Our intuitive status system provides easy management of your GPU Server Blibs. Servers transition between statuses – currently at status A, and moving towards status B – allowing for seamless operation and control.
Possible statuses
Active Lifecycle States (normal operation):
- Running: The Blib is actively processing tasks and is accessible.
- Stopped: Halts the server. Resources (GPUs, CPU Cores, RAM) remain allocated, enabling rapid restarts within seconds. Recommended for cost optimization.
- Restarting: The Blib is in the process of powering on after being stopped. This may be referred to as a forced power reset. We recommend initiating a system reboot using the
sudo reboot command first.
Infrastructure & Maintenance States (Trooper.AI specific):
- Frozen: The server is archived, releasing resource allocation. Startup times may vary, and port reassignment may be required. Using the stopped state is recommended if predictable startup times are critical! Freezing is not recommended for enterprise clients due to potential port changes and increased downtime.
- Migrated: The Blib is being moved to different hardware; this process may take 10 to 90 minutes.
Destructive & Irreversible States:
- Terminated: Permanently deletes the server and all associated data. This action cannot be undone.
- Reset: Deletes all data on the server and reinstalls the initial configuration. To exclude templates from reinstallation, remove them from the server before resetting. This function does not affect your prepaid budget or predefined prices, but ports may change and provisioning a new Blib can take some time. ALL DATA WILL BE LOST! This action cannot be undone. Contact us with any questions regarding this reset function!
Q&A about status system
Why do A100 servers sometimes take longer to resume from a frozen state compared to other server types?
A100 servers are equipped with substantial ECC Video RAM. Upon resuming from a frozen state, a rigorous diagnostic test is performed on this memory to ensure data integrity and system stability. While this test is essential, it can occasionally result in a longer resume time.
How to minimize startup time of frozen servers?
The most efficient method for minimizing startup times is to stop the server instance rather than freeze it. Please note that freezing a server will incur a brief delay before it becomes available again or even a migration of the server is needed which changes ports, underlying hardware like CPU model and CPU speed and takes extra time.