Can someone help me understand HDD lights on hot swap cases?

[scurrier]

I am considering the X10SL7-F board to put in various random eBay cases with hot swap drive bays.

How can I ensure that a hot swap case + motherboard, chosen independently, will support all of the HDD lights on the front of the case?

It seems that this has something to do with SGPIO pins, but I don't understand those. Can someone give me a brief overview?

Is there anything else important in ensuring compatibility between a board and drive bays?

 

[Z300M]

I am considering the X10SL7-F board to put in various random eBay cases with hot swap drive bays.

How can I ensure that a hot swap case + motherboard, chosen independently, will support all of the HDD lights on the front of the case?

It seems that this has something to do with SGPIO pins, but I don't understand those. Can someone give me a brief overview?

Is there anything else important in ensuring compatibility between a board and drive bays?

What cases, for example? Of all the cases I've ever owned, only one had more than one HDD light. I used one connected to the motherboard HDD pins and the other connected to the HDD pins on a PCI SCSI card.

I have that motherboard on order -- should be here tomorrow. I see nothing in the manual to suggest that it has more than one set of pins for HDD lights.

 

[scurrier]

Hi Z300M. Check the manual again and search for SGPIO. You will find the pins I'm talking about.

Maybe I'm over complicating this and you really just connect the SGPIO pins on the board to some SGPIO pins on the chassis hot swap backplane.

I agree that consumer grade cases have only one light, usually. But many server grade chassis have individual lights for each HDD. Look at supermicro chassis for an example.

 

[Z300M]

Hi Z300M. Check the manual again and search for SGPIO. You will find the pins I'm talking about.

Maybe I'm over complicating this and you really just connect the SGPIO pins on the board to some SGPIO pins on the chassis hot swap backplane.

I agree that consumer grade cases have only one light, usually. But many server grade chassis have individual lights for each HDD. Look at supermicro chassis for an example.

OK, I see now. I assume that if you have a case with connectors for that many drives you can connect them to those motherboard pins.

I am using what is intended as a gaming case, but with iStarUSA drive cages each holding five 3.5" drives in the space of three 5.25" drive bays. Each drive has its own power/activity LED that changes color when the drive is accessed -- no LED connections to the motherboard. Perhaps all drive cages work that way.

 

[scurrier]

So theres no connection between the cage and the mobo/HBA besides the data cable? And the lights still work?

I wonder why SGPIO pins are necessary at all, then?

I also wonder how the red "drive failure" lights work on certain systems in terms of wiring. And what part of the system triggers the light to turn on.

 

[Z300M]

So theres no connection between the cage and the mobo/HBA besides the data cable? And the lights still work?

Only the data cable, and the lights still work.

I wonder why SGPIO pins are necessary at all, then?

I also wonder how the red "drive failure" lights work on certain systems in terms of wiring. And what part of the system triggers the light to turn on.

Maybe the SGPIO pins and the drive failure feature are for some Supermicro-specific devices in their server chassis.

 

[jyavenard]

I am considering the X10SL7-F board to put in various random eBay cases with hot swap drive bays.

How can I ensure that a hot swap case + motherboard, chosen independently, will support all of the HDD lights on the front of the case?

It seems that this has something to do with SGPIO pins, but I don't understand those. Can someone give me a brief overview?

Is there anything else important in ensuring compatibility between a board and drive bays?

If you use FreeNAS or FreeBSD, then it won't.
There's no SGPIO drivers.
The only LED that is going to work is the activity LED.

With Linux it's another matter, there are various ways to do it.

 

[jyavenard]

So theres no connection between the cage and the mobo/HBA besides the data cable? And the lights still work?
I wonder why SGPIO pins are necessary at all, then?

The SGPIO interface allows you to control the other LEDs on the chassis, one per disk (or two depending on the backplanes). So you for example turn on a LED when you're trying to locate a drive.

 

[scurrier]

Shame on me for thinking that this topic would be too obscure for a wikipedia article.
http://en.wikipedia.org/wiki/SGPIO

Judging from that, it seems like you basically plug it in and it works. I'm not 100% confident on that.

 

[scurrier]

Thanks jyavenard. So it sounds like I should not concern myself with the SGPIO for my FreeNAS build.

 

[AlVal]

does this not suggest sgpio drivers do exist for freebsd? I guess there's a difference between saying they exist for freebsd and they exist for freenas?

https://www.freebsd.org/cgi/man.cgi?query=ahci(4)&sektion=

 

[Ericloewe]

does this not suggest sgpio drivers do exist for freebsd? I guess there's a difference between saying they exist for freebsd and they exist for freenas?

https://www.freebsd.org/cgi/man.cgi?query=ahci(4)&sektion=

That's FreeBSD 10 and thus may not apply to FreeNAS 9.X.

Additionally, the driver itself doesn't seem to do anything beyond provide the interface, it would have to be separately coded. It also only works with AHCI controllers, which means SAS controllers would need separate support.

 

[jgreco]

Actually getting the details right for a lot of this stuff is a frustrating and annoying experience which is why businesses typically outsource this to a company that specializes in making stuff work right. As someone who owns such a company, I'll just say that a lot of it is relative magic, ranging from software management and mapping out drives to building custom printed circuit boards to act as interface between host and backplane. The latter we haven't had to do with the recent popularity of smarter backplanes, thankfully.