SuperMicro BMC unresponsive after shutdown?

[i386]

I just booted to the Proxmox (Debian) installer no problem.

Can you reach ipmi via browser?

Yeah it's an onboard M.2 SSD.

I think this could be problematic as the x10sdv boards seem to have problems with many "untested" m.2 ssds. My x10sdv-2c for example doesn't recognize the 990 pro like every two months...
But I doubt that it has something to do with the bmc/ipmi being unreachable ._.

 

[altano]

you have set NVMe FW source to "AMI native" ?

I can't tell because I can't get into the BIOS anymore. If I try by pressing "DEL" during bootup I get stuck at a black screen that has a cursor and the text "A9" and nothing else. This happens every time.

I can install an OS from USB and boot into it. So the two things that are not working right now are IPMI and BIOS editing.

 

[RolloZ170]

clear CMOS

To solve the problem, perform the following steps:

1. Disconnect all power cables so that no power supply is connected to power..
2. Remove the BIOS battery from the motherboard.
3. Carry out a reset by short-circuiting the two metal connections on the battery socket with a metal tool for ~5 seconds.
4. Replace the BIOS battery and restore power to the system.

 

[altano]

Can you reach ipmi via browser?

No it is unreachable from IPMIView and web browser.

clear CMOS ...

These steps aren't letting me into the BIOS, same problem. But after these steps IPMI was responsive again, because I power cycled. But it's unresponsive again after I rebooted.

 

[altano]

I don't think I successfully cleared the CMOS because the "Recovery from AC Power Loss" is still set to "Power-On" (instead of the default of "Last State"). That should be reset back to default, right?

I'm definitely unplugging both PSUs in the system, pulling the CMOS battery, and shorting the JBT1 jumper on the battery socket. Not sure what I'm missing.

 

[i386]

I'm definitely unplugging both PSUs in the system, pulling the CMOS battery, and shorting the JBT1 jumper on the battery socket. Not sure what I'm missing.

Manuals usually say something like "remove power for 5 minutes".
Take a coffee break and wait

 

[RolloZ170]

I'm definitely unplugging both PSUs in the system, pulling the CMOS battery, and shorting the JBT1 jumper on the battery socket. Not sure what I'm missing.

short the battery socket longer with a coin or something similar.

 

[altano]

I left the CMOS battery out of the motherboard overnight and now it won't power up at all. I can't power it up from the chassis or from IPMI (which is responsive atm). I tried changing the CMOS battery to a brand new CR2032 and it still won't post. =[

 

[RolloZ170]

I tried changing the CMOS battery to a brand new CR2032 and it still won't post. =[

wrong or bad BIOS ? flash chip corrupted. this can happen if you install not matching BIOS<>BMC FW.

 

[altano]

So I've been experimenting with my second X10SDV board (the one that I haven't bricked) and it looks like it is experiencing all the same symptoms:

• I often cannot get into BIOS setup (black screen w/ A9)
• BMC often becomes totally unresponsive
• Startup sometimes hangs

These symptoms all seem to go away when I remove the P4801X Optane SSD I was using as my boot drive. That makes sense because it was the only common factor between these two boards, both of which worked fine with another drive.

@RolloZ170 I've looked all through the BIOS and I don't see any option to change the NVMe FW source. Do you know where it might be?

 

[oneplane]

What power supplies are you using? What firmware versions are you running? What other internal devices do you have connected? Is the ID button cramped?

 

[ericloewe]

I've looked all through the BIOS and I don't see any option to change the NVMe FW source

Does the P4801X even have an extension ROM?

 

[RolloZ170]

've looked all through the BIOS and I don't see any option to change the NVMe FW source. Do you know where it might be?

this setting if present is only for onboard M.2 SDD without firmware. PCIe SSD has OpROM

 

[RolloZ170]

Does the P4801X even have an extension ROM?

if you BOOT in EFI mode the OpROM should be set to EFI too.

 

[RolloZ170]

Does the P4801X even have an extension ROM?

wait: is this SSD M.2 format ?
edit:
P4801X in onboard M.2
Change M.2 OPROM to EFI

P4801X in AddOonCard you will be required to change slot# OPROM to "EFI" and M.2 OPROM to EFI also.

 

[itronin]

@altano

EDIT - DO NOT DO THIS see post by @ericloewe

Do you have the molex power connection hooked up on the motherboard? My reading of the SM manual says for power to "onboard drives" which I have always read to mean the DOM's and the nvme slot. I had wonky/bizarre behaving from my X10SDV-4C's with dom and m2. nvme until I added power to the molex. Note that without doms and using an m.2 to u.2 adapter card I've not had to hook up the molex since the u.2 is drawing power at the drive.


As stated previously the oprom setting is key if using EFI too!

 

[ericloewe]

@altano

Do you have the molex power connection hooked up on the motherboard?

I added power to the molex

Don't do that. It's a power output for when the +12V power connector is used, so that a disk or two can be connected without needing a full ATX PSU. Yeah, the pins are, rather bizarrely, reversed relative to what everyone agrees on (female for power out, no touchy live pins).

 

[itronin]

Don't do that. It's a power output for when the +12V power connector is used, so that a disk or two can be connected without needing a full ATX PSU. Yeah, the pins are, rather bizarrely, reversed relative the what everyone agrees on (female for power out, no touchy live pins).

well eff me gently with a chainsaw. its a wonder no magic smoke escaped from my boards. yikes. need to go back and check mine that are sitting on the shelf to make sure I didn't leave 'em cabled up.

 

[altano]

What power supplies are you using? What firmware versions are you running? What other internal devices do you have connected? Is the ID button cramped?

• PSUs: PWS-920P-1R redundant PSUs
• BIOS: v2.3 (latest)
• Other hardware:
  • 128GB Memory (4 x 32GB DDR4-2400MHz HMA84GR7MFR4N-UH Hynix RAM)
  • Optane 900p PCIe add-in card
  • Lots of 3.5" SAS and SATA drives connected to onboard SAS controller

if you BOOT in EFI mode the OpROM should be set to EFI too.

Yep it's an M.2 drive, and I had the oprom set to EFI. Was still experiencing issues (e.g. couldn't enter bios setup).

 

[oneplane]

I think this means what we have to specifically look for bus related issues or flash related issues.
The BMC is connected to the same bus as any PCH and CPU on the board would be, so if something is causing issues, neither would be very happy.

If the BMC starts up but then crashes, that's probably bad flash, or a bad call from the UEFI DXE for BMC communication, since once the 'entering' screen for the firmware configuration utility loads it needs to find some data, including (but not always): MAC address and IP configuration for the BMC, BMC status, BMC version etc. and while some of that can be located in flash, some of it is requested over I2C.

I would perform firmware recovery for both devices (UEFI flash, BMC flash) before trying much more. If a naked boot (all devices removed except CPU, memory) has the same problems as with everything plugged in, you'll know for sure.