Closed mscdex closed 1 month ago
Update: I think what's happening now is the backplane is wired up starting with the third onboard SATA connector and so the status LEDs being changed are offset by two.
However, the problem still remains for the second AMD controller. This particular motherboard (H12SSL-i) has eight onboard SATA ports, two 7/8-pin SGPIO connectors, and one SlimSAS x8 connector.
Four of the onboard SATA ports (and one of the SGPIO connectors I'm assuming since I don't have physical access yet) seemingly work fine (or will once I move the connections around so the correct status LEDs are lit up), but what I don't understand is how ledctl is supposed to be able to address the sideband portion of the SlimSAS x8 connection to control the status LEDs for those attached bays. I would have assumed a second SGPIO would have been necessary along with the SlimSAS connection, but no such cabling exists.
So my question is, should Linux/ledctl be able to somehow utilize the special signaling from the SAS connection via the AMD controller or is it not possible because the signaling may be simply disconnected between the SlimSAS connection and the AMD controller?
Hi @nfont could you look at this issue, please?
No feedback from @nfont , closing.
I am attempting to use ledctl with a 12-bay Supermicro server (3 rows with 4 bays each), but only 2 of the bays ever "locate" properly. All 12 bays work fine as they are all filled with working SATA drives and the activity LEDs for all of them also work correctly.
When I try to locate
/dev/sda
the first bay on the first row flashes.When I try to locate
/dev/sdb
the third bay of the third row flashes.When I try to locate
/dev/sdc
through/dev/sdl
, nothing flashes on any of the bays.I've also tried different indicators (e.g.
failure=/dev/...
) and while they seem to work as intended for the already working bays, they also do not work for the unresponsive bays.The other strange thing is that when I attempt to list the slots via ledctl it shows nothing for either of the (AMD) controllers.
You can ignore the output below regarding
/sys/devices/pci0000:00/0000:00:03.4/0000:01:00.0
as that is an M.2 NVMe slot and not a part of this issue.Here is the output for various commands (using ledctl v0.97):
For the locate output, I've removed all of the preceding locate_off commands that get sent for brevity but I can post the entire output for any of the devices if necessary.