Monitoring and Managing Instances

Monitoring GPU Performance

You can monitor GPU performance metrics in real-time using Remote's API. Retrieve GPU metrics for each provisioned instance.

# Retrieve GPU performance metrics
for instance in instances:
    performance_metrics = client.get_gpu_metrics(instance.id)
    print(f"Instance ID: {instance.id}, GPU Metrics: {performance_metrics}")

Example of output

print("GPU Performance Metrics:")
print("========================")
print(f"Instance ID: {instance_id}")
print(f"GPU Utilization: {performance_metrics['utilization']}%")
print(f"Temperature: {performance_metrics['temperature']}°C")
print(f"Memory Usage: {performance_metrics['memory_usage']} MB")
print(f"Fan Speed: {performance_metrics['fan_speed']} RPM")

Managing Instances

Remote's API allows you to manage GPU instances programmatically. You can start, stop, and terminate instances as needed.

# Start GPU instances
for instance in instances:
    client.start_instance(instance.id)

# Stop GPU instances
for instance in instances:
    client.stop_instance(instance.id)

# Terminate GPU instances
for instance in instances:
    client.terminate_instance(instance.id)

Example:

Start GPU Instance

# Specify the instance IDs of the GPU instances you want to start
instance_ids = ['instance123', 'instance456']

# Start GPU instances
for instance_id in instance_ids:
    client.start_instance(instance_id)
    print(f"Started instance {instance_id}")

Stop GPU Instance

# Specify the instance IDs of the GPU instances you want to stop
instance_ids = ['instance123', 'instance456']

# Stop GPU instances
for instance_id in instance_ids:
    client.stop_instance(instance_id)
    print(f"Stopped instance {instance_id}")

Terminate GPU Instance

# Specify the instance IDs of the GPU instances you want to terminate
instance_ids = ['instance123', 'instance456']

# Terminate GPU instances
for instance_id in instance_ids:
    client.terminate_instance(instance_id)
    print(f"Terminated instance {instance_id}")