Network switch failure

This topic describes how to handle network switch failure.

In an FPO cluster, if Auto recovery is enabled and there are more than maxFailedNodesForRecovery non functional nodes, auto recovery does not recover the nodes. By default, maxFailedNodesForRecovery is three nodes. You can change this number depending on your cluster configuration.

A switch network failure can cause nodes to be reported as non functional. If you want auto recovery to protect against switch network failures, careful planning is required in setting up the FPO cluster. For example, a network switch failure must not bring disks (with metadata) down from 3 or more failure groups, and maxFailedNodesForRecovery must be configured to a value that is larger than the number of down nodes that will result from a switch network failure.