Symptoms
Affected version:
PowerStoreOS 2.0.0.0 Build 1376722
This issue is observed when PowerStoreOS is upgraded to 2.0.0.0 Build 1376722 or a node is rebooted or restarted with PowerStoreOS 2.0.0.0 Build 1376722.
During an NDU, or when a node is rebooted or restarted, hosts lose access to any LUNs that are configured as LUN 0. All other LUNs are not be affected. In order to regain access to LUN 0, the host must be logged out and logged back in again.
If a LUN 0 is provisioned to the host, the host loses access to LUN 0 after nondisruptive upgrade (NDU) to PowerStoreOS version 2.0.0.0 Build 1376722.
Nondisruptive Upgrade (NDU) to PowerStoreOS version 2.0.0.0 Build 1376722 causes data unavailable situation for a LUN 0.
The connectivity to a LUN 0 on a PowerStore node cannot be restored after node is rebooted.
If ESX host uses LUN 0, in the /var/log/vmkernel.log file on the ESXi host, entries are seen similar to the below:
NMP: nmp_ThrottleLogForDevice:3861: Cmd 0xa3 (0x4598c9d3e8c0, 0) to dev "naa.68ccf098001eb7ef7b75d019f90f1ff2" on path "vmhba64:C0:T9:L0" Failed:
NMP: nmp_ThrottleLogForDevice:3869: H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x25 0x0. Act:NONE. cmdId.initiator=0x45388721bbd8 CmdSN 0x0
Cause
Code defect in PowerStoreOS 2.0.0.0 Build 1376722.
Even after node is rebooted or restarted, PowerStore keeps responding with ASC: 25h ASCQ 00h (LOGICAL UNIT NOT SUPPORTED) to host I/Os for LUN 0 that can trigger a Permanent Device Loss (PDL) condition.
Resolution
Workaround:
As a workaround, all host and host groups should not have a PowerStore LUN with LUN ID 0.
Permanent fix:
This issue is addressed in PowerStoreOS 2.0.0.0 (Build: 1397847).