Cluster Retraction


This guide provides instructions on how to configure fallback in case of endpoints behind a cluster that are not healthy. Cluster groups a set of endpoints together into a resource pool that is used in the load balancer configuration. To learn more about endpoints and load balancers, see Load Balancing and Service Mesh.

When a route points to multiple clusters as part of a weighted cluster configuration, and any of the clusters have no healthy endpoints, then traffic to the route is distributed only among the remaining clusters that have one or more healthy endpoints. This is referred to as cluster retracting. An endpoint is deemed unhealthy in case a health-check fails or endpoint discovery fails.

Using the instructions provided in this guide, you can enable or disable cluster retraction for an HTTP or TCP load balancer.

Note: Cluster retraction is enabled by default for HTTP and TCP load balancers. This means that weighted cluster configuration of route and panic-threshold configuration for a cluster may not work when any cluster gets retracted because of no healthy endpoints.


The following prerequisites apply:


Cluster retraction is enabled by default for HTTP and TCP load balancers. You can disable it while creating a new load balancer, or you can enable or disable it for an existing load balancer. This example shows how to update an existing load balancer to enable or disable cluster retraction for an HTTP load balancer.

Perform the following steps to enable or disable cluster retraction:

Step 1: Navigate to the load balancer configuration.
  • Log into F5® Distributed Cloud Console (Console).

  • Click Multi-Cloud App Connect.

Figure: Console Homepage
Figure: Console Homepage
  • Change to your application Namespace.

  • Click Manage > Load Balancers > HTTP Load Balancers.

Figure: HTTP Load Balancers
Figure: HTTP Load Balancers
  • Click ... > Manage Configuration > Edit Configuration for your HTTP load balancer from the displayed list.
Figure: Edit HTTP Load Balancers
Figure: Edit HTTP Load Balancers
Step 2: Start editing the route configuration.

The only route configurations that allow a cluster retraction configuration is Simple Route and Custom Route.

These instructions show Simple Route. If Custom Route is needed, noted below.

  • In Routes section of HTTP Load Balancers, click Configure to open the route configuration page.

  • Click + Add Item button.

  • Confirm Route Type is Simple Route.

Note: Cluster retraction only works with Simple Route if Prefix is entered for HTTP, but not required for TCP.

  • Enter Prefix.

  • If Custom Route Object.

    • Reference to Custom Route requires Multi-Cloud App Connect > Manage > Virtual Host > Routes > Configure in Routes box.

    • + Add Item > Destination List in Actions > Configure link in Destination List box.

    • Toggle Show Advanced Fields in Destination Actions box > in Select Cluster retract option select Disable cluster retraction.

    • Apply button > Save and Exit button.

    Note: See full Routes setup in, Virtual Hosts.

Step 3: Configure cluster retraction.
  • Under the Advanced Options section, click Configure for the route entries for which you want to enable the cluster retraction.

  • Select Cluster retract option drop-down menu in Miscellaneous Options section option:

    • Select Retract cluster with no healthy endpoints to enable cluster retraction.

    • Select Disable cluster retraction to disable cluster retraction.

Cluster Retraction Settings
Figure: Cluster Retraction Settings
  • Click Apply.

  • Click Apply to apply the cluster retraction setting to the route configuration.

  • Click Apply in the route configuration page to apply the changes to the load balancer configuration.

Step 4: Complete applying cluster retraction to load balancer.

Click Save and Exit.

Note: Any subsequent endpoint failures result in performing cluster retraction.


API References