ESXi Host Port Channel Connectivity Fails After Power Outage

Products

• VMware vSphere ESXi

Issue/Introduction

After a power outage and subsequent restoration, ESXi hosts experience network connectivity failures. Port channels between ESXi hosts and physical switches fail to establish proper Link Aggregation Control Protocol (LACP) negotiation, resulting in degraded or complete loss of network connectivity.

The issue manifests when physical network switches complete their boot sequence before ESXi hosts during power restoration. Network tunnels show as down, workload connectivity is disrupted, and vMotion operations fail with timeout errors.

Symptoms observed:

• One port in a port channel shows as active while the other is inactive or isolated
• Both ports in a port channel show as down when both should be active
• ESXi host network connectivity to infrastructure components fails
• vMotion failures with "Connection closed by remote host" errors
• Management network intermittent connectivity or complete failure
• Virtual machine network traffic disruption

To verify port channel status and confirm this issue, use Testing VMkernel network connectivity with the vmkping command. If vmkping tests show connectivity through one physical path but not the other, or no connectivity when redundant paths should exist, this maintenance procedure is required.

Environment

• VMware vSphere ESXi
• Physical network switches with LACP port channel configuration
• Redundant network uplinks (vmnic) configured in port channels

Cause

Physical network switches initialize faster than ESXi hosts during power restoration. The switch initializes port channel configuration but does not receive LACP negotiation from the host side because ESXi network services are still initializing. This timing mismatch leaves the port channel in an inconsistent state where the switch does not properly recognize the aggregated links.

The port channel remains in this state because:

• LACP handshake between switch and host is incomplete
• Switch port channel member ports are isolated or in individual mode
• The bonding protocol state machine on the switch side is not synchronized with the host side

Resolution

Administratively disable and re-enable affected port channels on the physical switch to force LACP renegotiation:

1. Identify affected ESXi hosts and their corresponding switch port channels
   • Note the host management IP address
   • Document which vmnics connect to which switch ports
2. Access the physical switch management interface
3. For each affected port channel, administratively disable the port channel interface
4. Wait 10 seconds for the interface to fully shut down
5. Re-enable the port channel interface
6. Verify LACP status shows all member ports as active and bundled using appropriate show commands for the switch vendor
7. From the ESXi host, verify network connectivity:
   • Check vmnic status shows "Up" in vSphere Client under Configure > Networking > Physical adapters
   • Test connectivity using vmkping between affected hosts
   • Confirm vMotion and management network operations function correctly
8. Repeat steps 3-7 for each affected ESXi host port channel

Note: This procedure causes a brief network disruption to the affected ESXi host. During outage recovery scenarios, this is the fastest resolution method and avoids host reboots.

To prevent recurrence:

• Configure LACP rate fast on switch interfaces for quicker convergence
• Implement switch boot delay timer if available on the platform
• Review and adjust LACP timeout values

If the error persists after following these steps, contact Broadcom Support for further assistance.

When opening a support request with Broadcom for this issue, provide:

• ESXi host names and management IP addresses
• Physical switch vendor, model, and firmware version
• Port channel configuration from switches
• LACP neighbor status output
• ESXi host network configuration details
• Time of power outage