Overview of the Life Cycle of EC2 Instances
The EC2 instance lifecycle defines how an EC2 example transitions through different states starting at launch and ending at termination.
PendingWhen the first instance is launched, it enters into the state of pending
RunningAfter the instance has been launched, it enters the running state
For each second that the instance is running, charges are incurred with a minimum of one minute.
Only EBS-backed instances can be stopped or started.
Instance store-backed instances cannot be stopped or started.
Instances can be stopped and restarted if they fail a status check, or are not running as expected.
StopAfter an instance has been stopped, it enters a stopping state and then a stopped state.
The EBS storage charges are not subject to hourly or data transfer charges.
The instance can be stopped while its root volume is being modified. Repair file system problems, update software, or modify the instance type, user information, EBS optimization attributes, or both
Volume can be detached from a stopped instance and attached to a running instance. It can also be modified and detached from the running instance before being reattached. It should be reattached using its storage device name, which is specified in the block device mapping.
StartWhen an instance is started, it enters in pending and then into running
When an instance is stopped and started, it launches on a new host
Any data on an instance volume (not root volume), would be lost, while data on the EBS volume would persist.
The Elastic IP address is also retained by EC2 instances.
If an instance has an IPv6 adress, it retains its IPv6 addresses.
The public IP address would be released if it was assigned instead of the ElasticIP address.
Instance Hibernate charges per second for each instance that is transitioned from running to stopped. The minimum charge is one minute every instance is started.
Instance hibernation signals to the operating system to perform hibernation (“suspend-to disk”), which saves the contents of the instance memory (RAM), to the EBS root volume.
The EBS root volume for the instance and any attached EBS data volumes for the instance are persistent, along with the RAM’s saved contents.
All EC2 instance store volume volumes are retained attached to the instance. However, the data on the instance store volumes is lost.
The instance is restarted and the EBS root volume is restored back to its original state. The instance retains its instance ID and any previously attached data volumes will be reattached.
After the instance has been hibernated it enters in a stopping state and then to a stopped state.
The instance is restarted and it enters the pending status. In some cases, the instance may be moved to a different host computer.
EBS root volume is restored back to its original state
Reload RAM contents
The instance is now running again.
The instance ID is retained and any previously attached data volumes are reattached.
Instance keeps private IPv4 addresses as well as any IPv6 addresses
Instance keeps its Elastic IP address
Instance releases its public IPv4 address, and would receive a new one
Hibernation prerequisitesSupported instance families – C3, C4, C5, M3, M4, M5, R3, R4, R5, & T2
RAM size for an instance must not exceed 150 GB
For bare metal instances, the instance size is not supported.
Supported AMIs must support hibernation and be HVM-compatible
Root volume type – must not be instance store but EBS volume
EBS root volume size – must be large enough for RAM contents
To ensure that sensitive content is protected in memo, the root volume of EBS must be encrypted
