Supermicro SAS-836TQ Sideband/SGPIO

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

adgenet

Member
Apr 12, 2016
40
17
8
So, I have an SAS-836TQ and up until last week, had everything connected with SFF-8087 <> SATA cables with flashed H310s (9211-8i).
These cables however didn't have the 8-pin SGPIO wiring.
I've just switched to an SAS3 HBA (Flashed M1215, 9340-8i) and this time, bought cables that include the 8-pin SGPIO block.
I've confirmed that the backplane is set to SGPIO by setting JP84 on the backplane to 1-2.
The 8-pin header from the cable has a tiny triangle marking pin 1, and I've aligned that according to pin1 on the backplane for the corresponding sideband header.
IMG_7175RS.JPG

Unfortunately it doesn't look like the connection does anything.
Trying to use locate from either sas3ircu, sg_ses, or ledctl doesn't do anything.
sas3ircu says it's blinking but doesn't actually blink.
sg_ses doesn't work at all, all the entries in /dev/bg are "disk device (not an enclosure)" when you run against them.

Supermicro provides a pinout description in the backplane manual, but there's no way to check it against the cable...supermicro_sideband.png

To further the confusion, someone asked on supermicro's support site about SGPIO, and received a response about various jumpers. However, none of these jumpers seem to actually exist:
JP76 pins 1-2
JP92 open
JP103 pins 1-2
JP104 pins 1-2
JP94 closed
FAQ Entry | Online Support | Support - Super Micro Computer, Inc.

While I know that it's sort of a useless feature to have given that I'm just passing disks to ZFS using IT mode, It's bugging me a lot.

Am I missing something obvious?
Does IT mode firmware not support SGPIO/enclosure connections?

I know that tons of people use Supermicro hardware with LSI controllers so I was hoping someone would have some insight into what's going on.
 

i386

Well-Known Member
Mar 18, 2016
4,221
1,540
113
34
Germany
I think I read in their documentation that Adaptec 12gb/s controllers and hbas support enclosure management only with SFF-8643 to SFF-8643/8087 links. Maybe this is also true for LSI based controllers.
 

adgenet

Member
Apr 12, 2016
40
17
8
I think I read in their documentation that Adaptec 12gb/s controllers and hbas support enclosure management only with SFF-8643 to SFF-8643/8087 links. Maybe this is also true for LSI based controllers.
Interesting, though I wonder why...
I feel like sideband is sideband. As long as the wiring is there, it should work.
Maybe they just say that because sff-8643 to sata with a sideband backplane is not common or standard for sas3?
 

Terry Kennedy

Well-Known Member
Jun 25, 2015
1,140
594
113
New York City
www.glaver.org
Am I missing something obvious?
Does IT mode firmware not support SGPIO/enclosure connections?

I know that tons of people use Supermicro hardware with LSI controllers so I was hoping someone would have some insight into what's going on.
I use the 9201-16i with the SAS-836A backplane (so, 8087-8087 cables) and everything worked out-of-the-box via:
Code:
sas2ircu 2 locate 1:0 on
The 9201-16i is the 3rd controller in my systems - the first is a PCIe SSD and the second is an external HBA for the tape library. That's why there's a "2" on the command line instead of "0".

Are you sure your SGPIO cable is pinned correctly?
 

adgenet

Member
Apr 12, 2016
40
17
8
No idea if the cable is pinned correctly.
All the sata connectors seem to work at least.
I'm tempted to try a different cable but that's extra money...

Anybody have a pinout for sff-8643? I want to check woth my multimeter. Searched everywhere but couldn't find one.
 
Last edited:

Terry Kennedy

Well-Known Member
Jun 25, 2015
1,140
594
113
New York City
www.glaver.org
No idea if the cable is pinned correctly.
All the sata connectors seem to work at least.
Well, the SATA pinout is a standard, so I'd expect that to work. I don't know about the SGPIO connector - my copy of SFF-8485 doesn't mention any specific pinout. Normally I would expect to see a split ribbon cable there, not individual wires - it is a lot easier for automated equipment to attach the ribbon cable to the multi-lane end than it is to attach individual wires.

Do you have a link for information on the cable you're using? If it says "Supermicro" in there, I'd give it a better chance of working than if it mentioned other brands but not Supermicro. My copy of the 836 manual says there are 10 pins for the SGPIO connector on the TQ, with 9 and 10 unused (Manual Version 2.1b, Page C-12) and a left-to-right connector (Page C-10). Your picture seems to show an 8-pin connector with another connector labeled "I2C#1 JP371". Normally, Supermicro doesn't use 3-digit connector numbers on their backplanes. Can you provide the full part number and revision of the backplane so I can look it up?
I'm tempted to try a different cable but that's extra money...
Let's make sure you're not "digging in the wrong place" first.
Anybody have a pinout for sff-8643? I want to check woth my multimeter. Searched everywhere but couldn't find one.
Take a look at Page 11 of this PDF document. It has a table with descriptions of all of the pins as well as the mapping to SFF-9401 (which is the main focus of that document). Page 10 has an expanded version of that table which also includes SFF-8087.
 

adgenet

Member
Apr 12, 2016
40
17
8
With the help of that document you linked, I was able to compare the wiring on the cable to what "should be" the pinout.
The cable is a no-name cable I managed to source locally. Looks like the standard china-made variety with the blue SATA wires.
Anyway, interestingly, all the pins for GPIO were wired, and in the right order, just "on the wrong side" (at least for the supermicro backplane).

Turns out, the top row of pins needs to be swapped to the bottom, and the bottom to the top:
SBx is the "sideband signal #, the A/B/C/D# refers to the pin# of the SFF-8643 connector the GPIO pin should be wired to.
Colors just happen to follow the cable I have.
Screen Shot 2016-11-18 at 22.02.00.png

Blinking now works correctly.
I wonder why the cable is manufactured this way.
Are Supermicro GPIO cables wired differently from other manufacturers or is this simply a mis-wired cable?
I'd be very interested to know.

Thanks for the help! I couldn't have done it without that PDF! I searched everywhere and could only come up with mechanical specs and diagrams of the actual connector - no pinouts. You'd think it would be easier to find.

In case you were still curious, the backplane is an SAS836-TQ Rev 3.2
The pin header "looks like" 10 pins as you mention according to the manual, but this is not really the case.
The PCB has solder pads for 10 pins, but pin headers for only the first 8 pins:
IMG_7177RS.JPG
 

Terry Kennedy

Well-Known Member
Jun 25, 2015
1,140
594
113
New York City
www.glaver.org
Turns out, the top row of pins needs to be swapped to the bottom, and the bottom to the top:
SBx is the "sideband signal #, the A/B/C/D# refers to the pin# of the SFF-8643 connector the GPIO pin should be wired to.
Colors just happen to follow the cable I have.
View attachment 3854

Blinking now works correctly.
I wonder why the cable is manufactured this way.
Are Supermicro GPIO cables wired differently from other manufacturers or is this simply a mis-wired cable?
I'd be very interested to know.
I expect they do what everybody else does, and whoever created the specs for your cable didn't realize they were looking at the backplane pinout instead of the cable pinout and simply mirror-flipped one side to the other.
Thanks for the help! I couldn't have done it without that PDF! I searched everywhere and could only come up with mechanical specs and diagrams of the actual connector - no pinouts. You'd think it would be easier to find.
Most of the specs (at least at draft level) are on the Seagate site. Once the spec gets ratified, the way to get the official spec is to send a boatload of money to the standards body. Fortunately, Seagate leaves the drafts around for people to find useful stuff in.
 
  • Like
Reactions: dontwanna and abq

giusesco

New Member
Jan 8, 2021
1
0
1
With the help of that document you linked, I was able to compare the wiring on the cable to what "should be" the pinout.
The cable is a no-name cable I managed to source locally. Looks like the standard china-made variety with the blue SATA wires.
Anyway, interestingly, all the pins for GPIO were wired, and in the right order, just "on the wrong side" (at least for the supermicro backplane).

Turns out, the top row of pins needs to be swapped to the bottom, and the bottom to the top:
SBx is the "sideband signal #, the A/B/C/D# refers to the pin# of the SFF-8643 connector the GPIO pin should be wired to.
Colors just happen to follow the cable I have.
View attachment 3854

Blinking now works correctly.
I wonder why the cable is manufactured this way.
Are Supermicro GPIO cables wired differently from other manufacturers or is this simply a mis-wired cable?
I'd be very interested to know.

Thanks for the help! I couldn't have done it without that PDF! I searched everywhere and could only come up with mechanical specs and diagrams of the actual connector - no pinouts. You'd think it would be easier to find.

In case you were still curious, the backplane is an SAS836-TQ Rev 3.2
The pin header "looks like" 10 pins as you mention according to the manual, but this is not really the case.
The PCB has solder pads for 10 pins, but pin headers for only the first 8 pins:
Hi adgenet,
you saved me, I was losing my mind. I had an original Supermicro cable that worked and another, no Supermicro, that didn't work.

I have two Supermicro CSE-M14TQC enclosure wired to Controller SAS D2607-a21 equivalent to LSI9211-8i and in Flash in IT Mode, and one Supermicro CSE-M35TQ enclosure wired to motherboard Supermicro X8DTE-F that have SGPIO connector separately from sata cables.
Total disks 5 x 3.5" + 8 x2.5"

It was impossible for me to check the wiring because I deals a SFF-8087 cable with sideband cable as in the picture. On the one hand it is impossible to understand the internal wiring.
Supermicro CBL-0097L-02 0.5m Mini-SAS (SFF-8087) to 4x SATA Internal Cable - is OK
Delock Cable Mini SAS SFF-8087 > 4 x SATA 7 pin + Sideband 1 m metal cod. 85682 - is WRONG

I resolved as you suggest, see photo.

Now all locate led blinking correctly.
unnamed.jpg


WhatsApp Image 2021-01-08 at 17.55.23.jpeg
For now it's a dirty job but it works, then I'll do a nicer harness.

Thanks thanks thanks