RAV Migration Fails in HCX Due to Asymmetric Routing and Physical Network Packet Drops
search cancel

RAV Migration Fails in HCX Due to Asymmetric Routing and Physical Network Packet Drops

book

Article ID: 434872

calendar_today

Updated On:

Products

VMware HCX

Issue/Introduction

Symptoms:

  • VMware HCX Replication Assisted vMotion (RAV) migrations fail, whereas standard vMotion migrations complete successfully.
  • HCX is configured with two network profiles as below:
    1. Subnet X with tag "Management", "HCX Uplink", "vSphere Replication"
    2. Subnet Y with tag "vMotion"
  • Packet captures reveal that the replication TCP session with port 31031 fails to establish between ESXi and HCX IX appliance, workload VM is in the mentioned ESXi. The physical network drops the TCP SYN ACK packets sent from the HCX Interconnect (IX) appliance, as well as the ESXi host's subsequent TCP ACK and retransmissions.

Environment

VMware HCX 4.x
VMware vSphere 8.x

Cause

The failure is caused by asymmetric routing on the HCX IX appliance.

The ESXi replication vmkernel interface (Subnet Y) initiates a TCP SYN to the HCX IX appliance's vNic_0 (Subnet X). Because the IX appliance possesses a second interface, vNic_1(Subnet Y), which is directly connected to Subnet Y, its internal routing table prefers this direct connection for the return path.

The IX appliance transmits the TCP SYN ACK reply out of vNic_1. The resulting Ethernet frame contains the Source IP address of vNic_0 but the Source MAC address of vNic_1. Physical network detect this MAC/IP mismatch and drop the packets.

Resolution

Realign the HCX network configuration to ensure symmetric routing for replication traffic.

  1. Log in to the HCX Manager UI.

  2. Navigate to Infrastructure > Network Profiles.

  3. Modify the configuration to remove the "vSphere Replication" tag from the Subnet X profile and attach this tag to the Subnet Y profile (associated with vNic_1).

  4. Navigate to Infrastructure > Compute Profiles.

  5. Edit the active Compute Profile to map the Replication traffic to the updated vNic_1 Network Profile.

  6. Navigate to Infrastructure > Service Mesh.

  7. Select the affected Service Mesh and click Resync to push the updated symmetric vNIC configuration to the IX appliance.

  8. Retry the RAV migration.
Powered by Wolken Software