The Circuit Breaker Example demonstrates a generic pattern for reporting the failure of a service and then limiting access to the failed service until it becomes available to handle requests. This helps prevent cascading failure in other services that depend on the failed services for functionality.
This example shows you how to implement a Circuit Breaker and Fallback pattern in your services.
By default, this example will show how to use the community version of the Opossum Circuit Breaker
To use the fully supported version of Opossum, there are a couple steps that should be completed. Note: All the steps should be performed in the greeting-service sub-directory
-
Make sure there is a
.npmrcfile in the root of thegreeting-servicedirectory. -
Uninstall the community version first with
npm uninstall opossumthen install the fully supported version with this command:npm install @redhat/opossum -
In the app.js file:
/* Remove or comment out the line below */
const Opossum = require('opossum');
/* Uncomment the line below to use the fully Red Hat Supported version of Opossum */
// const Opossum = require('@redhat/opossum');
For more information on Red Hat's Node.js offerings, click here
The Circuit Breaker is a pattern intended to mitigate the impact of network failure and high latency on service architectures where services synchronously invoke other services. In such cases, if one of the services becomes unavailable due to network failure or incurs unusually high latency values due to overwhelming traffic, other services attempting to call its endpoint may end up exhausting critical resources in an attempt to reach it, rendering themselves unusable. This condition is also known as cascading failure and can render the entire microservice architecture unusable.
Essentially, the Circuit Breaker acts as a proxy between a protected function and a remote function, which monitors for failures. Once the failures reach a certain threshold, the circuit breaker trips, and all further calls to the circuit breaker return with an error or a predefined fallback response, without the protected call being made at all. The Circuit Breaker usually also contains an error reporting mechanism that notifies you when the Circuit Breaker trips.
In an architecture where multiple services depend on each other for functionality, a failure in one service can rapidly propagate to its dependent services, causing the entire architecture to collapse. Implementing a Circuit Breaker pattern helps prevent this. With the Circuit Breaker pattern implemented, a service client invokes a remote service endpoint via a proxy at regular intervals. If the calls to the remote service endpoint fail repeatedly and consistently, the Circuit Breaker trips, making all calls to the service fail immediately over a set timeout period and returns a predefined fallback response. When the timeout period expires, a limited number of test calls are allowed to pass through to the remote service to determine whether it has healed, or remains unavailable. If these test calls fail, the Circuit Breaker keeps the service unavailable and keeps returning the fallback responses to incoming calls. If the test calls succeed, the Circuit Breaker closes, fully enabling traffic to reach the remote service again.
- Enables a service to handle the failure of other services it invokes.
- Optimizing the timeout values can be challenging
- Larger-than-necessary timeout values may generate excessive latency.
- Smaller-than-necessary timeout values may introduce false positives.
You can run this example as node processes on your localhost or as pods on an Openshift Cluster. OpenShift Local can be used to try out Openshift locally.
To run the basic application on your local machine, just run the
start-localhost.sh script.
$ ./start-localhost.sh
This will launch the greeting service on port 8080 and the name
service on port 8081. To kill the servers, run ./shutdown-localhost.sh.
The cluster should be started, and you should be logged in with a currently
active project. Then run the ./start-openshift.sh script.
$ oc new-project circuit-breaker-example # Create a project to deploy to
$ ./start-openshift.sh # Launch the example appThis link contains instructions on how to install the OpenShift Distributed Tracing Platform and enable tracing.