Retrieving command invocation in AWS SSM

Question

I am trying to send command to a running ubuntu ec2 instance. I have configured the appropriate role and I have an ssm agent running on the ec2 instance. Using the boto3 SDK I am able to use the client.send_command() function to successfully send a shell command and was subsequently able to get the command Id. Now the challenge is polling for the result. I am trying to use the client.get_command_invocation() function but keep getting a InvocationDoesNotExist error. I am confident that I am using the correct command ID and instance ID because I have tested these using the AWS CLI as in aws ssm list-command-invocations --command-id #### --instance-id i-##### and this worked. Here is a snippet of my code:

`ssm_client = boto3.client('ssm')
target_id = "i-####"
response = ssm_client.send_command(
            InstanceIds=[
                target_id 
                    ],
            DocumentName="AWS-RunShellScript",
            Comment="Code to run" ,
            Parameters={
                    "commands": ["ls -l"]
                        }
            )
cmd_id= response['Command']['CommandId']
response = ssm_client.get_command_invocation(CommandId=cmd_id, 
InstanceId=target_id)
print(response)`

Here is the returned error: botocore.errorfactory.InvocationDoesNotExist: An error occurred (InvocationDoesNotExist) when calling the GetCommandInvocation operation

Thanks in advance.

score 17 · Answer 1 · answered May 11 '18 at 00:52

17

I had the same issue, I fixed it by adding a time.sleep() call before calling get_command_invocation(). A short delay should be enough.

answered May 11 '18 at 00:52

Drake Petersen

171
4

1

Ahhh, Thanks very helpful. – Remilekun Basaru May 11 '18 at 11:04

score 8 · Answer 2 · answered Jan 09 '20 at 20:12

Just add these 2 line.

import time
time.sleep(2)

Then It'll work properly generally it take only 0.65sec but it's better to give 2 sec. To make it better you can add some cool stuffs like some print statement in for loop and sleep inside it something like that.

score 4 · Answer 3 · answered May 31 '22 at 03:50

The following worked for me. Check out the API documentation for how to update the default delays and max attempts at https://boto3.amazonaws.com.

from botocore.exceptions import WaiterError 
ssm_client = boto3.client(
    "ssm",
    region_name=REGION,
    aws_access_key_id=KEY,
    aws_secret_access_key=SECRET,
)

waiter = ssm_client.get_waiter("command_executed")
try:
    waiter.wait(
        CommandId=commands_id,
        InstanceId=instance_id,
    )
except WaiterError as ex:
    logging.error(ex)
    return

jellycsc · Answer 4 · 2022-05-30T19:34:45.003

time.sleep does the job 99% of the time. Here is a better way to do this:

MAX_RETRY = 10

get_command_invocation_params = {
    'CommandId': 'xxx',
    'InstanceId': 'i-xxx'
}

for i in range(MAX_RETRY):
    try:
       command_executed_waiter.wait(**get_command_invocation_params)
       break
    except WaiterError as err:
        last_resp = err.last_response
        if 'Error' in last_resp:
            if last_resp['Error']['Code'] != 'InvocationDoesNotExist' or i + 1 == MAX_RETRY:
                raise err
        else:
            if last_resp['Status'] == 'Failed':
                print(last_resp['StandardErrorContent'], file=sys.stderr)
                exit(last_resp['ResponseCode'])
        continue

Retrieving command invocation in AWS SSM

4 Answers4