Emulate the typical LED and alarms one would see with a hardware-based RAID controller so an appliance that uses software-based RAID has same look-and feel …

• Sound the audible alarm once mdstat indicates a logical disk becomes critical.
• Light the FAULT LED
• Query the O/S to find out the surviving disks, then set them to blink. This way user knows what disks are being rebuilt.
• Use the ID LED to indicate which hot spare the O/S chose to rebuild the LUN with.

Make pretty patterns with the LEDs in a RAID array for trade shows.

Mute audible alarms that drive people nuts, especially ones that indicate a fault that can’t be addressed until next week some time.

Query enclosure to determine the serial number or WWN of a device given the row/col of a slot or vise-versa.

Test all SES components to just make sure they work, without having to manually break things or unplug power supplies and/or disk drives.

Notes
SES will let you light the LED(s) associated with the SLOT the drive is installed in. This is a better mechanism then lighting the LED on a disk drive, since the technique will always work, even if the disk drive is dead or dying.

root@morph smartmon]# ./smartmon-ux -E+ /dev/sdc
SMARTMon-ux [Release 1.23, Build 30-NOV-2003] - Copyright 2003 SANtools, Inc. http://www.SANtools.com SEAGATE ST336753FC S/N "3HX00LE3" on /dev/sdc [SES] (Not Enabling SMART)(35003 MB)
Discovered XYRATEX  RS1600-FC2-FFX2  WWN=20-00-00-50-CC-00-7B-8E:
Configuration switches numbered from 1-12 as viewed from rear, top to bottom
Vendor-specific features (Notes) [SWITCH SETTING]
SoftSelect Mode             (Disabled) [SW11-OFF]
Drive Speed                 (2 Gbit FC Mode)
Loop Config                 (1 x 16 loop) [SW1-ON]
Hub Mode                    (Enclosure in hub mode) [SW3-ON]
Reserved                    (reserved) [SW4-OFF]
SES Report                  (REPORT bit set on single)
Power Redundancy Indication (Enclosure indicates redundancy)
Ops Panel Muted Mode        (Enclosure in REMIND mode)
Drive Addressing Mode:      0 (1 x 16 JBOD)
Unit Select Switch:         1
Model is: Goshawk - Mylex FFX2 RAID 2Gbit dual port controller
Master LRC Firmware level:  35
SFP Host 0 Present (LoopA): YES
SFP Host 0 Good (A):        NO
SFP Host 1 Present (A):     YES
SFP Host 1 Good (A):        NO
SFP Expansion Present (A):  NO
SFP Expansion Good (A):     NO
SFP Host 0 Present (LoopB): YES
SFP Host 0 Good (B):        NO
SFP Host 1 Present (B):     NO
SFP Host 1 Good (B):        NO
SFP Expansion Present (B):  YES
SFP Expansion Good (B):     NO
Device #0 OK SelID=04h [Row=1 Col=1]
Device #1 Not Installed SelID=05h [Row=1 Col=2]
Device #2 OK SelID=06h [Row=1 Col=3]
Device #3 Not Installed SelID=07h [Row=1 Col=4]
Device #4 Not Installed SelID=08h [Row=2 Col=1]
Device #5 Not Installed SelID=09h [Row=2 Col=2]
Device #6 Not Installed SelID=0ah [Row=2 Col=3]
Device #7 Not Installed SelID=0bh [Row=2 Col=4]
Device #8 Not Installed SelID=0ch [Row=3 Col=1]
Device #9 Not Installed SelID=0dh [Row=3 Col=2]
Device #10 Not Installed SelID=0eh [Row=3 Col=3]
Device #11 Not Installed SelID=0fh [Row=3 Col=4]
Device #12 OK SelID=10h [Row=4 Col=1]
Device #13 Not Installed SelID=11h [Row=4 Col=2]
Device #14 OK SelID=12h [Row=4 Col=3]
Device #15 OK SelID=13h [Row=4 Col=4]
Power Supply #0 Critical DC Undervoltage AC failure DC failure [LED ON]
Power Supply #1 OK
Cooling Element #0 OK fan at speed 4
Cooling Element #1 OK fan at speed 4
Temperature Sensor #0 OK 104F/40C
Audible Alarm #0 OK ENABLED sounding CRITICAL
SESElectronics Processor #0 OK [ACTIVE]
SESElectronics Processor #1 OK [PASSIVE]
 
Threshold Information                   Warning Range   Critical Range
Temperature Sensor #0:                    30 - 74          20 - 78
 
Module Locations - Front View
 Col-1     Col-2     Col-3     Col-4
+--------------------------------------+
|Dev #00 | Dev #01 | Dev #02 | Dev #03 | Row-1
|Dev #04 | Dev #05 | Dev #06 | Dev #07 | Row-2
|Dev #08 | Dev #09 | Dev #10 | Dev #11 | Row-3
|Dev #12 | Dev #13 | Dev #14 | Dev #15 | Row-4
+--------------------------------------+
 
Module Locations - Rear View
+--------------------------------------+
| PSU /    | 2nd | 1st |OPS  | PSU /   |
| Cooling  | LRC | LRC |Panel| Cooling |
| First    | 'B' | 'A' |     | Second  |
|   #0     | #1  | #0  |     |  #1     | <--SES ID#
+--------------------------------------+
 
Legend for Below: SN=Serial#, (optional)SC=Status Code
 LRC-A: SN=PMT317000005619
 LRC-B: SN=PMT317000005396
 Power Supply#1: SN=IMS4204300008BB
 Power Supply#2: SN=IMS4204300007F7
 
Program Ended.

The command set is defined as an ANSI specification, and utilizes the SCSI commands to obtain configuration information for the enclosure as well as reprogram it to light external indicator LEDs or audible alarms to inform the user about the health and maintenance information for attached RAID controllers and disk drives.

An enclosure services device (circuit board with CPU) typically manages all these enclosure elements (fans, power supplies, etc…) and communicates with the processor via application software. 

Note that the “SCSI” in SCSI Enclosure Services refers to the protocol, and not any type of physical connection.  Enclosures that connect via Fibre Channel and SAS cabling are classified as SCSI devices. They speak the SCSI command set, and use a serial SCSI interface. Traditional “SCSI” disk drives also use the SCSI protocol, but use parallel SCSI connectors.

SES devices will generally have the ability to flash firmware updates, but the mechanism for doing this is sometimes vendor/product specific.

ANSI-Defined SES Element Types and Description Table

Below is a table from the ANSI SES programming specification which shows how one might package the bytes to control aspects of a device.  We will send a harmless command which will enable the fault light for a device in a particular slot. Every element type has a different 4-byte structure and options, so you should consult either the ANSI programming specification or your particular vendor’s documentation.  Remember, an enclosure manufacturer is free to not support certain functions as well as add vendor-unique functionality.

To enable the request fault light, we must set bit 5 in byte #3 (i.e., 20 hex), so the wwxxyy sequence must be 000020.

As we are controlling the device element, we must send a 01 to indicate a disk device.  For our example, we’ll select the third device in the enclosure (corresponding to element # 2).

Put it all together, and you would send out -EP20102000020.

If we were to send out -EP20102000200, this will turn off the fault light, but turn on the identify light (assuming one exists).  Note that the fault light goes off because byte 3 (the yy field) has all zeros in it.  The SES enclosure will stay in whatever state you put it in, until either the enclosure decides to override that state or power is reset to the enclosure.  Everything is volatile.  (There may be some exceptions for vendor-unique SES elements).

The SES specification has a great deal of intelligence, which allows for not only reporting health, but changing fan speeds, manipulating the LEDs, and controlling audible alarms.   SAF-TE is a rather dumb interface that reports the most basic information in more of a pass/fail form.