Handle failure of service dependencies

2.1 System dependency initially healthy

  1. You already observed that all three EC2 instances are successfully serving requests

  2. In a new tab navigate to ELB Target Groups console

  3. Click on the Targets tab (bottom half of screen)

  4. Under Registered Targets observe the three EC2 instances serving your web service

  5. Note that they are all healthy (see Status and Description)

    • In this state the ELB will route traffic to any of the three servers


  6. From the Target Groups console, now click on the the Health checks tab

    • Note here that the Path is configured to /healthcheck
  7. Copy the URL of the web service to a new tab and append /healthcheck to the end of the URL

    • The new URL should look like:

    • Refresh several times and observe the health check on the three servers

    • Note the check is successful

  8. The EC2 servers receive user requests (for a TV show recommendation) on the path / and they receive health check requests from the Elastic Load Balancer on the path /healthcheck

    • The health check always returns an http 200 code for any request to it.
    • The server code running on each EC2 instance can be viewed here, or you can view the health check code excerpt below:
# Healthcheck request - will be used by the Elastic Load Balancer
elif self.path == '/healthcheck':

      # Return a healthy code
      self.send_header('Content-type', 'text/html')

2.2 Simulate dependency not available

2.2.1 Disable RecommendationService

You will now simulate a complete failure of the RecommendationService. Every request in turn makes a (simulated) call to the getRecommendation API on this service. These will all fail for every request on every server.

  1. In a new tab, navigate to the Parameter Store on the AWS Systems Manager console
  2. Click on RecommendationServiceEnabled
  3. Click Edit
  4. In the Value box, type false
  5. Click Save Changes
    • A status message should say Edit parameter request succeeded

The RecommendationServiceEnabled parameter is used only for this lab. The server code reads its value, and simulates a failure in RecommendationService (all reads to the DynamoDB table simulating the service will fail) when it is false.

2.2.2 Observe behavior when dependency not available

  1. Refresh the test web service multiple times

    • Note that it fails with 502 Bad Gateway
    • For each request one of the servers receiving the request attempts to call the RecommendationService but catastrophically fails and fails to return a reply (empty reply) to the load balancer, which in turn presents this as a http 502 failure.
  2. You can observe this by opening a new tab and navigating to ELB Load Balancers console:

  3. Click on the Monitoring tab (bottom half of screen)

    • Observe the ELB 5XXs (Count) and HTTP 502s (Count) errors for the load balancer
    • It will take a minute for the metrics to show up. Make sure you refresh the web service page multiple times in your browser
    • These are the error codes the load balancer returns on every request during this simulated outage


  4. Compare these metrics to those for the target group (the EC2 servers themselves)

    • Return to the Target Groups console and click the Monitoring tab there
    • Observe HTTP 5XXs ( Count ) errors shows no data
    • The servers themselves are not returning actual http error codes, they are failing to return any data at all


    • We need to update the server code to handle when the dependency is not available

2.3 Update server code to handle dependency not available

The getRecommendation API is actually a get_item call on a DynamoDB table. Examine the server code to see how errors are currently handled

  1. The server code running on each EC2 instance can be viewed here

  2. Search for the call to the RecommendationService. It looks like this:

     response = call_getRecommendation(self.region, user_id)
    • What happens if this call fails?
  3. Choose one of the options below (Option 1 - Expert or Option 2 - Assisted) to improve the code and handle the failure

2.3.1 Option 1 - Expert option: make and deploy your changes to the code

You may choose this option, or skip to Option 2 - Assisted option

This option requires you have access to place a file in a location accessible via https/https via a URL. For example a public readable S3 bucket, gist (use the raw option to get the URL), or your private webserver.

  1. Download the existing server code from here: server_basic.py
  2. Modify the code to handle the call to the RecommendationService
  3. When the call to RecommendationService fails then instead of using the response data you requested and did not get, return a static response:
    • Instead of user first name return Valued Customer
    • Instead of a personalized recommended TV show, return I Love Lucy
    • Try to also return some diagnostic information on the cause of the error
  4. Put your updated server code in a location where it can be downloaded via its URL using wget
  5. In the AWS Console go the HealthCheckLab CloudFormation stack and Update it:
    • Leave Use current template selected and click Next
    • Find the ServerCodeUrl parameter and enter the URL for your new code
    • When stack status is CREATE_COMPLETE (about four minutes) then continue

If you completed the Option 1 - Expert option, then skip the Option 2 - Assisted option section and continue with 2.3.3 Error handling code

2.3.2 Option 2 - Assisted option: deploy workshop provided code

  1. The new server code including error handling can be viewed here
  2. Search for Error handling in the comments (occurs twice). What will this code do now if the dependency call fails?
Deploy the new error handling code
  1. Navigate to the AWS CloudFormation console

  2. Click on the HealthCheckLab stack

  3. Click Update

  4. Leave Use current template selected and click Next

  5. Find the ServerCodeUrl parameter and enter the following:

  6. Click Next until the last page

  7. At the bottom of the page, select I acknowledge that AWS CloudFormation might create IAM resources with custom names

  8. Click Update stack

  9. Click on Events, and click the refresh icon to observe the stack progress

    • New EC2 instances running the error handling code are being deployed
    • When stack status is CREATE_COMPLETE (about four minutes) then continue

2.3.3 Error handling code

This is the error handling code from server_errorhandling.py. The Option 2 - Assisted option uses this code. If you used the Option 1 - Expert option, you can consult this code as a guide.


Click here to see the code:

2.3.4 Observe behavior of web service with added error handling

  1. After the new error-handling code has successfully deployed, refresh the test web service page multiple times. Observe:

    • It works. It no longer returns an error
    • All three EC2 instances and Availability Zones are being used
    • A default recommendation for Valued Customer is displayed instead of a user-personalized one
    • There is now Diagnostic Info. What does it mean?
  2. Refer back to the newly deployed code to understand why the website behaves this way now

The Website is working again, but in a degraded capacity since it is no longer serving personalized recommendations. While this is less than ideal, it is much better than when it was failing with http 502 errors. The RecommendationService is not available, so the app instead returns a static response (the default recommendation) instead of the data it would have obtained from RecommendationService.

Well-Architected for Reliability: Best practice
Implement graceful degradation to transform applicable hard dependencies into soft dependencies: When a component’s dependencies are unhealthy, the component itself can still function, although in a degraded manner. For example, when a dependency call fails, instead use a predetermined static response.