Supermicro X10SAT doing 6 short beeps

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

zir_blazer

Active Member
Dec 5, 2016
355
128
43
Last week I decided to clean my computer, which involved a full dissambly and assembly. In the meantime I got an used 4U rackmount case that I got for cheap that I intended to use to test the format since I don't like standard consumer Towers, but usually have to rely on them anyways since I don't have enough disposable money for solid, robust functional containers.
As I got a CM Hyper 212 EVO, I knew that chances were that it wouldn't fit since it is a tower heatsink and a 4U case is borderline for it, but I didn't mind running without a top cover, or trying to make a taller cover that I could stack on top of the rack. Googling results about a Hyper 212 EVO in a 4U rack proved inconclusive, so I had to try myself.
After positioning the Motherboard, I figured out that it would never fit since it was using smaller case standoffs than what was required for a CM Hyper 212 EVO (It requires to install a taller backplate on the back of the Motherboard than what is usual), and if using taller standoffs then the I/O Shield and expansion slots would not fit, so I decided to go back to the older Tower. I didn't tried to screw it or anything, noticed quickly that the thing was being supported by the CM backplate so I didn't flexed it or brute force it at all.
Point it, after assembling it again in the old Tower, the computer properly POSTed once. I got into the BIOS Setup, noticing that the settings had been resetted (Except UEFI Boot entries, since those are NVRAM), then after configuring some options so that I could boot as usual, I did a save and reset, then the computer never POSTed again.


At first, whenever I turned the computer on, the Keyboard Numpad light turned on, then after a while off then on again, as if it was on an endless POST loop. As I have experienced some difficulty getting this Motherboard out of a cold boot I left it several minutes doing that cycle, expecting it to eventually boot (I recall a time where it took a whole 2 minutes before coming to life). However, the Motherboard didn't beeped at all.
First step was to try to clear BIOS settings, as I expected than that was related since that was the last thing I touched. I tried with both the Clear CMOS button on the Back Panel of the Motherboard, and also shorting the JBT1 solder pads (It doesn't have a Jumper, you're supposed to use a screwdriver of something). No changes. Left it an entire day disconnected from the wall and without the CR2025 battery. No changes.

When I began to test the Motherboard on a table, I ended up reinstalling the Processor and testing with a single DIMM. Socket pins seems all good. No changes.
I noticed that the Motherboard did made noise if I removed all RAM modules, so the Motherboard was not entirely dead. This made me to look around for other possible culprits, and that is where I found the JBR1 BIOS Recovery Jumper. If I switched it to Recover mode, the Motherboard actually beeped with a whole 6 short beeps.
Looking around for AMI Aptio 4 BIOS Beep Codes I didn't found anything for 6 short beeps, but in some older versions it seems to point out for a Gate A20 failure, which seems a bit... extreme. Somehow it points out to a major chip (Super I/O or Chipset) not working as expected.

Looking for alternatives, I noticed that there is a Jumper, the JBR1, that manages the Supermicro BIOS Recovery Mode (Dual Boot Block). I made an USB Flash Drive with a SUPER.ROM file, but since Supermicro doesn't provide an archive of BIOS versions, the SUPER.ROM was based on the latest BIOS (3.3 I think) instead of the 2.x that I previously had. I don't know if this may cause conflicts due to the typical two steps flashing procedure thanks to Intel ME.
The Motherboard does read the USB Flash Drive since it has a light and I see it on for a few seconds, then the Motherboard makes two short beeps as if entering recovery mode. However, I never get to see anything on screen, nor I see further reads to the Flash Drive, and the Motherboard doesn't reacts again (At least it isn't in the endless loop phase). I tried touching the up and down arrows and Enter in case that the BIOS VBIOS is corrupted and the Xeon IGP is not going online, but either it is not working, or I didn't got a good key combination to flash blind.


I have pretty much exhausted my options, and getting a Motherboard replacement would actually force me to either downgrade or to spend a ton of money on another X10SAT (Even used X10SAT are expensive as hell...). I'm a bit surprised that I somehow managed to kill the Motherboard. Since it is my main and only computer I have been disconnected from the online world since about a week already, and I'm not sure if I can expect a miracle revival, or repair of sorts...

This one would be a sad birthday...
 

zir_blazer

Active Member
Dec 5, 2016
355
128
43
Been there, done that. It is actually 6 short beeps, not 5 shorts and 1 long. I don't know if the 6 beep code for Flash update is failed (From the DXE Beep Codes) is the correct one since I don't have debugging info available, so I don't know if it gets to that stage at all.
At this time I wish that I had one of those serial port debuggers, I think that the X10SAT does provide a port for that...

I do suspect a corrupted BIOS since the last time it worked was when I saved settings then rebooted. The BIOS had a weird tendency of autocreating repeated UEFI Boot Entries on every damn boot for unknow reasons (I always suspected it was a bug), and it has been a long time since I wiped the unneeded entries. I recall that there were some Notebooks that would become bricked if the NVRAM got filled, so there is that, too.
Since the NVRAM is not affected by removing the battery (It should be stored in the Flash itself, not the RTC SRAM), and automatic BIOS Recovery routines seems to have failed for some reason, the only option to fix that is to find someone that wants to desolder the Flash chip (For some reason, Supermicro doesn't like to use sockets for those...) and reprogram it. If it is not that, it is physical damage.

Does someone with plenty of bad experiences can tell me how often does the BIOS can magically corrupt itself? If this is it, seems that I have hit an edge case of sorts.
 

zir_blazer

Active Member
Dec 5, 2016
355
128
43
I'm partially happy. Sent it to a local repair shop and they successfully revived it after desoldering the BIOS Flash chip, burning the Firmware ROM and soldering it back it. This means that it was a mere BIOS corruption issue and no physical damage at all. Total cost was 15 U$D. Problem: The guy that did the job didn't noticed that it was a X10SAT since the PCB says C7Z87-OCE, so I got it back with a C7Z87-OCE BIOS. Now I have overclocking options but I'm missing a lot of the feature set of the X10SAT. I will have to sent it back again...

I think that I will try to undervolt my Xeon E3, having this Motherboard with some tuning options was my wet dream when I purchased it 5 years ago. Giving that it POSTed fine with its sister Motherboard BIOS, should it be possible to create a sort of hybrid so that I can have both overclocking and undervolting options with the X10SAT Server feature set? I think I will ask in WinRAID about that. Makes me think that I will have to ask if a custom mod to get a ROM socket in there as well is possible, as I expect it to get bricked quite often...


BTW, I think that I discovered what I touched that bricked the Motherboard. The X10SAT has a SMBIOS log that logs in things like power on and offs, and some Hardware failures like PCI errors. These are found in the NVRAM. I touched at most four options in the entire BIOS before rebooting and discovering that it transformed into a brick. One of these options was to clear the SMBIOS log at the next boot, since it had a ton of stuff due to months of operation. I think that maybe the procedure took longer hat I expected (Or there was an obscure bug involved that crashed it midway, since it left it for like 2 minutes or so before noticing that it wasn't rebooting and powering it off) and may have bricked the BIOS when I shut down it before NVRAM flashing was complete and left corrupted, half-empty or whatever other undefined data state.
 

zir_blazer

Active Member
Dec 5, 2016
355
128
43
Mediocre news. I finally have the damn thing working after around two weeks with no computer. However, the latest version of the X10SAT BIOS, 3.2, seems to have somehow removed the Graphics Configuration submenu in the Chipset Configuration menu.
If I boot with my Radeon 5770 plugged in, it gets always used as primary video output while the Intel IGP gets completely disabled, as Linux can't see it with lspci at all. Booting with no discrete Video Card results in the X10SAT picking the IGP as main output, but the Graphics Configuration submenu is still gone. I need to access it to configure the IGP to be always enabled and picked first cause I use the Video Card for VGA Passthrough to a QEMU VM, which is easier if the BIOS and Linux uses the IGP, and doing it with only the Radeon would require some reconfiguration and automatic scripts to do it headless. As such, I'm currently using my QEMU VM, but with an emulated Video Card.

Moreover, the MAC Address of the Chipset NIC (Not the i210, the i217-LM I think) seems to have been resetted to 88:88:88:88:87:88. I suppose that any data specific to my particular Motherboard model (Custom SMBIOS info) should have been blown away, too.

What other steps should I take for a full recover after getting the Motherboard Flash ROM zapped by an external reprogrammer? I suppose that I should mail Supermicro.

I'm still interesed in modding this Motherboard with a ROM socket for the BIOS and purchasing a reprogrammer. This can't happen me again, a stupid BIOS corruption completely ruined my last two weeks.
 

RyC

Active Member
Oct 17, 2013
359
88
28
Mediocre news. I finally have the damn thing working after around two weeks with no computer. However, the latest version of the X10SAT BIOS, 3.2, seems to have somehow removed the Graphics Configuration submenu in the Chipset Configuration menu.
If I boot with my Radeon 5770 plugged in, it gets always used as primary video output while the Intel IGP gets completely disabled, as Linux can't see it with lspci at all. Booting with no discrete Video Card results in the X10SAT picking the IGP as main output, but the Graphics Configuration submenu is still gone. I need to access it to configure the IGP to be always enabled and picked first cause I use the Video Card for VGA Passthrough to a QEMU VM, which is easier if the BIOS and Linux uses the IGP, and doing it with only the Radeon would require some reconfiguration and automatic scripts to do it headless. As such, I'm currently using my QEMU VM, but with an emulated Video Card.

Moreover, the MAC Address of the Chipset NIC (Not the i210, the i217-LM I think) seems to have been resetted to 88:88:88:88:87:88. I suppose that any data specific to my particular Motherboard model (Custom SMBIOS info) should have been blown away, too.

What other steps should I take for a full recover after getting the Motherboard Flash ROM zapped by an external reprogrammer? I suppose that I should mail Supermicro.

I'm still interesed in modding this Motherboard with a ROM socket for the BIOS and purchasing a reprogrammer. This can't happen me again, a stupid BIOS corruption completely ruined my last two weeks.
If you submit a support ticket, Supermicro should give you a BIOS file that restores the Graphics Configuration submenu for the X10SAT. I did this in 2016, I don't know why Supermicro doesn't just push out the fixed BIOS. If they won't give it to you, I think I still have it saved somewhere.
 

zir_blazer

Active Member
Dec 5, 2016
355
128
43
I exchanged some mails today with Supermicro support about end user BIOS restoration and they responded very quickly. Besides the classic warnings about not touching the BIOS and RMAing the Motherboard, they mentioned something which I find rather... interesing:

Unfortunately, we can’t provide any utility to the customer that can re-program the mac address on the LAN port due to the regulation from the government. Custom BIOS can’t fix this issue. Thank you for your understanding!



If you submit a support ticket, Supermicro should give you a BIOS file that restores the Graphics Configuration submenu for the X10SAT. I did this in 2016, I don't know why Supermicro doesn't just push out the fixed BIOS. If they won't give it to you, I think I still have it saved somewhere.
That is interesing to hear. At first I thought that it could be a problem with the blank static SMBIOS info, but I recall seeing the submenu when the technician gave me the X10SAT with the C7Z87-OCE BIOS installed, so I doubted about that.
Do you remember in which specific BIOS version you didn't had it? I had one of the 2.x (Maybe 2.0? Need to check BIOS filename) installed and it worked properly, now I'm using the latest 3.2 and it is missing. Based on what you said, looks like Supermicro may have hided it for some reason. There are some BIOS modding tools that I may be able to use to unlock that (And other options that seems to be missing. I only remember X2APIC Opt Out Bit, but maybe there are more). What I'm scared about is bricking the Motherboard...

I have a Thread going on on WinRAID forums, which caters for the BIOS modding audience. Seems like unlocking options is rather simple. The problem will be attemping to figure out what the missing data is, since I have things like Serial Number and MAC Address, but I'm not sure how much more could be missing.


BTW, are you still a X10SAT owner? If you don't mind that I get your Motherboard Serial Number and MAC Address, or whatever else there could be in the NVRAM (SMBIOS Event Log, UEFI Boot Entries, maybe BIOS password), it may be possible that I request a full dump of your BIOS so that I (Or some of the BIOS gurus there) can do a comparison between a BIOS devoid of unique factory programmed data and whatever factory programmed static SMBIOS data yours had, so at least I known what it is missing and what I need to replace with whatever data in stickers I can find.
 

RyC

Active Member
Oct 17, 2013
359
88
28
I exchanged some mails today with Supermicro support about end user BIOS restoration and they responded very quickly. Besides the classic warnings about not touching the BIOS and RMAing the Motherboard, they mentioned something which I find rather... interesing:







That is interesing to hear. At first I thought that it could be a problem with the blank static SMBIOS info, but I recall seeing the submenu when the technician gave me the X10SAT with the C7Z87-OCE BIOS installed, so I doubted about that.
Do you remember in which specific BIOS version you didn't had it? I had one of the 2.x (Maybe 2.0? Need to check BIOS filename) installed and it worked properly, now I'm using the latest 3.2 and it is missing. Based on what you said, looks like Supermicro may have hided it for some reason. There are some BIOS modding tools that I may be able to use to unlock that (And other options that seems to be missing. I only remember X2APIC Opt Out Bit, but maybe there are more). What I'm scared about is bricking the Motherboard...

I have a Thread going on on WinRAID forums, which caters for the BIOS modding audience. Seems like unlocking options is rather simple. The problem will be attemping to figure out what the missing data is, since I have things like Serial Number and MAC Address, but I'm not sure how much more could be missing.


BTW, are you still a X10SAT owner? If you don't mind that I get your Motherboard Serial Number and MAC Address, or whatever else there could be in the NVRAM (SMBIOS Event Log, UEFI Boot Entries, maybe BIOS password), it may be possible that I request a full dump of your BIOS so that I (Or some of the BIOS gurus there) can do a comparison between a BIOS devoid of unique factory programmed data and whatever factory programmed static SMBIOS data yours had, so at least I known what it is missing and what I need to replace with whatever data in stickers I can find.
IIRC, it was a 3.0 BIOS which lost the the Graphics submenu. The BIOS Supermicro gave me in 2016 reports version "3.0.T201511121937" in ESXi. I do still run the X10SAT, but it's at work in a production environment now and so it may be a while before I can take it down to get BIOS dumps.
 

zir_blazer

Active Member
Dec 5, 2016
355
128
43
IIRC, it was a 3.0 BIOS which lost the the Graphics submenu. The BIOS Supermicro gave me in 2016 reports version "3.0.T201511121937" in ESXi. I do still run the X10SAT, but it's at work in a production environment now and so it may be a while before I can take it down to get BIOS dumps.
Long history made short: I contacted Supermicro, they acknowledge the issue 45 minutes after my mail, then send a 3.2 based BIOS with the submenu working as intended right after the weekend. You can get version T201903151640 here. I have it working since a month ago or so.

I still want to have an open door should I want to restore it to factory defaults which would involve some BIOS modding to add back whatever missing info there is. I have taken photos of the stickers in the Motherboard which have Serial Number, apparently UUID, MAC Address, and what seems to be a 1394 Firewire serial. You may read the latest exchange with the modding guru here, there are some dumping instructions but there is still missing data about the position of some Jumpers that may not allow to dump the ME region.
The thing is, on ideal scenarios, we should be using both the same BIOS version so that there are less differences between your dump and my dump. Do you still have at hand the beta that Supermicro sent you? I could try to downgrade if it makes it easier for you.

I suppose that this would be a time sink for when you have nothing to do, since I have no reason to hurry thus I can try to coordinate to whenever you're going to take offline the X10SAT and give some mainteinance.