Random storage path loss occurs when using HP blades and Qlogic FCoE adapters using the qedf driver

Products

• VMware vSphere ESXi
• VMware vSphere ESXi 6.0
• VMware vSphere ESXi 7.0
• VMware vSphere ESXi 8.0

Issue/Introduction

- Storage paths go dead and then recovery almost immediately
- /var/log/vmkernel.log will have entries similar to this:

2024-07-11T17:24:29.847Z cpu39:2097273)qedf:vmhba65:qedfc_scsi_completion:1826:Error: dropped frame detected xid[(0x2da]  lba=0x1000 lbc=0x1 cmd 28:0:0:0:10 (512 of 512 bytes).
2024-07-11T17:24:29.865Z cpu39:2097595)qedf:vmhba65:qedfc_scsi_completion:1826:Error: dropped frame detected xid[(0x216]  lba=0x1000 lbc=0x1 cmd 28:0:0:0:10 (512 of 512 bytes).
2024-07-11T17:24:32.197Z cpu39:2098160)qedf:vmhba65:qedfc_scsi_completion:1826:Error: dropped frame detected xid[(0x42b]  lba=0x1000 lbc=0x1 cmd 28:0:0:0:10 (512 of 512 bytes).

- The "Dumped frames" and "Invalid Tx Word Count" counters from "localcli storage san fc stats get" will both have a value other than zero and likely the exact same value:

FcStat:
   Adapter: vmhba65
   Tx Frames: 13962354945
   Rx Frames: 186712322673
   Lip Count: 0
   Error Frames: 0
   Dumped Frames: 41759881
   Link Failure Count: 0
   Loss of Signal Count: 0
   PrimSeq Protocol Err Count: 0
   Invalid Tx Word Count: 41759881
   Invalid CRC Count: 0
   Input Requests: 0
   Output Requests: 0
   Control Requests: 0

Cause

There is a bug in older firmware on the HPE Virtual Connect Module that is causing the Invalid Tx Word Count and Dumped frames

Resolution

HPE recommends to apply Synergy Service Pack SY-2024.07.02 on the blade enclosures which will update the Virtual Connect Module firmware. Please contact HPE support for assistance.