This rule raises an issue when custom polling loops are used instead of AWS waiters for checking resource states.

Why is this an issue?

When working with AWS resources that require time to reach a desired state, such as EC2 instances starting up, S3 buckets being created, or DynamoDB tables becoming available, developers often implement custom polling loops to check the resource status repeatedly. However, this approach is inefficient and error-prone. Custom polling requires manual handling of timing intervals, retry logic, error conditions, and timeout scenarios. It also tends to be verbose and duplicates logic that AWS already provides in a more robust form. AWS SDK provides waiters, which are purpose-built abstractions specifically designed to poll AWS resources until they reach a desired state. Waiters handle the complexities of polling automatically, including appropriate delays, exponential backoff, maximum wait times, and proper error handling.

What is the potential impact?

Custom polling implementations can lead to inefficient resource usage due to inappropriate polling intervals, increased complexity and maintenance overhead, potential race conditions and timing issues, and unreliable error handling that may cause application failures or infinite loops.

How to fix it

Replace custom polling loops with AWS waiters. Identify the AWS resource you need to wait for, then use the appropriate waiter method from the boto3 client. Waiters are available for most AWS services and common state transitions. Use the get_waiter() method on your boto3 client to obtain the waiter, then call wait() with the appropriate parameters.

Noncompliant code example

import boto3
import time

ec2_client = boto3.client('ec2', region_name='us-east-1')

while True:
  response = ec2_client.describe_instance_status(  # Noncompliant
    InstanceIds=[instance_id],
    IncludeAllInstances=True
  )

  instance_status = response['Statuses'][0]['InstanceStatus']['Status']
  if instance_status == 'ok':
    break

  time.sleep(10)

Compliant solution

import boto3

ec2_client = boto3.client('ec2', region_name='us-east-1')

ec2_client.get_waiter('instance_status_ok').wait(
    InstanceIds=[instance_id],
    IncludeAllInstances=True
)

Resources