Managing Running Instances - Start, Stop, and Reboot Operations
Introduction
Managing instance lifecycle operations is fundamental to maintaining healthy cloud infrastructure. Whether you need to apply updates, troubleshoot issues, or optimize resource usage, understanding the differences between start, stop, soft reboot, and hard reboot operations ensures you make the right choice for each situation.
This guide explains each operation, when to use them, and best practices for managing your OpenStack instances without risking data loss or unexpected downtime.
Understanding Instance States
Before performing any instance operation, you need to understand the basic instance states in OpenStack:
Active - Instance is running and accessible. CPU cycles are being consumed, and resources are fully allocated.
Shutoff - Instance is powered off but still exists. Resources like floating IPs and volumes remain attached, but no compute resources are consumed.
Reboot - Temporary state during restart operations. The instance transitions through this state before returning to Active.
Error - Instance encountered a problem and cannot operate normally. This state requires investigation and manual intervention.
Each operation you perform transitions the instance between these states. The method you choose determines how the transition happens and what risks are involved.
Stop Operation (Graceful Shutdown)
The stop operation performs a graceful shutdown of your instance. This is equivalent to running a clean shutdown command from within the operating system.
What Happens During Stop
When you stop an instance, OpenStack sends an ACPI shutdown signal to the guest operating system. The OS receives this signal and initiates its normal shutdown procedure, closing applications gracefully, flushing disk buffers, and shutting down services in the proper order.
After the OS completes its shutdown, the instance enters the Shutoff state. Attached volumes remain connected, networking configuration persists, and your floating IP addresses stay associated with the instance.
When to Use Stop
Use the stop operation when you need to power down an instance temporarily but plan to restart it later. Common scenarios include scheduled maintenance windows, cost optimization by shutting down development instances overnight, or preparing for infrastructure changes.
Stop is the safest way to power down an instance because it allows the operating system to shut down cleanly. Applications have time to save state, databases can close connections properly, and file systems flush all pending writes to disk.
How to Manage Instance Lifecycle in Horizon Dashboard
Horizon dashboard provides a visual interface for managing instance lifecycle operations. Most users find the dashboard approach easier than CLI commands, especially when managing multiple instances or viewing instance status at a glance.
Stopping an Instance
- Log into the Horizon dashboard
- Navigate to Project > Compute > Instances
- Locate the instance you want to stop in the instances list
- Click the dropdown arrow on the right side of the instance row
- Select Shut Off Instance from the dropdown menu
- Confirm the action in the confirmation dialog
The instance transitions to the Shutoff state. The status column updates to show "Shutoff" when the operation completes.
Starting an Instance
- Log into the Horizon dashboard
- Navigate to Project > Compute > Instances
- Locate the stopped instance in the instances list
- Click the dropdown arrow on the right side of the instance row
- Select Start Instance from the dropdown menu
The instance begins booting and transitions to the Active state. The status column shows "Active" once the boot sequence completes.
Soft Reboot an Instance
- Log into the Horizon dashboard
- Navigate to Project > Compute > Instances
- Locate the instance you want to reboot
- Click the dropdown arrow on the right side of the instance row
- Select Soft Reboot Instance from the dropdown menu
- Confirm the action in the confirmation dialog
The instance performs a graceful reboot. The status column briefly shows "Reboot" during the restart process, then returns to "Active" when the reboot completes.
Hard Reboot an Instance
- Log into the Horizon dashboard
- Navigate to Project > Compute > Instances
- Locate the instance you need to hard reboot
- Click the dropdown arrow on the right side of the instance row
- Select Hard Reboot Instance from the dropdown menu
- Confirm the action in the confirmation dialog
The instance performs an immediate power cycle without waiting for the OS to shut down gracefully. Use this only when soft reboot fails or when the instance is unresponsive.
How to Manage Instance Lifecycle Using OpenStack CLI
For users who prefer command-line tools or need to automate instance operations, the OpenStack CLI provides direct access to all lifecycle management functions.
Stopping an Instance
Via OpenStack CLI:
1openstack server stop <instance-name-or-id>
Important Considerations
A graceful shutdown relies on the guest operating system responding correctly to the ACPI signal. If the OS is frozen, unresponsive, or has ACPI support disabled, the stop operation may not complete. In these cases, you will need to use a hard reboot or contact support.
Stopped instances do not consume compute resources, but you continue to be billed for attached volumes, snapshots, and reserved IP addresses.
Start Operation
The start operation powers on a stopped instance and returns it to the Active state.
What Happens During Start
When you start an instance, OpenStack allocates compute resources on the hypervisor, attaches the instance's root disk and any additional volumes, restores the networking configuration, and powers on the virtual machine. The guest operating system then boots using its standard boot sequence.
The instance goes through the same boot process it would during initial creation, running init scripts, starting services, and mounting file systems.
When to Use Start
Use start whenever you need to bring a stopped instance back online. This is the counterpart to the stop operation and should be used after planned shutdowns.
Starting an instance is a safe operation with minimal risk. The instance boots cleanly from its last known good state.
Starting an Instance
Via OpenStack CLI:
1openstack server start <instance-name-or-id>
Important Considerations
Boot time depends on your operating system and the services configured to start automatically. Linux instances typically boot within 30 to 60 seconds, while Windows instances may take several minutes.
If an instance fails to start, check the console log for boot errors. Hardware failures, corrupted file systems, or misconfigured services can prevent successful startup.
Soft Reboot Operation
The soft reboot operation restarts an instance gracefully, similar to running the reboot command from within the operating system.
What Happens During Soft Reboot
OpenStack sends a reboot signal to the guest operating system. The OS initiates its normal shutdown procedure, stopping services cleanly and flushing disk buffers. Once shutdown completes, the instance immediately powers back on and boots through its standard boot sequence.
The instance remains in the Reboot state during this process, then returns to Active once the boot completes. Network connections are preserved, and attached volumes remain connected throughout the reboot.
When to Use Soft Reboot
Use soft reboot when you need to restart an instance to apply configuration changes, clear memory issues, or resolve application problems. Soft reboot is the preferred method for routine restarts because it gives the OS time to shut down cleanly.
Common scenarios include applying kernel updates that require a reboot, restarting services that cannot be reloaded, or recovering from application crashes that have left the system in an unstable state.
Soft Reboot an Instance
Via OpenStack CLI:
1openstack server reboot --soft <instance-name-or-id>
Important Considerations
A soft reboot typically completes within 1 to 2 minutes for Linux instances. Windows instances may take longer depending on the number of services configured.
If a soft reboot hangs or the instance becomes unresponsive during the reboot process, you may need to use a hard reboot to recover.
Hard Reboot Operation
The hard reboot operation immediately powers off and restarts the instance without waiting for the operating system to shut down gracefully. This is equivalent to pressing the physical reset button on a server.
What Happens During Hard Reboot
OpenStack forces the instance to power off at the hypervisor level, regardless of the guest OS state. The instance is then immediately powered back on. No shutdown signal is sent to the operating system, applications are not given time to close cleanly, and disk buffers may not be flushed.
The instance transitions directly from its current state to powered off, then boots from a cold start.
When to Use Hard Reboot
Use hard reboot only when a soft reboot fails or when the instance is completely unresponsive. Hard reboot is the emergency option for instances that have frozen, crashed, or stopped responding to ACPI signals.
Specific scenarios include kernel panics that prevent graceful shutdown, hung processes that cannot be killed, or situations where the instance console shows the system is frozen.
Hard Reboot an Instance
Via OpenStack CLI:
1openstack server reboot --hard <instance-name-or-id>
Important Considerations
Hard reboot carries risk of data loss or file system corruption because the OS does not have time to flush pending writes or close files properly. Modern file systems like ext4 and XFS are journaled and handle unexpected shutdowns relatively well, but the risk still exists.
Use hard reboot as a last resort. Always try soft reboot first unless the instance is completely unresponsive.
After a hard reboot, check system logs and run file system checks if you suspect corruption. Monitor application behavior to ensure services started correctly.
Best Practices for Instance Operations
Follow these best practices to minimize risk when managing instance lifecycle operations:
Always prefer soft operations over hard operations. Soft reboot and graceful stop allow the OS to shut down cleanly, reducing the risk of data loss or corruption.
Monitor instance state during operations. Use the Horizon dashboard or CLI to verify that operations complete successfully. If an instance remains in a transitional state for more than a few minutes, investigate before attempting additional operations.
Schedule maintenance windows for production systems. Do not reboot production instances during peak usage unless absolutely necessary. Plan restart operations during low-traffic periods.
Test reboot procedures on development instances first. If you are uncertain how an application will handle a reboot, test the procedure on a non-production instance before applying it to production.
Document your instance dependencies. Know which services depend on each instance so you can coordinate restart operations and avoid unexpected downtime for dependent systems.
Use snapshots before major operations. Before performing any operation that could result in data loss, create a snapshot of the instance. This provides a recovery point if something goes wrong.
Check for pending disk writes before shutdown. If you are running databases or applications with heavy disk I/O, ensure all pending writes have completed before stopping or rebooting the instance.
Review console logs after unexpected reboots. If an instance required a hard reboot, check the console output for errors that explain why the soft reboot failed. This information helps prevent future issues.
Troubleshooting Common Issues
Instance Won't Stop or Soft Reboot
If an instance does not respond to stop or soft reboot commands, the guest OS may be frozen or unresponsive. Verify the instance state using openstack server show <instance-id>. If the instance remains in a transitional state for more than 5 minutes, try a hard reboot.
Instance Won't Start After Stop
If an instance fails to start after being stopped, check the console log for boot errors using openstack console log show <instance-id>. Common causes include corrupted file systems, misconfigured fstab entries, or hardware allocation failures.
Instance Stuck in Reboot State
If an instance remains in the Reboot state for more than 10 minutes, the hypervisor may have lost communication with the instance. Try a hard reboot to force a clean restart. If the problem persists, contact support.
Data Loss After Hard Reboot
If you experience data loss after a hard reboot, check for file system errors using fsck (Linux) or chkdsk (Windows). Restore missing data from snapshots or backups. Review application logs to identify which data was affected.
Summary
Managing instance lifecycle operations effectively requires understanding the differences between start, stop, soft reboot, and hard reboot. Each operation serves a specific purpose and carries different levels of risk.
Use stop and start for planned shutdowns and restarts. These operations are safe and allow the OS to shut down cleanly. Use soft reboot for routine restarts when you need to apply updates or recover from application issues. Reserve hard reboot for emergency situations when the instance is completely unresponsive.
Following best practices like preferring soft operations, monitoring instance state, and creating snapshots before major changes will help you manage instances confidently while minimizing risk to your infrastructure.
Understanding these fundamental operations gives you the control you need to maintain healthy, reliable cloud infrastructure on OpenStack.