etcdserver: user name already exists

Products

• VMware vCenter Server

Issue/Introduction

You are seeing log errors similar to - 

{index=db-vmware_263, message=<5>2025-11-14T17:11:21Z <hostname> clusterAgent[########]: ERROR Failed in call to etcd {"opID": "NAM-domain-c######-####-####-####", "cli": "0xc0002bc4e0", "func": "UserAdd", "error": "etcdserver: user name already exists", "error value": "rpctypes.EtcdError{code:0x9, desc:\"etcdserver: user name already exists\"}", "username": "####-#####-###-###.databank.host"}, fields={gl2_accounted_message_size=589, protocol=1, level=6, port=55028, gl2_remote_ip=##.###.###.###, gl2_remote_port=45186, gl2_message_id=####################, gl2_source_node=#######-####-####-####-############, type=vmware-esxi, gl2_source_input=##################, facility=gelf-rb, tags=vmware-esxi, _geoip_lookup_failure}, id=#######-####-####-####-############, timestamp=2025-11-14T17:11:21.202Z, source=####-#####-###-###.databank.host, stream_ids=[##################]}

Environment

vCenter 8.x

Cause

DKVS (Distributed Key-Value Store) is a feature where vCenter picks a subset of hosts in each inventory cluster to directly form a cluster of Etcd instances, then it uses that database to store some resilient config data.

In this case, vCenter is trying to add a host to a cluster that it had already added before. Which means that it could be retrying a failed addition, or that a removal had failed in the past. But in isolation, the ClusterAgent (our per-host control plane service) will just forward the UserAlreadyExist error back to vCenter, which will notice that case and recreate the user so that it has a fresh password, then try again.

Resolution

Contact Broadcom support for further assistance