NSX Edges Not Visible in UI After Upgrade to Version 9.1.0

Products

• VMware NSX

Issue/Introduction

• After completing an NSX upgrade from 4.2.x to 9.1.x, existing Edge nodes may not be visible on the UI under System > Fabric > Nodes.
• This issue typically impacts the first manager node upgraded (often the VIP node), while the remaining manager nodes in the cluster display the Edges correctly.
• Reviewing the search manager logs on the impacted NSX manager node at /var/log/search/search-manager.log reveals OpenSearch indexing errors and connection refusals, with symptoms including:

  2026-07-14T07:36:00.937Z ERROR NSX 8504 [nsx@4413 comp="nsx-manager" errorCode="MP60507" level="ERROR" logger="SearchServiceImpl" reqId="92f2####-9270-####-a840-cde####e7f7b" subcomp="manager" threadName="http-nio-127.0.0.1-7440-exec-24009" username="user\\/* AND resource_type:PolicyEdgeNode"}},{"bool":{}}],"must_not":[{"match":{"_meta.isOverridden":{"query":true}}}],"should":[{"bool":{"must":[{"query_string":{"query":"path.aggregation:\\/*"}}]}},{"bool":{"must":[{"bool":{"must":[{"query_string":{"query":"(!resource_type:(Alarm OR MonitoringEvent OR IDSEvent OR ConsolidatedRealizedStatus OR VirtualMachine OR VirtualNetworkInterface OR GenericPolicyRealizedResource) AND !(_exists_:path))"}}]}}]}}]}}
  
  2026-07-14T07:36:00.940Z ERROR NSX 8504 [nsx@4413 comp="nsx-manager" errorCode="MP60506" level="ERROR" logger="SearchServiceImpl" reqId="92f2####-9270-####-a840-cde####e7f7b" subcomp="manager" threadName="http-nio-127.0.0.1-7440-exec-24009" username="svc-###-nsx"] [QueryExecution: Execution] Exception while searching, query: dsl: null query: NOT path:\/global-infra\/* AND resource_type:PolicyEdgeNode\njava.net.ConnectException: Connect to http://localhost:9200 [localhost/127.0.0.1] failed: Connection refused
  
  2026-07-14T07:36:29.446Z ERROR NSX 8504 [nsx@4413 comp="nsx-manager" errorCode="MP60524" level="ERROR" logger="SearchServiceWatchDog" subcomp="manager" threadName="SearchHealthCheck"] [Search: WatchDog] OpenSearch health status is RED

• However, the NSX Manager cluster will be stable on all processes, including Search.

Environment

VMware NSX 9.1.0

Cause

In the logical migration during the upgrade path from 4.2.x to 9.1.x, the Edge Transport Node (TN) path changes from */edge-transport-node/* to */edge-transport-nodes/*. As part of the NSX 9.1 hitless upgrade design, the first node to be upgraded intentionally skips rebuilding its search index.

Consequently, this first node's search index is left with a stale Edge TN path within the PolicyEdgeCluster Data Transfer Object (DTO). When the UI attempts to perform a join operation between PolicyEdgeCluster and PolicyEdgeTransportNode, the edges with the stale path fail to populate.

Resolution

The issue is planned to be resolved in upcoming releases.

Workaround

• Log in to the impacted NSX manager node via SSH or console as an admin user.
• Execute the following command to manually rebuild the search index: start search resync all
  (Note: The UI will be unavailable to the user during this resync period, and the duration will depend on the number of objects in the environment.)