Issue

Use this article to troubleshoot cases in which VM migration is not triggered after a NIC goes down under a ZSphere HA policy.

Symptoms

• After a NIC is manually brought down in the environment, VM migration is expected but not triggered.
• The scenario is used to demonstrate or verify high availability policy functions.

Environment

• Product: ZVF
• Version: 4.x
• Component: ZMigrate
• Affected object: HA
• Scenario: High availability issues. Confirm the current version, deployment topology, storage type, and operation timeline.

Possible causes

• Manually bringing down one NIC does not necessarily meet HA trigger conditions.
• Failures of the service, management, and storage networks have different HA migration-trigger semantics.
• The HA policy, monitoring network, host-isolation determination, or shared-storage prerequisites are not satisfied.

Resolution

Perform the following checks in sequence, and record the affected objects, occurrence time, complete error messages, and current state.

1. Define whether the test target is host disconnection, a management-network failure, a storage-network failure, or a service-network failure.
2. Confirm that the HA policy is enabled and that the VM and host meet migration prerequisites.
3. Check HA-related logs and alerts to confirm whether the failure is determined to be host isolation.
4. Simulate the failure in a controlled manner. Do not directly bring down management or storage networks in a production environment.

Remediation

1. Based on the failure type, confirm that the HA determination requirements for host isolation and migration prerequisites are met.
2. After correcting the HA policy, host network, or storage-network configuration, revalidate the failure scenario in a controlled test environment.

Verification

• In a controlled failure scenario, HA recognizes the failure according to policy and triggers the expected recovery or migration behavior.
• VM service continuity, logs, and alert results match expectations.