I contacted ASRock Rack’s support about the missing Network Bond Configuration menu entry on one of my units yesterday including a reference to the “known” solution in the ServeTheHome thread.
I haven’t heard back from them - maybe I rubbed someone the wrong way - but I’ll post a copy of the instructions here for increasing the knowledge value of this forum hopefully beating out any other threads on the interwebs 
I’m not the guy from the other forum, I just happened to know a solution to a problem some of you guys were having, so I registered an account here and pointed at it :-).
I’m not sure it’s mine to share, that’s all. I can check tomorrow if there’s anything related to that mentioned in their documentation.
The tool is called bmcmac btw.
I think it took them a couple of days to get back to me.
I’m working with AsRock Rack on this right now, and I am being told:
NC-SI bonding selection option may have been eliminated in the production BMC firmware. However, I have an ipmitool command that you can try to disable it manually. Please see the command and syntax below and let me know if it works for you. Thank you!
Bond disable
ipmitool -H <ip> -U admin -P admin raw 0x32 0x71 1 0 0 1 0x64 0 3 0
This was in regards to me asking about my X470D4U2-2T. I then also asked if the behavior would be the same on my X470D4U as well as my X399D8A-2T. William was nice enough to respond with:
If the previous command worked properly on your X470D4U, it should also work in other unit that is also using ASpeed AST2500. But check for the NC-SI disable availability before you do so because every BMC for each board might be different.
How are you trying to contact them? I did a test on their support. When sending them an email to their netherlands support address they’re not responding. When using the support form of their website I got an answer the next day.
I also use the proper support form (Europe/English), only then responding to the emails they’ll send me with my support ticket number.
If the support’s initial response isn’t that “on-point” with its first try (“Try another PSU” for a corrupted BMC image) and I clarify things with feedback I basically never get further responses (has become a pattern since July).
Could you please describe what’s the underlying issue in a normie-proof way and what these commands are doing?
Sure, in your previous posts you shared a screenshot where you have the option in network settings of your BMC called “Network Bond Configuration”. In there, you can remove the Intel nic from your bond, and isolate your BMC nic adapter. I don’t have that option in my BMC, so I contacted AsRock to ask what gives as I also want to Isolate my BMC nic from the Bond, breaking the bond.
This was the answer I got. That perhaps they are removing that option from the BMC, however using ipmitool (which I am previously familiar with for fan control and remote turn on / shutdown), we seem to have the command that can remove the nic from the bond as a stand alone on AST2500 boards.
I have not tested yet, but that’s the short of where it’s going.
So ASRock Rack’s current fix is just a workaround using the command line tool what should be possible via GUI?
At first I thought that after doing something with command line tool xyz you’ll enable the GUI option as it should be.
Any ideas how such a bug can survive a BMC restore flash where the exact same file is used for two motherboards - and one shows the Network Bond Configuration option and one doesn’t?
Not sure… Like you said, it feels like a workaround if the info I am getting is the “Official” response, and not just supports workaround to help me along.
I get the impression from the wording that there exists the possibility to just not offer the user the ability to disable bonding altogether? Seems like that could be the path forward… Not trying to chicken little, just passing verbatim what I am getting.
careful with this… it did not work for me but it did make the http interface unavailable. I had to reboot the machine to get it back.
Thanks for the heads up, I will let AsRock know in the ticket. By chance are you on the U or U2-2T board?
Got the response detailing the procedure today (for the X470D4U):
This is the procedure we are using …
Latest firmware
BMC image:
https://www.dropbox.com/s/3mlhkncdpkrq503/X470D4U_P1.90.00.zip?dl=0
BIOS image:
https://www.dropbox.com/s/4q37sdstpubncxp/X47D4U325.ZIP?dl=0
about the bonding :
It is indeed possible to disable the bonding but the problem is that if you do so you will get nic ports with 2 same mac addresses
SOP for attachment tool:
Prepare a bootable usbDecompress the file into usbExecute S_BMCMAC.exe 0= {customer’s MAC address} in this case you can find the bmc mac address in the bios ( server management )MAC1 and MAC2 should be set automaticallyCheck MAC in BIOS after update successhttps://www.dropbox.com/s/8376pxeylsu37b8/2nd%20Mac%20with%201st%20Mac%20%2B%20offset.zip?dl=0
Regards,
ASRockRack TSD
What’s not clear to me: Do I get multiple NICs with the same MAC address or can I fill in the Realtek NIC’s MAC address where it says “{customer’s MAC address}” and it’ll be OK?
im on the X470D4U
Looks like they say if you fill in the MAC for the address that it is now, the utility will set the new one automatically.
Bananas, why are we getting different solutions?
Maybe ASRock Rack’s marketing team in association with Intel has launched a large-scale study and all customers of non-Epyc AMD “Workstation”/“Server” motherboards automatically participate…?
Group A: Buggy firmware, solution A is being tested on them
Group B: Buggy firmware, solution B is being tested on them
Group C: Annoying, nitpicky Germans where 9 of 10 messages get automatically redirected to the spam folder
Und so ist das!
Has anyone tried to run a Ryzen 9 with one of these yet? I know the 3000 series had some teething issues but with the new 3950X crushing the benchmarks I’m wondering if my 1700X is really starting to show it’s age here.
The 3900X is on the official CPU compatibilty list and I can’t think of anything that would speak against it.
Also, by logical extrapolation, the 3950X won’t be a problem since its power consumption is about the same as a 3900X’s since its two 8-core chiplets are of a higher silicon quality meaning they need less voltage for a certain frequency target.
With the propagated “ECO-Mode” (new feature in AGESA 1004 B), reducing its TDP to 65 W it would even be less of an issue.
The only issue regarding the 3950X is: When are we going to have a BIOS with AGESA 1004 B? At the launch of Ryzen 4000?
That’s what still keeping me on the fence about this board. The support seems to be way behind (I know it’s a server board, but still) a lot of the regular consumer boards and it just having long standing bugs that haven’t been fixed. I don’t really “need” IPMI since the server is in my basement, and I’m thinking I might be better served with an Asus X470-Pro or X570-Pro.
Hopefully, they make some public releases in the near future that fixes some of these issues. I don’t feel comfortable going with beta BIOS and BMC firmware releases to get a motherboard to do what I bought it for.