Starting a Google Compute instance with Python
Question:
I am trying to start a Google Compute instance with the Google API Python Client Library. This is so that a cheap instance (running on a single core) can periodically start and stop a more expensive instance (with many cores) periodically, to keep costs down.
I have successfully installed the different components and run Google’s example script create_instance.py (which creates an instances, runs a startup script, and deletes the instance). Inspecting the PyDoc reference for the Compute Engine API, and cross-referencing how the other instances() functions work in the create_instance.py example, I would expect the start instance command to be:
python compute.instances().start(project=*, zone=*, instance=*).execute()
The above command gives me the error “An expression was expected after ‘(‘. at line:1 char:34” – this is the first parenthesis.
a. What have I done wrong?
b. Is using the Google API with Python a good way to start instances from other instances, programmatically?
Answers:
-
Generally I would expect you would need to import the api library with an import statement or perhaps a runtime flag (-m somemodule?). Running a line of python directly from the command line is usually not the best way to proceed. Instead Google provides the gcloud command line tool.
-
An authentication/login function is usually called before sending the API actual commands. On a Google VM, this can be either an id/private key or a blank id/key if the VM is specifically authorized to call the API or act as a specific account. This authorization can be set up from the compute engine web control panel when creating a Google VM the first time. On an external VM, it will need an id/private key to supply to the Google API. So, a one liner in python probably won’t work as it is missing this step.
-
The compute.instances().start() function takes required parameters to start a specific instance that has stopped. That means:
- the VM instance has been created previously
- the VM instance is in the stopped state
- the instance to be restarted is identified by a specific project id, a (geo) zone, and an instance name that is supplied in the call to
start
From Google Cloud Python Documentation
start(project=, zone=, instance=*) Starts an instance that was
stopped using the instances().stop method. For more
information, see Restart an instance.
Args: project: string, Project ID for this request. (required)
zone: string, The name of the zone for this request. (required)
instance: string, Name of the instance resource to start. (required)
…
Below is the code needed to start a compute engine instance
from googleapiclient import discovery
service = discovery.build('compute', 'v1')
print('VM Instance starting')
# Project ID for this request.
project = 'project_name'
# The name of the zone for this request.
zone = 'zone_value'
# Name of the instance resource to start.
instance = 'instance_name'
request = service.instances().start(project=project, zone=zone, instance=instance)
response = request.execute()
print('VM Instance started')
This is the code I used for starting my VM instance from a cloud function.
An important thing to note here is that this can only start an instance if the instance is in stopped state, which suits my requirements perfectly.
I used the code shared by @user570778, and for me it worked fine.
`from googleapiclient import discovery
service = discovery.build(‘compute’, ‘v1’)
print(‘VM Instance starting’)
Project ID for this request.
project = ‘project_name’
The name of the zone for this request.
zone = ‘zone_value’
Name of the instance resource to start.
instance = ‘instance_name’
request = service.instances().start(project=project, zone=zone, instance=instance)
response = request.execute()
print(‘VM Instance started’)
`
I’m wondering, is it possible to start multiples instance in the same function?
from google.cloud import compute_v1
project = ""
zone = ""
instance_client = compute_v1.InstancesClient.from_service_account_file("ServiceAccount.json")
instance_list = instance_client.list(project=project, zone=zone)
for instance in instance_list:
print(instance.name)
instance_client.start(project=project, zone=zone, instance=instance.name)
Requires iam role Compute Admin
Source available here: https://github.com/googleapis/python-compute
I am trying to start a Google Compute instance with the Google API Python Client Library. This is so that a cheap instance (running on a single core) can periodically start and stop a more expensive instance (with many cores) periodically, to keep costs down.
I have successfully installed the different components and run Google’s example script create_instance.py (which creates an instances, runs a startup script, and deletes the instance). Inspecting the PyDoc reference for the Compute Engine API, and cross-referencing how the other instances() functions work in the create_instance.py example, I would expect the start instance command to be:
python compute.instances().start(project=*, zone=*, instance=*).execute()
The above command gives me the error “An expression was expected after ‘(‘. at line:1 char:34” – this is the first parenthesis.
a. What have I done wrong?
b. Is using the Google API with Python a good way to start instances from other instances, programmatically?
-
Generally I would expect you would need to import the api library with an
importstatement or perhaps a runtime flag (-m somemodule?). Running a line of python directly from the command line is usually not the best way to proceed. Instead Google provides thegcloudcommand line tool. -
An authentication/login function is usually called before sending the API actual commands. On a Google VM, this can be either an id/private key or a blank id/key if the VM is specifically authorized to call the API or act as a specific account. This authorization can be set up from the compute engine web control panel when creating a Google VM the first time. On an external VM, it will need an id/private key to supply to the Google API. So, a one liner in python probably won’t work as it is missing this step.
-
The
compute.instances().start()function takes required parameters to start a specific instance that has stopped. That means:- the VM instance has been created previously
- the VM instance is in the stopped state
- the instance to be restarted is identified by a specific project id, a (geo) zone, and an instance name that is supplied in the call to
start
From Google Cloud Python Documentation
start(project=, zone=, instance=*) Starts an instance that was
stopped using the instances().stop method. For more
information, see Restart an instance.Args: project: string, Project ID for this request. (required)
zone: string, The name of the zone for this request. (required)
instance: string, Name of the instance resource to start. (required)…
Below is the code needed to start a compute engine instance
from googleapiclient import discovery
service = discovery.build('compute', 'v1')
print('VM Instance starting')
# Project ID for this request.
project = 'project_name'
# The name of the zone for this request.
zone = 'zone_value'
# Name of the instance resource to start.
instance = 'instance_name'
request = service.instances().start(project=project, zone=zone, instance=instance)
response = request.execute()
print('VM Instance started')
This is the code I used for starting my VM instance from a cloud function.
An important thing to note here is that this can only start an instance if the instance is in stopped state, which suits my requirements perfectly.
I used the code shared by @user570778, and for me it worked fine.
`from googleapiclient import discovery
service = discovery.build(‘compute’, ‘v1’)
print(‘VM Instance starting’)
Project ID for this request.
project = ‘project_name’
The name of the zone for this request.
zone = ‘zone_value’
Name of the instance resource to start.
instance = ‘instance_name’
request = service.instances().start(project=project, zone=zone, instance=instance)
response = request.execute()
print(‘VM Instance started’)
`
I’m wondering, is it possible to start multiples instance in the same function?
from google.cloud import compute_v1
project = ""
zone = ""
instance_client = compute_v1.InstancesClient.from_service_account_file("ServiceAccount.json")
instance_list = instance_client.list(project=project, zone=zone)
for instance in instance_list:
print(instance.name)
instance_client.start(project=project, zone=zone, instance=instance.name)
Requires iam role Compute Admin
Source available here: https://github.com/googleapis/python-compute