While utilizing virtual volumes (vVols) within your VMware environment you notice that virtual machines (VMs) begin to report "LOGICAL BLOCK ADDRESS OUT OF RANGE" errors. This in turn may result in the VM crashing and/or reporting that the virtual disk(s) is no longer present.
This message is reported in the Guest OS logs, the ESXi host vmkernel logs and vvold log files:
Example log lines from a Linux Guest OS:
May 26 15:10:18 vm-name kernel: [9184116.167289] sd 4:0:0:0: [sdi] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE May 26 15:10:18 vm-name kernel: [9184116.167293] sd 4:0:0:0: [sdi] Sense Key : Illegal Request [current] May 26 15:10:18 vm-name kernel: [9184116.167296] sd 4:0:0:0: [sdi] Add. Sense: Logical block address out of range May 26 15:10:18 vm-name kernel: [9184116.167313] sd 4:0:0:0: [sdi] CDB: Write(10): 2a 00 00 48 f8 60 00 00 20 00 May 26 15:10:18 vm-name kernel: [9184116.167319] end_request: critical target error, dev sdi, sector 4782176
Example log line from the ESXi host vmkernel log:
2020-05-26T20:10:18.095Z cpu12:2097239)ScsiDeviceIO: 3449: Cmd(0x459a9b0cc2c0) 0x2a, CmdSN 0xfb from world 2102564 to dev "naa.624a937024c7e2ab435e44c8000113f4" failed H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x21 0x0.
Example log line from vvold:
2020-05-26T20:10:18.095Z cpu12:2097239)VVol: 2210: [vvolDev: rfc4122.ba996349-3524-4d04-g10r-9e94f374b599] WRITE command (ioFlags 0x111, cmd 0) failed: Not supported (0xbad0020)
This issue is specific to Cisco UCS servers utilizing nfnic drivers 184.108.40.206 - 220.127.116.11. In order to resolve the issue an upgrade to Cisco nfnic driver 18.104.22.168 or later.
For additional information please refer to Cisco CSCvr67972 for additional information on this issue.
This issue only impacts Cisco UCS servers utilizing nfnic drivers in ESXi 6.7+. The standard fnic driver is not susceptible to this issue.