An automatic takeover in a NetApp ONTAP High-Availability (HA) pair is a critical process designed to maintain service continuity when a node fails. For this to happen safely and automatically, two key conditions must be met.

First, the NVRAM, which contains uncommitted write operations, must be synchronized between the two nodes. This ensures that the surviving node has a complete copy of the in-flight data from the failed node, preventing any data loss (zero Recovery Point Objective). If the NVRAM is not synchronized, automatic takeover is disabled to preserve data integrity.

Second, the surviving node must be able to reach the shared mailbox disks. These disks serve as a secondary communication channel or tie-breaker. If the primary HA interconnect fails, the nodes use the mailboxes to verify each other's status, which is essential for preventing a "split-brain" scenario.

A. The combined utilization of both nodes must be less than 100%.

This is a performance consideration, primarily for the giveback process. A takeover is prioritized for availability and is not blocked by high utilization.

B. The cluster network must be available.

The HA takeover decision relies on the dedicated HA interconnect and the mailbox disks, not the general-purpose cluster interconnect network.

1. NetApp ONTAP Documentation, "High-Availability configuration," Section: "How takeover and giveback work."

This section explains the role of the HA interconnect and NVRAM mirroring: "The contents of each node's NVRAM are mirrored to its partner's NVRAM. This mirroring ensures that in the event of a takeover, the surviving node can write all of the failed node's cached data to disk, ensuring data consistency." This directly supports option D.

2. NetApp ONTAP Documentation, "High-Availability configuration," Section: "How takeover and giveback work."

The same section details the mailboxing mechanism: "If the heartbeat messages stop [on the HA interconnect], the node attempts to verify the partner's status through an alternate path called the HA mailbox... The use of mailboxing prevents a 'split-brain' scenario." This directly supports option C.

3. NetApp ONTAP Documentation, "Commands for managing automatic takeover and giveback," Command: storage failover show

The output of this command explicitly lists reasons why takeover might be disabled. Common reasons shown in documentation examples include "NVRAM log not synchronized" and "partner mailbox disks not accessible," reinforcing the critical nature of options C and D.