Supermicro system initializing B7 B9 BA BF | Really weird problem with X9DRH-7F
ฝัง
- เผยแพร่เมื่อ 21 ก.ค. 2022
- I wanted to share with you guys a really bizarre problem I encountered while testing out some Supermicro X9DRH-7F motherboards. The system is stuck during POST with "system initializing... B7". Usually, when you get codes like this, in particular, the B7, B9, BA, and BF code pertain to memory initialization problems. However, this motherboard was POSTing just fine with all DIMM slots occupied until I added a Nvidia Quadro P400 in one of the PCIe slots. In fact, I later found that the same problem would manifest with the P400 in ANY of the PCIe slots, even PCIe slots not related to the B7 code, which is the first CPU socket.
Furthermore, I found a really bizarre way of reproducing this problem, which I will demonstrate in this video. The P400 card used in this demonstration tests just fine without problems in another fully working X9DRH-7F, but on this particular X9DRH-7F, I seem to have problems.
Anyway, just wanted to share some troubleshooting tips here if or when you get "System initializing B7 B9 BA BF" codes. But also wanted to share with you guys this bizarre issue you will see in the video.
For a full in-depth review of the Supermicro X9DRH-7F, see this: • Supermicro X9DRH-7F in...
If you want to flash the LSI controller on the X9DRH-7F to IT mode, watch this: • How to turn a LSI SAS2...
== Links to products and my eBay store front ==
You can buy this X9DRH-7F already modified with LSI IT mode firmware here: ebay.to/3aZ9XMO
If you'd like to support this channel, please consider shopping at my eBay store: ebay.to/2ZKBFDM
eBay Partner Affiliate disclosure:
The eBay links in this video description are eBay partner affiliate links. By using these links to shop on eBay, you support my channel, at no additional cost to you. Even if you do not buy from the ART OF SERVER eBay store, any purchases you make on eBay via these links, will help support my channel. Please consider using them for your eBay shopping. Thank you for all your support! :-) - วิทยาศาสตร์และเทคโนโลยี
I had ran into exactly the same situation with my X9DRD board and you saved me a day, thanks.
Glad I could help
I love your videos man, been watching for years. Thanks for everything!
I appreciate that! Thanks for watching! :-)
This is amazing. I have a Supermicro server I've just received that had exactly this error. When it arrived the heat sink on the processor in question was installed at 90 degrees so wasn't doing much so I figured the problem had something to do with that, and as long as the ram was installed into any slot other than one particular blue one the problem went away, so, not ideal, but it worked.
While I was away doing another data center install I saw your video. Exactly what this server was doing. I've just removed the processor, nothing obviously wrong, nothing I could see on the pins, but, I've just fired it up with the ram back in the correct slots and boom, off we go.
Thank you!! Would never even have tried without this video.😁
Glad this was helpful? Be sure to try reseating the CPU and dimms first.
@@ArtofServer Bloody thing just locked up and is back to B9. Can only get it to boot with one stick of ram on each side of the CPU's (one of the blue ram sockets). Running a NAS with 8 GB isn't much fun.
Those PCIe slots are also wired directly to the CPU. So you may have some pins in the CPU socket not making proper contact which are for PCIe which is why when the PCIe slots are all empty it passes the power on self test (POST). But when a slot has a device it doesn't. Similar to the RAM slot issue.
This is why on newer Intel Scalable XEON's and AMD SP3 EPYC sockets they recommend the use of a specific pound-force inch limited screwdriver. In the case of the SP3 socket (Threadripper and EPYC) AMD recommends 14 pound-force inches. You can buy screwdrivers pre-set to this or buy one that you can adjust, when going over that amount of force it will disengage mechanically and not transfer any more torque into the screw guaranteeing all four screws are at the perfect level.
yes, the PCIe slots are connected to the CPUs. the strange thing is at the time of this video, the P400 is in a slot connected to socket #2, yet the B7 code is for socket #1. I definitely think something to do with the PCIe bus (and why I suspect SMBus perhaps, which often can manifest as memory fault), but it was a really strange one...
Amazing! it worked! thanks a lot!!
awesome! glad this helped!
After spending countless hours working with my super micro x9 d r i over the past 3 years it's been an absolute nightmare. Any upgrades done or just such a time-consuming troubleshooting mess. I really wish I would have videotaped myself all the time struggling with motherboards because they are very very temperamental. Best of luck.
Sorry that was your experience. I've had rock solid reliability with my Supermicro servers and I run 4 of them, not including the lab machines.
Thank you so much dear bro
You're welcome. Hope this helped you out! :-)
I bought one of these about a month ago, one of the slots doesn't clip closed fully(expect usually the clips close fully when pushing down on the ram sticks) when populated; the card was seated enough that I'd expect it would still be fine but had this sorta issue(the slot doesn't look like it has anything particularly wrong with it; and it's been running stable till now after the fact)....cause I have a mix of 3-4 sets/ 2 brands(hynix, hynix, micron) sets of ram in it, I swapped around the cards and it ended up working, definitely a funky board series but not a bad one; if I had to guess maybe they skimped on quality sockets/slots for these boards
I've built several systems with this board and have talked about it in other videos. It is generally a very good board, stable and reliable. There are some samples of this board on the used market that are not in the best condition, like this particular one in this video.
I find DDR3 ECC U/R/LR-DIMMs are generally a bit tricky to get seated correctly across the board, not just with a particular brand.
Nice Job 💪💪💪💪❤❤❤❤❤
Thanks ✌
Crazy thought tried putting in 2 gpus? 1 for each cpu assigned by user manual
My old s5000xlar had weird pcie issues but putting in 2 k600i booted fine. Taking one out or both resulted in the same issues you have had.
Good luck .
Even stranger!
For me is a bad case of grounding. When fully populating all the ram, I will have B7, B9 or BA as what you have told. Find everywhere for help and in Russian website saying it’s a bad case of ground. Put a static bag behind the motherboard and all boot up fine and working fine
Thanks for sharing 🙏. That's a very interesting remedy.
Yes, I experienced something similar. I kept receiving these errors until I removed the motherboard and cleaned out the dust and debris from the chassis. Something was causing a ground to short issue and all of my troubleshooting of seating with CPUs and DIMMs was for not!
hello, is there any chance that a dh61ww (LGA 1155) can run a h200 / h310/ 9211 8i/ 9217-414e (in JBOD/IT).
If all of them are compatible which one is recommended?
I don't know that board specifically. for best compatibility, choose one of the genuine LSI cards. if you must use a Dell or other OE branded card, there's a chance of SMBus conflict. If so, you'll have to apply the workaround. Search my channel for "smbus" to find the video showing how to do that.
@@ArtofServer Great, I did miss the SMBbus, shall surely check that out. Thank you for finding time to resolve my query.
You saved my day. Thanks why this board is design like this?
Glad I could help! I don't think the board is designed to behave with this problem.
how do you restart the IPMI when it some how mine I can't access it from web gui. or any other way its like it has crashed. I could just shut down the entire server pull power and maybe go that route. bur really dont want to do that.
You can use IPMIview to connect to the IPMI of the server and from there you can run a BMC cold reset. It will reset the IPMI and the GUI will be accessible again.
www.supermicro.com/manuals/other/IPMIView20.pdf
Or if you have ssh access to the server, then install openipmitool and run a reset command to reset the ipmi
eg: "ipmitool mc reset cold"
As Teffin has mentioned, there are other tools to reset the BMC, but it requires that your BMC is accessible in some way or another. if it is not, then a power reset might be the only way. From within the OS, you can usually communicate via local IPMI interfaces. Supermicro has a tool called "ipmicfg" that you can use. If you can still access it via the network IPMI port, then you can use Teffin's suggestion or any other IPMI over LAN client.
@@ArtofServer thanks I found out why I could not access it. I have nord VPN and for some reason I could not access that IP Address, I have it on a different Network, and Nord VPN was not allowing me at the time connect to that network. So I had to do some Configuring and then was able to to see it again. The IP Address for the different switches and servers are all on a Different IP Address Example My normal IP Address is 192.168.1.1 Were I set all my other devices I don't want Access from the out side world Like My security cameras, switches ect servers are all on a different address Example 192.168.20.1 So once I got Nord reconfigured I was able to. Some day I will stop being lazy and get pFsence set up and get nord on that and protect all my network the way I really want it protect.
It could be warping the board when the mounts are fully tightened down. Just a thought but Steve on GamersNexus mentioning about installing a spacer to even out pressure on a CPU quite recently.
this is the style of socket that is suppose to have less of any issue or not have the issue, 2 levers LGA2011, but it could still be the same issue.....
@@plasmar1 Yes but its not the clipping down of the cpu that was highlighted, it was the retainers for the cpu cooler causing uneven pressure.
yes, I think there's something to that. also, I believe some of the DIMM slot traces run under the location of the heatsink screws. other working samples of this particular motherboard are not so sensitive to how tight the heatsinks are screwed down, so it's not a design flaw in my opinion. I just think there's some weird defect with this board.
the strange thing is, without the P400 card, all DIMM slots work just fine and I can run memtest86+ without issue with all DIMM slots occupied. the introduction of the P400 (in any PCIe slot) triggers the problem, which is then alleviated by loosening the socket #1 heatsink screws!
@@ArtofServer Strange indeed it might be a board layer fault. Just to add a google for the board brings up a similar model (Supermicro X9DRH-7TF) on server at home but the board diagram shows all but one PCI-E port coming off CPU 2.
Anyhow best replaced if you can as it may surface again once out of warrenty.
Mine is the X9SRH-7F the same but Single CPU.
I was having the same problem, and I tried everything including what you do here.
in the end it only worked with a single memory module.
and, OH SURPRISE!, when I disassembled it, I noticed that where the screws that hold the heatsink go, there are tracks underneath, and the screws reached them. breaking several.
oh yeah, I've heard of that problem before... someone using aftermarket CPU cooler with screws that are too long and end up damaging traces under the screw hole. You can remove the CPU bracket stuff and expose the damage to repair it if you are handy.
@@ArtofServer 😖Yeah, I think I can fix it, but it's going to hard work. It was the 4 screws. 6 tracks on each side.
I get an AB error and a blue screen when trying to boot into the bios. Cleared CMOS and tried to change the date in the EFI shell but it won't save the date so still won't boot to bios. X9scm-f Have you ever seen that happen?
I'm not familiar with that problem. However, I used to have an X9SCM-F too, and I had so many problems with it, I got rid of it. Supermicro tech support even admitted they knew of some of the problems, but since the product was EOL, they have no plans to fix them.
@@ArtofServer Its a bios issue. Something to do with the clock if you update it which I had. It has to be rolled back in UEFI before Dec 31, 2020 or in an OS like UnRaid. Then you can boot into the bios again. Its a good starter board which I just made my backup server. thanks. As always love the videos.
Im Stack in 91 ? What I have to do
code 91 usually means there's a faulty PCIe device causing a problem. Try removing all PCIe cards and see if it resolves. If so, add 1 card back at a time until the problem is reproduced. Then you know which card is causing the problem. If you have no PCIe devices, it might mean something on the motherboard is faulty and needs to replace the motherboard.
I have the exact same. issue with an X79 EVGA board. Posted fine and then got stuck.
try re-seating the CPUs in addition to DIMM modules.
new subscriber here i have a supermicro x9dri-ln4f+ thats so picky with ram its unreal so far have only been able to get to ram sticks to work anything else and it gets stuck at either A9 or B7 lol
have you tried my first suggestion from this video? I find that if the CPUs are not seated perfectly, the pins to the DIMM slots might have poor contact and that can manifest as RAM error. it's also possible you have a bad motherboard where the traces to the DIMM slots from CPU might be damaged somewhere along the line. Especially under the screw holes of the CPU mounting brackets. If someone used a screw that is too long, it could have contacted the traces under there and damaged them.
@@ArtofServer my uncle whos a pc tech spent weeks messing with it he told me its picky with memory the only thing we havent tried is swapping the cpus around from one socket to the other and testing out the memory to see if the issue related to the dimms for socket 2 can be replicated for socket 1 thanks for the reply never seen a board so picky
update swapped cpus no dice same issue
You mentioned that the issues happens with that Quadro P400 in any slot but you didn't mention if this issue happened with any other PCIe card or not. Also maybe a card x8 x4 x2 will help narrow down the problem.
I test the other PCIe slots with some Supermicro branded HBA cards and that did not trigger the problem. The P400 did, but I did not try anything else.
That's very weird.
How are the screw mounts on these?
Can we sneak one into an ATX case to build an interesting home work station? You have me a little intrigued with these boards :D
it doesn't have through holes instead it just has a plate with threaded holes(bolts going to far would go into the board that is behind it), having said that I mounted 2 Cooler master 212's by expanding the heatsinks x brackets holes a bit and using my own bolts with nuts to adjust spring tension and bolt height; there doesn't seem to be a wide selection of official supported heatinks:P
as for atx case these are EATX form factor and so you need a super empty case (12in~ by 12in~ for the board & whatever else) or something that is big
@@plasmar1 Oh yeah I can see it, compared to some of the enthusiast boards I have operated it appears a little bigger like the EVGA X58 Classified boards that were quite big for their day.
@@CMDRSweeper if you're gonna try it out, Phanteks Enthoo Pro:) if you get the version with the lsi controller be warned that it's a tight fit for the sff-8087's but doable.... X9DRH-7TF has dual 10gbe(intel x540-t2)
@plasmar1 thanks for answering!
I have the BA issue. Now I will trie this. sounds strange. My IPMI is also not starting every time I plug in the Server into Power
Sounds like there are some problems with that board...
But with this tip it worked. Strange what on this board is that often wrong
the supermicter gave Me B0 when motherboard was placed in the table, all three times, and when i was installed in the case of motherboard b0
That's a totally different code. I believe B0 means you are using wrong type of ram modules.