ESXi Host Flapping in vSAN ESA Cluster After SDDC Upgrade Due to ENS Mismatch

Products

• VMware vSAN

Issue/Introduction

After an SDDC upgrade (from ESXi build 25067014 to build 25429399), an ESXi host in a vSAN ESA stretch cluster exhibits persistent network and cluster flapping, continuously disconnecting and reconnecting.

Associated symptoms:

• Virtual machine performance degradation, frozen or unresponsive VMs
• vMotion failures reporting "Failed waiting for data. Error bad0021. Timeout." 
• Recurring ENS: Ens_CreateSwitch activity indicating a failure to correctly initialize the Enhanced Networking Stack (ENS) on the host.

Errors for the related vMotion and ENS activity are found in ESXi vmkernel.log:  

• • 2026-08-05T09:13:12.479Z Db(167) Hostd[2102403] [Originator@6876 sub=Vigor.Vmsvc.vm:/vmfs/volumes/[vsan:UUID]/[Name Space UUID]/[VM Name].vmx] VMotionPrepare: MigrateFromDest message: Failed waiting for data.  Error bad0021. Timeout. 

 

• • 2026-08-04T09:41:02.913Z In(182) vmkernel: cpu26:2097583)ENS: Ens_CreateSwitch:2478: Create ENS switch DvsPortset-0: maxPorts 128, swID 0, mode: polling

 

Environment

• Product: VMware SDDC Manager / NSX
• Version: SDDC upgrade from ESXi build 25067014  to 25429399
• Configuration: vSAN ESA Stretch Cluster

Cause

The issue is caused by a configuration discrepancy in the Enhanced Networking Stack (ENS) where physical network adapters report ENS Capable: True but ENS Driven: False. This mismatch prevents the host from successfully completing the ENS switch creation required for network traffic, leading to the observed flapping and communication timeouts.

Resolution

To resolve the issue, perform the following steps:

1. Isolate the affected host: Power off the flapping host to stabilize the cluster and restore VM responsiveness.
2. Verify NIC status: Run the command esxcfg-nics -e on the affected host. Compare the output with healthy nodes in the cluster. Identify adapters where ENS Capable is True but ENS Driven is False.
3. Review logs: Check the /var/log/vmkernel.log for ENS: Ens_CreateSwitch errors and vMotion failure logs (e.g., "Failed waiting for data. Error bad0021") to confirm the datapath initialization failure.
4. Reconcile ENS Configuration: Engage NSX support to verify transport node configuration state. It may be necessary to remove the host from the NSX transport node configuration and re-apply it to force a re-initialization of the ENS stack.
5. Validate: Once the ENS status is corrected (all relevant adapters show ENS Driven: True), perform a controlled reboot of the host and verify cluster stability.

Additional Information

This behavior is a known post-upgrade configuration inconsistency where the ENS datapath fails to attach to the physical NICs after an SDDC infrastructure upgrade. For additional context on NSX configuration and troubleshooting, refer to KB 448955.