the official state list does not include it. (docs.aws.amazon.com) AWS separately notes that scheduled events can include reboot, stop or terminate actions tied to hardware issues, software updates or maintenance. That makes reboot relevant operationally even when it is not modeled as one of the six primary states. (boto3.amazonaws.com) ### What changes when an instance stops instead of terminates? AWS says a stopped instance is shut down and can be started again later, while a terminated instance has been permanently deleted and cannot be restarted. For EBS-backed instances, attached EBS volumes and network interfaces can persist across stop-start cycles, while RAM contents, instance-store data and automatically assigned public IPv4 addresses do not. (docs.aws.amazon.com) The billing difference is also explicit in AWS documentation. Instance usage is billed in the running state, not in pending, stopping, stopped, shutting-down or terminated, although AWS says hibernating instances can still incur charges while in stopping and stopped instances can still generate storage and Elastic IP costs. (docs.aws.amazon.com) ### Where do these transitions matter in production? AWS says EC2 emits “EC2 Instance State-change Notification” events with values such as pending, running, stopping, stopped, shutting-down and terminated. Those events can be consumed by monitoring and automation systems that need to trigger alerts, cleanup jobs or replacement workflows. (docs.aws.amazon.com) Auto Scaling also uses state and health as signals. AWS says that if an instance in an Auto Scaling group is in any state other than running, or if its status checks become impaired, Auto Scaling considers the instance unhealthy and replaces it. (docs.aws.amazon.com) ### What should an engineer take from the diagram? The practical use of a lifecycle diagram is to make state handling explicit. An instance that is pending is not yet ready for traffic, a stopped instance is recoverable, and a terminated one is final under AWS’s model. Those distinctions affect retries, health checks, deployment timing and incident response. (docs.amazonaws.cn) AWS’s next step for engineers who want to operationalize that model is in its own tooling: `DescribeInstanceStatus`, state-change notifications and Auto Scaling health checks all expose the same transitions that the diagram summarizes. (docs.aws.amazon.com 1) (docs.aws.amazon.com 2) (docs.aws.amazon.com 3)