Boot hang: IPMI system interface?

[TeslasPigeon]

I have recently put together a small NAS from the following components:

Motherboard: DFI-ITOX BE171 (AMD AMD RX-427BB SoC)
DRAM: 2x Micron 8GB PC3L-14900S 1866 MHz DDR3L SODIMM
SSD: Micron M600 MSATA 64GB
RAID controller: Startech PEXSAT34RH (Marvell 88SE9230)
HDD: 4x 3TB HGST (HUS724030ALE641)

I have installed Freenas on the SSD several times, and each time the installation completes without any errors. However, when it boots, it hangs at the following message, shortly after the LAN interfaces are initialized:


ipmi0: <IPMI System Interface> Port0xc000-0xc0ff mem
0xfe815000-0xfe815fff, 0xfe80400-0xfe807fff irq 39 at
device 0.3 on pci3
ipmi0: Using KSC interface


The system does not boot to the console setup menu, the system does not request a DHCP lease from the server, I cannot access the web GUI, and I cannot establish a serial connection through the com port. The motherboard has two LAN interfaces (Intel I210 and Realtek RTL8111EP), and the motherboard manual specifies that the RTL8111EP chipset supports Realtek DASH remote management, however the BIOS does not have an option to set a static IP for either LAN interface, only to disable DASH, which does nothing.

I am not familiar with IPMI, but my understanding is that it is a protocol used for remote management, although I am not sure how to utilize it (I assume it is related to the Realtek DASH remote management functionality?). Can anyone who is familiar with IPMI provide more insight into the situation, and a possible solution? I have exhausted my limited knowledge, and would appreciate any suggestions.

 

[jro]

I believe DASH is Realtek's implementation of IPMI. The second LAN port will be the dedicated IPMI port, but you shouldn't need something connected there for the system to boot.

I'm not sure what to make of the console message you shared... it looks more like standard reporting rather than an IPMI-related error that would halt the boot sequence. Are there any other console messages that look more error-y? Are you sure you have the boot sequence set correctly (i.e., it's set to try to boot from the SSD)?

edit: I just found your post on r/homelab... there could be an issue with your motherboard. Have you tried emailing DFI? Are you on the latest BIOS?

edit2: Another thought... have you tried plugging an ethernet cable into both the Intel LAN port and the Realtek/IPMI port? This might get the IPMI to load in a different mode that doesn't generate an error.

one more edit: Look at the manual for the motherboard, page 33: http://54.84.194.202/Upload/Product/Manuals/Mini-ITX_BE170_BE171_BE173_Manual.pdf

Here's the menu for disabling IPMI/DASH functionality (Advanced > MCTP Config). Try disabling DASH and MCTP.

 

[TeslasPigeon]

I believe DASH is Realtek's implementation of IPMI. The second LAN port will be the dedicated IPMI port, but you shouldn't need something connected there for the system to boot.

I'm not sure what to make of the console message you shared... it looks more like standard reporting rather than an IPMI-related error that would halt the boot sequence. Are there any other console messages that look more error-y? Are you sure you have the boot sequence set correctly (i.e., it's set to try to boot from the SSD)?

edit: I just found your post on r/homelab... there could be an issue with your motherboard. Have you tried emailing DFI? Are you on the latest BIOS?

Thank you for your response. I agree that it doesn't seem to be an error message, and I have tried disabling the DASH functionality in the BIOS, I have tried to boot it with and without a cable in either LAN port, and it still stops with the IPMI message. None of the other messages seem to be erroneous, and it is booting from the M600 just fine (it is detected in BIOS as well). I had to set jumpers on the board to run the slot in SATA mode, however I doubt that has anything to do with the issue. I haven't emailed DFI for the BIOS, since I bought the board off ebay and am not sure they would send me the BIOS since I didn't purchase it from them directly (since it is an industrial embedded SoC board). I think I will send them an email anyways, and see if they respond.

Edit: I have tried each port individually as well as simultaneously. I have also tried disabling DASH and MCPT, however the issue persists. Perhaps it is a bug in the BIOS?

 

[jro]

Very strange... Hopefully it's just a BIOS update to fix it and your board isn't broke. They should provide BIOS updates regardless of where/how you purchased the board. Whether they provide support is another question...

Have you tried booting any other OSs on it? Might be worth a shot to install Ubuntu or something on your SSD and see if that misbehaves as well.

 

[TeslasPigeon]

Very strange... Hopefully it's just a BIOS update to fix it and your board isn't broke. They should provide BIOS updates regardless of where/how you purchased the board. Whether they provide support is another question...

Have you tried booting any other OSs on it? Might be worth a shot to install Ubuntu or something on your SSD and see if that misbehaves as well.

Yeah, it strikes me as suspicious that they would have all the necessary manuals, specsheets, and drivers for their motherboards but require you to email them for the BIOS image.

Thanks for the suggestion, I will try installing 18.04 LTS later and see if I can get it working.

 

[droeders]

Isn't there a verbose option from the FreeNAS boot menu? If so, can you try that to see if there are any other messages of interest?

Lots of times a kernel hang is related to ACPI, but I wouldn't expect the install to succeed if that were the case.

 

[TeslasPigeon]

Isn't there a verbose option from the FreeNAS boot menu? If so, can you try that to see if there are any other messages of interest?

Lots of times a kernel hang is related to ACPI, but I wouldn't expect the install to succeed if that were the case.

I have booted it in verbose mode, and I can't identify anything else atypical.

 

[droeders]

I'm not sure how comfortable you are with CLI/FreeBSD. Can you boot with the installation media and then get to a shell?

If so, maybe try mounting /boot from the SSD and renaming/moving all the ipmi*.ko modules from /boot/kernel (I believe this is the path) and see if it still hangs.

It would be nice to know if IPMI is really the culprit.

 

[TeslasPigeon]

I'm not sure how comfortable you are with CLI/FreeBSD. Can you boot with the installation media and then get to a shell?

If so, maybe try mounting /boot from the SSD and renaming/moving all the ipmi*.ko modules from /boot/kernel (I believe this is the path) and see if it still hangs.

It would be nice to know if IPMI is really the culprit.

Thanks for the suggestion, I will be sure to try this. Will removing the IPMI kernel modules cause any issues?

 

[droeders]

Thanks for the suggestion, I will be sure to try this. Will removing the IPMI kernel modules cause any issues?

It will not harm anything to move/rename them for debugging.

That said, the IPMI module is certain useful. For example, many people (including myself) use IPMI to monitor and control system fans based on CPU and hard drive temperature. Nothing critical for operation though.

 

[TeslasPigeon]

It will not harm anything to move/rename them for debugging.

That said, the IPMI module is certain useful. For example, many people (including myself) use IPMI to monitor and control system fans based on CPU and hard drive temperature. Nothing critical for operation though.

Alright, following your advice, I booted into single-user mode and proceeded to enter the following commands


#su -
#mount -u -w /
#cd /boot/kernel
#rm ipmi.ko
#rm ipmi_linux.ko
#exit
#reboot


After a reboot, I was able to pass the point where the boot message displayed, and I am now sitting at the FreeNAS console setup menu! So it appears IPMI was in fact the issue! I'm not sure that I will be needing IPMI in my use case anyways, since the only fan that is PWM controllable is the CPU fan, which has temperature threshold settings controlled in BIOS. All that being said, I wonder if there is perhaps a conflict between the Realtek RTL8111EP chipset and the IPMI kernel modules?