High Latency Accessing Shared Storage using iSCSI adapters.

Products

• VMware vCenter Server

Issue/Introduction

 

Intermittent storage connectivity or "Path Selection" flapping.

vmkernel: cpu7:2097889)NMP: nmp_ThrottleLogForDevice:3893: Cmd 0x89 (0x45ba69d10a40, 2097176) to dev "naa.6xxxxxxxxxxxxxxxxx" on path "vmhba64:C4:T0:L2" Failed:
vmkernel: cpu7:2097889)NMP: nmp_ThrottleLogForDevice:3898: H:0x0 D:0x2 P:0x0 Valid sense data: 0x6 0x29 0x0.  
vmkernel: cpu7:2097889)ScsiCore: 2000: Power-on Reset occurred on naa.6xxxxxxxxxxxxxxxxxxx
 

Transient storage conditions leading to latencies observed on the virtual machines

vmkernel: cpu5:2106690)VMW_SATP_ALUA: satp_alua_issueCommandOnPath:966: Path (vmhba64:C5:T0:L2) command 0xa3 : Failed with transient error status Transient storage condition, suggest retry. sense data: 0x6 0x29 0x0.
vmkernel: cpu5:2106690)VMW_SATP_ALUA: satp_alua_issueCommandOnPath:972: Path (vmhba64:C5:T0:L2) command 0xa3 : Waiting for 20 seconds for the transient error to change before marking the path down

 

Storage arrays failing to map LUNs to specific hosts due to initiator conflicts.

Environment

ESXi 8.X and above

Cause

The software iSCSI initiator name was likely modified manually or cloned during a host deployment without refreshing the unique identifier. By default, VMware generates a unique IQN; however, manual overrides do not automatically update when a host is renamed or cloned.

- hostname: esxi01.local  
- SW iSCSI initiator name
  iqn.1998-01.com.vmware:esxi01:123456789:64
  
- hostname: esxi02.local  
- SW iSCSI initiator name
  iqn.1998-01.com.vmware:esxi01:123456789:64

Resolution

Resolution Steps
    1. Enter Maintenance Mode: Ensure all VMs are migrated off the affected host (esxi0X.local).
    2. Navigate to Storage Adapters: Go to Configure > Storage > Storage Adapters.
    3. Identify Adapter: Select the Software iSCSI adapter (e.g., vmhba64).
    4. Edit IQN: Under Network Port Binding or Properties, select Edit next to the IQN.
    5. Revert to Default: * Uncheck "Enter Name Manually" to allow the system to generate a unique string based on the current hostname.
        ○ Alternatively, manually update the string to reflect the correct host: iqn.1998-01.com.vmware:esxi0X:[unique-id].
    6. Re-scan Storage: Perform a Rescan of the Storage Adapter and the VMFS volumes.
    7. Storage Array Update: Coordinate with the storage team to ensure the new IQN is correctly mapped in the storage array’s host group/initiator group.

Prevention & Best Practices
    • Verification: Always verify that the IQN matches the host’s naming convention during the post-build validation phase.
    • Automation: If using Host Profiles, ensure the iSCSI configuration is set to "User-specified" or "Determine dynamically" to avoid pushing a static IQN from the reference host to the entire cluster.