WSFC timeout during cluster failover

Products

• VMware vSphere ESXi

Issue/Introduction

Microsoft Windows Server Failover Clustering (WSFC) nodes using shared VMDKs on vSphere 8.x experience cluster timeouts or "Production SQL is down" symptoms following storage path redundancy degradation.

• SQL Cluster resources go offline.
• VMkernel logs show: Device <naa.id> performance has deteriorated or redundancy degraded.
• SCSI reservation conflicts observed during host rescan or path failover.

Environment

• VMware vSphere 8.x
• Clustered VMDKs on FC SAN
• Windows Server Failover Clustering

Cause

When storage paths fail or become degraded, the ESXi host attempts to scan LUNs for metadata. If these LUNs are held by a WSFC node via SCSI-3 Persistent Reservations and the "Perennial Reservation" flag is not set on the host, the scan operation hangs or delays, leading to cluster heartbeats failing.

Resolution

1. Identify all LUNs (NAA IDs) used for shared cluster disks.
2. Set Perennial Reservations on every ESXi host in the cluster: esxcli storage core device setconfig -d <naa.id> --perennial-reservation=true
3. Verify the setting: esxcli storage core device list -d <naa.id>
4. Ensure shared VMDKs are on a dedicated VMware Paravirtual SCSI controller with Physical Bus Sharing enabled.

Additional Information: Refer to Contact Broadcom Support if storage path redundancy remains degraded after host-level configuration. Refer to KB 206567 for log collection procedures.