Instance Actions¶

This page describes the different actions you can perform on instances in Safespring Compute, such as stopping, starting, rebooting, and shelving.

This page includes OpenStack CLI commands. See the API Access documentation for instructions on how to install and configure the command line client.

Billing implications¶

At Safespring, you are billed for an instance as long as its resources (CPU, RAM, compute node capacity) are reserved — even if the instance is shut off, paused, or suspended. The only way to reduce billing for an instance you want to keep is to shelve it. A shelved instance is removed from the compute node entirely and you are only billed for the storage used by its snapshot.

Overview¶

The following actions are available for instances:

Action	Description	Billed for compute	Resources freed
Shut off	Shuts down the instance gracefully. The instance remains allocated on a compute node.	Yes	None
Start	Boots a previously shut off instance.	Yes	N/A
Reboot	Restarts the instance. Can be soft (graceful) or hard (forced).	Yes	None
Pause	Freezes the instance in memory. Very fast to resume.	Yes	None
Unpause	Resumes a paused instance from memory.	Yes	N/A
Suspend	Saves instance state to disk, similar to hibernation.	Yes	CPU and RAM freed
Resume	Restores a suspended instance from disk.	Yes	N/A
Shelve	Shuts down the instance and removes it from the compute node. A snapshot is stored.	No — only snapshot storage is billed	CPU, RAM, and compute node
Unshelve	Restores a shelved instance from its snapshot onto a compute node.	Yes	N/A

Shut off and start¶

Shutting off an instance performs a graceful shutdown of the operating system, similar to pressing the power button on a physical server. The instance stays allocated on its compute node and all resources (disk, network, IP address) are preserved.

Using the Horizon dashboard¶

In the instance list, select Shut Off Instance from the dropdown menu. To start it again, select Start Instance.

Using the CLI¶

openstack server stop my-instance

openstack server start my-instance

Note

A shut off instance still occupies resources on the compute node and you will continue to be billed for it. If you want to free up compute resources and reduce billing while keeping the instance for later use, consider shelving instead.

Reboot¶

Rebooting restarts the instance operating system. A soft reboot sends a shutdown signal to the OS and then boots it again. A hard reboot is the equivalent of a power cycle and should be used when the instance is unresponsive.

Using the Horizon dashboard¶

Select Soft Reboot Instance or Hard Reboot Instance from the dropdown menu in the instance list.

Using the CLI¶

Soft reboot (graceful):

openstack server reboot my-instance

Hard reboot (forced):

openstack server reboot --hard my-instance

Warning

A hard reboot does not give the operating system a chance to shut down gracefully. Use it only when the instance is not responding to a soft reboot.

Pause and unpause¶

Pausing an instance freezes it in its current state in memory. The instance stops executing but its memory contents are preserved on the compute node. Unpausing restores execution immediately, making this the fastest way to temporarily stop and resume an instance.

Using the Horizon dashboard¶

Select Pause Instance from the dropdown menu. To resume, select Resume Instance.

Using the CLI¶

openstack server pause my-instance

openstack server unpause my-instance

Suspend and resume¶

Suspending an instance saves its entire state (including memory) to disk, similar to hibernating a laptop. This frees up CPU and RAM on the compute node while allowing you to restore the instance to exactly the state it was in when suspended.

Using the Horizon dashboard¶

Select Suspend Instance from the dropdown menu. To restore, select Resume Instance.

Using the CLI¶

openstack server suspend my-instance

openstack server resume my-instance

Shelve and unshelve¶

Shelving an instance shuts it down, takes a snapshot of it, and removes it from the compute node entirely. This frees up all compute resources (CPU, RAM, compute node capacity). The instance can be restored later from its snapshot using unshelve, though this takes longer than starting a shut off instance since it must be scheduled onto a compute node again.

Shelving is the only instance action that reduces your billing. While shelved, you are only billed for the storage consumed by the instance snapshot — not for compute resources.

Shelving works best for volume-based instances (b2 flavors), since the root disk already resides in the central storage and does not need to be uploaded to the image service. For l2-flavor instances with local disk, the shelve operation needs to upload the entire local disk as a snapshot to the image service, which can take a long time for large disks.

Using the Horizon dashboard¶

Select Shelve Instance from the dropdown menu. To restore, select Unshelve Instance.

Using the CLI¶

openstack server shelve my-instance

openstack server unshelve my-instance

Info

Shelving is useful for instances you want to keep but do not need running for an extended period. Since the instance is removed from the compute node, it does not count against your compute quota while shelved.

Rescue mode¶

If your instance has a broken root filesystem and cannot boot properly, you can use rescue mode to boot from a different image and repair the disk. This is covered in the Troubleshooting documentation.