Last Updated: Oct 28, 2025
Introduction
When working with Crusoe instances, particularly those with local NVMe storage (e.g., a100-pcie/sxm, h100-sxm, s1a), you may occasionally need to reboot to troubleshoot issues such as hanging operations or interruptions to the `sshd` service. It’s crucial to understand the implications of rebooting or resetting these instances to avoid losing important data stored on the ephemeral drives.
Prerequisites
- Access to the Crusoe CLI or Crusoe Console(Steps to Install the CLI)
- VM running on the Crusoe Cloud
Resetting a VM
The reset operation is akin to pulling the power cord on a virtual machine. This action will rerun any startup scripts associated with the VM, but will not run any shutdown lifecycle scripts if configured.
To reset a VM using the Crusoe CLI, use the following command:
# crusoe compute vms reset <vm-name>
Note: This operation will preserve data on the local NVMe drives.
Rebooting a VM (Preserves Local NVMe Storage)
Alternatively, you can perform a soft reboot from within the guest OS. This method preserves the local NVMe storage and allows the VM to reboot without wiping the ephemeral drives.
To reboot the VM via ssh or serial-console, run the following command once inside the instance(Steps to Access a VM):
# sudo reboot now
This command restarts the guest OS in place, preserving any local data stored on the NVMe drives, making it safer when you need to retain data across reboots.
Behavior in Both Scenarios
- The VM will remain in the RUNNING state in the UI throughout both operations (reset or reboot).
- The instance will remain scheduled on the same underlying host.
- A reboot or reset will not wipe ephemeral data
Restarting a VM
For cases where a hardware issue has been identified, the best thing to do is to STOP and START the affected VM to get it to migrate to a new host. You can do this through the Crusoe Console UI, or through the CLI e.g.
# crusoe compute vms stop <vm-name>
# crusoe compute vms start <vm-name>
Important Note: Performing this action will erase all local NVMEs if present and will result in data loss for the VM. Please be sure to use caution when shutting down VMs as part of migrations and ensure that any sensitive data is backed up.
Note
When deciding between resetting, rebooting, or restarting your Crusoe instance, be aware of the consequences, especially regarding data on local NVMe storage. If you're unsure which action to take or need further clarification, feel free to reach out to Crusoe Support for assistance.