In the vxconfigd debug logs a single set of messages is seen which is the disk claiming process. In the dmpevents logs file, the MAINTENANCE IN request that is sent to the device is treated as an illegal request by the eUSB disk, which in turn causes DMP to toggle the disk between Disabled and Enabled states.
Error from dmpevents.log:Sun Oct 23 19:07:37.772: SCSI error occurred on Path c2t0d0s2(214/66): opcode=0xa3 reported illegal request (status=0x2, key=0x5, asc=0x20, ascq=0x0) invalid command operation code
Interpretation of above line:Field Value Meaning
Opcode 0xa3 MAINTENANCE IN
Status 0x2 CHECK CONDITION
Key 0x5 Illegal Request
ASC 0x20 Invalid Command Operation Code
According to the dmpevents log message, MAINTENANCE IN Opcode is not supported by the eUSB disk.
/var/adm/messages:
Oct 24 03:15:04 sclt7-02 vxvm:vxconfigd: [ID 702911 daemon.error] V-5-1-13824 scsi3_claim_device: evpd inquiry failed- I/O error: I/O error Oct 24 03:15:04 sclt7-02 vxdmp: [ID 581508 kern.notice] NOTICE: VxVM vxdmp V-5-0-34 [Info] added disk array ALUAdisk, datype = aluadisk Oct 24 03:15:04 sclt7-02 vxdmp: [ID 681931 kern.notice] NOTICE: VxVM vxdmp V-5-0-148 [Info] enabled path 214/0x40 belonging to the dmpnode 323/0x38 Oct 24 03:15:04 sclt7-02 vxdmp: [ID 397033 kern.notice] NOTICE: VxVM vxdmp V-5-0-147 [Info] enabled dmpnode 323/0x38 Oct 24 03:15:09 sclt7-02 vxvm:vxconfigd: [ID 702911 daemon.error] V-5-1-13824 scsi3_claim_device: evpd inquiry failed- I/O error: I/O error Oct 24 03:17:34 sclt7-02 vxdmp: [ID 241556 kern.notice] NOTICE: VxVM vxdmp V-5-0-0 [Info] removed disk array ALUAdisk, datype = aluadisk Oct 24 03:17:34 sclt7-02 vxdmp: [ID 382146 kern.notice] NOTICE: VxVM vxdmp V-5-0-112 [Warn] disabled path 214/0x40 belonging to the dmpnode 323/0x38 due to path failure Oct 24 03:172:34 sclt7-02 vxdmp: [ID 381736 kern.notice] NOTICE: VxVM vxdmp V-5-0-111 [Warn] disabled dmpnode 323/0x38e
# vxdisk path | grep aluadisk
c2t0d0s2 aluadisk0_0 - - ENABLED
The eUSB disk supports SCSI Inquiry Opcode though, the vxscsiinq diagnostic utility can be used to send the SCSI Inquiry Opcode.
(Please check these only when dmpnode is enabled, if the dmpnode is disabled, vxscsiinq will fail.)
# /etc/vx/diag.d/vxscsiinq /dev/vx/rdmp/aluadisk0_0Inquiry for /dev/vx/rdmp/aluadisk0_0, evpd 0x0, page code 0x0
Vendor id : MICRON
Product id : eUSB DISK
Revision : 1112
The eUSB device is embedded in SPARC T7/M7/S7 as default, it cannot be removed.
The eUSB device doesn't support the SCSI MAINTENANCE IN Opcode.
Avoid the issue by excluding the path for eUSB device in vxdmpadm.
1. ‘prtconf –v’ shows eUSB disk as scsi class and sd device:
# prtconf -v | grep -i eusb value='Real SSD eUSB 2GB'
value='id1,sd@SMICRON__eUSB_DISK_______14F0022701018090'
value='eUSB DISK' location: /dev/chassis/SYS/MB/EUSB-DISK/disk
2. # echo | format 5. c2t0d0
/pci@308/pci@1/usb@0/hub@1/storage@3/disk@0,0 <<<<<<<<<<<<<<<<<<
/dev/chassis/SYS/MB/EUSB-DISK/disk
3. # vxdmpadm exclude path=c2t0d0
The new Oracle Solaris 11 boot process includes a new concept called a boot pool. A boot pool is a boot device that is used to store boot archives. In SPARC T7 processor–based servers, one or more embedded USB (eUSB) storage devices are grouped together to form a boot pool, making the boot pool accessible to the OpenBoot PROM firmware.
The eUSB storage is an internal USB flash memory device that is installed into the system at the factory. SPARC T7-1, T7-2, and T7-4 servers include one eUSB device. SPARC M7-8 and M7-16 servers include one eUSB device for every CPU, memory, and I/O unit (CMIOU) chassis board, allowing the boot pool to consist of multiple, striped eUSB storage devices.
The device is a USB device, and it's Embedded USB (eUSB) storage device that supports booting over the InfiniBand network on SPARC T7/M7/S7.
Additionally, the device is automatically detected in dmp as ALUAdisk.
From the docs.infoscale.com - HCL list, This MICRON-eUSB devices are not supported by our product.
(https://docs.infoscale.com)
ETrack: 3901850