In an NSX Federation environment, a high-severity alarm for "RTEP connectivity lost" remains stuck in an OPEN state within the NSX Manager dashboard.
The specific Edge node or Transport Node UUID referenced in the alarm is no longer present, having been previously decommissioned or deleted from the infrastructure inventory.
In the NSX Manager monitoring logs, you observe log entries similar to the following:
NSX 76534 MONITORING [nsx@6876 alarmId="68c4####-####-4f4a-####-4c3ec905####" alarmState="OPEN" comp="nsx-manager" entId="03c3####-####-4c3f-####-ccc506e6####" errorCode="MP701099" eventFeatureName="federation" eventSev="HIGH" eventState="On" eventType="rtep_connectivity_lost" level="ERROR" nodeId="3a99####-####-c3f0-####-93058f2a####" subcomp="monitoring"] Edge node edge#### lost RTEP (Remote Tunnel Endpoint) connectivity with remote location ##-###-###.
This issue occurs due to a database cache desynchronization within the NSX Manager core inventory service (Proton).
To flush the stale database cache and clear the orphaned alarm, you must perform a rolling restart of the management service across the NSX Manager cluster nodes.
IMPORTANT: Restarting the Proton service will temporarily disrupt access to the NSX Manager User Interface (UI) and the REST API for approximately 3 to 5 minutes per node. However, this management plane operation does not impact active data plane network traffic or VM packet forwarding.
Step-by-Step Instructions
1. Open an SSH session to the primary NSX Manager node displaying the open alert.
2. Log in with root privileges.
3. Restart the core management daemon by executing the following command: /etc/init.d/proton restart
4. Wait approximately 5 minutes for the API and UI architectures to fully reinitialize.
5. Log back into the NSX Manager web interface and verify that the stuck RTEP connectivity lost alarm has successfully cleared.