An attempt to delete an unused Edge node from an NSX-T Edge cluster fails with a "segment error" message. This occurs when a new Edge node is added to the cluster, and services are migrated to it without following the designated Edge replacement procedure. After manually moving services and attempting to remove the old Edge node from the cluster, the deletion fails.
An attempt to delete an Edge node (T1) fails due to a DHCP server still running on the node, which is connected to a segment that appears to be missing or untraceable in the configuration. This issue typically arises when a local test DHCP service is created and subsequently left in the configuration, causing it to persistently reference the Edge node. As a result, the Edge node deletion is blocked by the system due to the unresolved dependency.
Error Message:
"[Routing] Entity PolicyEdgeNode//infra/sites/default/enforcement-points/default/edge-clusters/{edge_cluster_id}/edge-nodes/{edge_node_id} cannot be deleted as it is being referenced by entity(s): LogicalDhcpServer/{dhcp_server_id} (Error code: 10026)"
Follow the steps in the VMware NSX documentation (link) to prepare the Edge node for removal, specifically addressing T1 DHCP and metadata proxy configurations.
Power off the old Edge node to initiate service failover to the newly configured standby Edge. This step ensures service continuity during the removal process.
After powering off the old Edge node, attempt to remove the node from the Edge cluster. If the deletion fails with an error related to DHCP or other services still referencing the Edge node, proceed with the following steps.
Delete the DHCP profile referencing the old Edge node. Use curl commands, as APIs are not preferred. If deleting the DHCP profile fails, verify that all services linked to the profile have been migrated to the new Edge and that no dependencies remain.
Re-attempt the Edge node deletion once the DHCP profile is successfully removed and no dependencies are detected.