Supermicro chassis front panel 'Power Fail' LED red

[DearestDreamer]

Hello,

I seem to have a weird issue that doesn’t make much sense to me. My 836BA-R920b chassis’ front panel has a LED that the manual labels as “Power Fail: Indicates a power supply has failed”. My server is connected to a 1000 watt UPS. I have the X10SRL-F motherboard and yes the chassis revision is ‘M’.

The LED is red – not blinking, but constantly illuminated as red after startup.

I checked the following:
1. Both PSUs have a green light in the back, I assume this means they’re working
2. I unplugged the top PSU and the server started from just the bottom one
3. I unplugged the bottom PSU and the server started from just the top one
4. I plugged both PSUs directly into the wall, instead of of the UPS
5. Simulated a power failure by unplugging one power cord and I did hear the internal switch to the other power supply, so I think the redundancy works just fine

All of this appears to work, the server is powered on and there’s “no issues”, but the ‘Power Fail’ LED is red. What’s going on here? How should I troubleshoot this? Could my front panel be broken (All the other LEDs work properly)? Do I have to enable the other PSU in the BIOS?

In IPMI I don't see a red exclamation or some alert about the PSU or power fail:

I also noticed through IPMI that I only see “PS1”, is this correct?

Here's some more images that might help understand what my setup looks like currently. While it looks like I have SATA drives installed, it's wrong. I also don't have any other drives installed, so the HBAs are just connected to the backplane through SFF8087, that's it.

Here's my build thread for reference: https://forums.freenas.org/index.ph...and-x9srh-7f-o-26tb-raidz3.39556/#post-245220

Any help would really be appreciated, since I don't know if the LED should be red normally or if I might have some kind of broken front panel that I need to RMA?

 

[jgreco]

Pull the power on one PSU. You should get an alarm. Pull the PSU module. The alarm should stop. Replace the PSU module. Alarm again. Connect power. No alarm. Red light probably disappears too.

 

[DearestDreamer]

Pull the power on one PSU. You should get an alarm. Pull the PSU module. The alarm should stop. Replace the PSU module. Alarm again. Connect power. No alarm. Red light probably disappears too.

I pulled out the power cord from the top module.
I didn't get an alarm, I just heard the internal switch to the bottom module I presume.
I now pulled the top module completely off.
I got the LED next to the "Power Fail" LED to blink red, which is the Information LED.
No alarm again of any sorts, I can see IPMI has an exclamation mark on the PSU image in the Sensors tab and there are two system event logs:
"Assertion: Power Supply | Event = Power Supply Failure detected"
"De-assertion: Power Supply | Event = Power Supply Failure detected"
So I just inserted the module back into its chamber.

I repeated this with the bottom module as well and I could see that the "PSU" tab in the "Sensors" tab actually showed the switch when one module was pulled out, because the PWS Serial Number changed to the one still in the chassis. It looks to me as though the PSUs are working just fine and detected properly, no?

Red light is however still on after all of this.

Could this all be an issue with the front panel LED connection? Am I supposed to hear a beeping alarm?

Thanks for your help jgreco.

 

[jgreco]

Yes, by "alarm," I'm sorry, I meant "shrill annoying alarm" plus the other indications you observed.

We notice that periodically the warning light may turn red for no apparent cause. I actually cleared one as a result of this message thread.

If the red light's still on, I don't really know what to tell you, sorry.

 

[DearestDreamer]

I'm not sure if this might be relevant, but I saw that this cable (SMB connector) was seemingly missing its 5th wire? Is this normal? Could this be what's causing the issue after all? If I remove it and start the server without it, the POWER FAIL LED is off. Surely this can't be the fix, though, since I'd like the monitoring SMB provides.

So I don't have any buzzing alarms.. weird? I saw that the 836BA-r920b had a chassis intrusion and I did connect it to JL1 on the motherboard, but I only get the IPMI logs, no buzzing sound here either. I don't seem to have a cable for JOH1 on the motherboard which is for Overheat/Fan Fail warnings. I did however find this in the manual:

Could this mean that the Power Fail will be on if there's overheating issues? Looking at my temps above, I don't see anything out of the ordinary? I mean the PSU seems to be at ~30C/90F and running its fans at 2k RPM. Could it be that the Power Fail LED doesn't blink but just stays on red, and this indicates overheating instead of PSU failure?

I just don't understand where the issues is currently.. everything is working just fine and either PSU doing its job. There's no other warnings except for that LED. If I have to exchange a part, which one do you think would be most likely? It seems a little odd that my motherboard isn't making buzzing sounds if a module is unplugged and the chassis intrusion isn't alarming either - do I need some kind of special connection for this to work? I always thought motherboards had an onboard noise thingie.. I checked next to the front panel connection on the motherboard and there's some pins and a jumper next to it. The manual indicates the following (my jumper is on pins 6+7):

JD1 Speaker/Buzzer (Pins 1-3: Power LED, Pins 4-7: Speaker)

This seems to be correct because it says: "To use the internal onboard speaker, close pins 6 and 7 with a jumper."

 

[jgreco]

You won't get any buzzer from the chassis intrusion.

I guess it could be overheating but I *thought* that was a transient alarm, and resets when things cool down. I could easily be wrong.

As for the PMBus, I was curious about the red wire. All I have available to me right now is a 1018R-WC0R (X10SRW) so it's similar but not horribly close to your setup. Also it's running so I did the best I could:

It's up there next to the blue/black connector. Only three wires, no +/gnd, red wire in the center. Seems like that's the color they use.

So in general I think you're doing all the right-ish sorts of things to figure this out; I have no further ideas for you, sorry.

 

[DearestDreamer]

Thanks for checking a configuration on your end, jgreco. Right now I'm assuming it's overheating issues, so I'll compile an e-mail to Supermicro and hopefully they can tell me what's going on here exactly and what I could do to fix it. I mean there's a few cables in the chassis, but what kind of sensors would pick this up and claim it as overheating? And what kind of temperatures should I aim for to avoid overheating (if that's what it is)? I tried setting the fans to heavy I/O and full, but the red Power Fail LED still isn't going away.

I'll report back once I hear from Supermicro, if anyone else has any ideas I'd more than happy to try things out or give more info. Thanks again jgreco!

 

[tvsjr]

If the second PSU isn't showing up in IPMI, you most likely have a data/communications issue some place. I wouldn't expect this to be overheating.

 

[DearestDreamer]

If the second PSU isn't showing up in IPMI, you most likely have a data/communications issue some place. I wouldn't expect this to be overheating.

Are you saying I should have a "PS2" tab here?

Show

As I mentioned above, the PWS serial number changes when either of the module is unplugged, which leads me to believe that both modules are recognized. The Sensors tab as you can see also shows two modules, I suppose to imply the redundancy - and it does throw a red exclamation on that image if I unplug one of the modules.

 

[Hobbel]

took a bit longer, but I had a look at the chassis and the red light shouldn't be on. Have you checked your backplane for the FAN jumper? For me all where set to "on" (don't have the exact word atm) and I set them to "not connected" (or similar). You should know what I mean, when you have a look at the manual and the backplane jumpers.

For the PSU tab in the IPMI section I currently can't look for it...

 

[DearestDreamer]

Oh alright, I'll check the jumpers then. I actually didn't look at the backplane at all, lol, I've been so busy with the motherboard and connecting everything that I totally forgot about checking the backplane more thoroughly.

I'll look at the jumpers, good idea, thanks Hobbel.

EDIT: So I checked the fan jumpers on the backplane and they're all on Pins 2-3, whereas Pin 1 is the square one. So it seems to be just in accordance with the manual.

I'm really not sure it's the backplane, I mean if I take out the PMBus SMB connector, the LED is gone but I also don't have PSU reporting. So it must have something to do with the PSU, but what it is exactly, nobody seems to know. Still waiting on Supermicro to get back to me, hopefully telling me if I need to RMA either the PSU modules or the PSU backplane, I guess that could be dysfunctional too.. let' see.