False Positive "NCP Plugin Down" Alarm in NSX Manager UI despite healthy Clusters

Products

• VMware NSX

Issue/Introduction

• NSX Manager UI displays an "NCP Plugin Down" alarm for one or more container clusters. However, the environment remains fully functional with no service disruption.
• Verification steps show:
  • API Validation: Running GET /api/v1/systemhealth/container-cluster/ncp/status returns a Healthy status for all clusters.
  • Pod Status: Manual inspection confirms all NCP pods are in a Running state.
• The following exception is found in the NSX Manager logs (var/log/syslog):

2026-04-19T17:15:23.952Z INFO SHA-NCP-Thread NcpMonitorServiceImpl 5297 MONITORING [nsx@6876 comp="nsx-manager" level="INFO" subcomp="manager"] SHA ncp check cluster ###################################, name domain-c##:###################################, status UNKNOWN
2026-04-19T17:15:23.978Z WARN SHA-NCP-Thread ObjectsView 5297 TXEnd[TX[eb1c]] Aborted Exception
org.corfudb.runtime.exceptions.TransactionAbortedException: TX ABORT | Snapshot Time = Token(epoch=2889, sequence=8446891223) | Failed Transaction ID = 71cca879-4133-4201-3972-01188984eb1c | Offending Address = ########## | Conflict Key = ########## | Conflict Stream = nsx$NcpClusterSummary | Cause = CONFLICT | Time = 25 ms
        at org.corfudb.runtime.view.StreamsView.append(StreamsView.java:180) ~[?:?]
        at org.corfudb.runtime.view.StreamsView.append(StreamsView.java:233) ~[?:?]
        at org.corfudb.runtime.view.StreamsView.append(StreamsView.java:244) ~[?:?]
        at org.corfudb.runtime.object.transactions.OptimisticTransactionalContext.getConflictSetAndCommit(OptimisticTransactionalContext.java:223) ~[?:?]

Environment

VMware NSX 

Cause

This issue is caused by an occasional Corfu database update failure. A concurrent update conflict within the nsx$NcpClusterSummary stream prevents the Manager from clearing the alarm state, resulting in a false-positive notification in the UI despite the plugin being healthy.

Resolution

This is a known issue  and a fix will be included in a future NSX release. 

Workaround: 

To clear the stale alarm state, restart the proton service on the affected NSX Manager nodes:

1. Log in to the NSX Manager CLI as admin.
2. Restart the service by executing the following command: restart service manager