Red Hat OpenShift Container Platform

Understanding EgressIP failover control

The reachabilityTotalTimeoutSeconds parameter controls how quickly the system detects a failing egressIP node and initiates a failover. This parameter directly determines the maximum time the platform waits before declaring a node unreachable.

When you configure egressIP with multiple egress nodes, the complete failover time from node failure to recovery on a new node is expected to be on the order of seconds or longer. This is because the new IP assignment can only begin after the reachabilityTotalTimeoutSeconds period has fully elapsed without a successful check.

To ensure traffic uses the correct external path, egressIP traffic on a node will always egress through the network interface on which the egressIP address has been assigned.

Configuring the EgressIP failover time limit

You can configure the reachabilityTotalTimeoutSeconds parameter to control how quickly the system detects a failing egressIP node and initiates a failover.

Prerequisites

You installed the OpenShift CLI (oc).
You logged in to the cluster as a cluster administrator.

Procedure

Edit the Network custom resource by running the following command:
```
$ oc edit network.operator cluster
```
Navigate to the egressIPConfig: {} section under spec:defaultNetwork:ovnKubernetesConfig:
Modify the block to include the reachabilityTotalTimeoutSeconds parameter with your chosen value, 5 seconds for example. Make sure to use the correct indentation:
```
  defaultNetwork:
    ovnKubernetesConfig:
      egressIPConfig:
        reachabilityTotalTimeoutSeconds: 5
```
The value must be an integer between 0 and 60. For details on possible values, see the "EgressIP failover settings" section.
Save and exit the editor. The operator automatically applies the changes.

Verification

Verify that the system correctly accepted the reachabilityTotalTimeoutSeconds parameter by running the following command:
```
$ oc get network.operator cluster -o yaml
```
Inspect the output and confirm that the reachabilityTotalTimeoutSeconds parameter is correctly nested under spec:defaultNetwork:ovnKubernetesConfig:egressIPConfig: with your intended value:
```
 # ...
  spec:
    # ...
    defaultNetwork:
      ovnKubernetesConfig:
        egressIPConfig:
          reachabilityTotalTimeoutSeconds: 5
        gatewayConfig:
  # ...
```

EgressIP failover settings

The reachabilityTotalTimeoutSeconds parameter defines the total time limit in seconds for the platform health check process before a node is declared down.

The following table summarizes the acceptable values and their implications:

Parameter Value (Seconds) Effect on reachability check Failover impact and use case

Parameter Value (Seconds)	Effect on reachability check	Failover impact and use case
`0`	Disables the reachability check.	No automatic failover: Use only if an external system handles node health monitoring and failover. The platform will not automatically react to node failures.
`1 - 60`	Sets the total time limit for reachability probing.	Directly controls detection time: This value defines the lower limit for your overall failover time. A smaller value leads to faster failover but might increase network traffic. Default: 1 second. The maximum accepted integer value is 60.

0

Disables the reachability check.

No automatic failover: Use only if an external system handles node health monitoring and failover. The platform will not automatically react to node failures.

1 - 60

Sets the total time limit for reachability probing.

Directly controls detection time: This value defines the lower limit for your overall failover time. A smaller value leads to faster failover but might increase network traffic. Default: 1 second. The maximum accepted integer value is 60.