Minisforum MS-01 PCIe Card and RAM Compatibility Thread

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

hsm

New Member
Jun 21, 2024
1
2
3
I just received my TL-D400S and I installed the included QXP-400eS-card in an MS-01 running proxmox. I lost my network-connectivity.
HELP
I'm using a 10G fibre-module in an SFP+ port. When I install the card, I don't have network-connectivity anymore. Removing it gets my network back. Any help ? At the moment, the DAS does not have any disks.
lspci sees the card.
The lights on my switch connected to the fiber-cable light and I have an ip-address (although with a fixed address that does not say much). Pinging does not work, not to local network, not to global network.
It's possible for devices to change names in Proxmox when new hardware is installed. Check the NIC names with `ip a` before you do anything else.
 
  • Like
Reactions: senso and Patrick

alphasite

New Member
Oct 11, 2021
17
15
3
I thought I had read that people were also experiencing the issue on their laptops, but maybe those laptops used desktop CPUs instead of mobile CPUs. If the mobile CPUs aren't affected then we're good.
 

Patrick

Administrator
Staff member
Dec 21, 2010
12,535
5,860
113
I just received my TL-D400S and I installed the included QXP-400eS-card in an MS-01 running proxmox. I lost my network-connectivity.
HELP
My best guess is that your network interfaces got renamed. At least that would be the most common cause I have run into. Check ip a versus /etc/network/interfaces and make sure the NIC you want is what is on the bridge.
 
  • Like
Reactions: Drizzik

Reduskaa

New Member
Sep 1, 2022
3
1
3
Good day! I recently purchased a Minisforum MS-01 (i5 12600H, 48GB) in hopes of replacing my old NAS. I’ve connected three 2TB M.2 drives, and an LSI 9207-8e card that worked perfectly in my old NAS is installed in the PCIe x8 slot. Proxmox is installed, and the idea was to create a VM, pass through a USB with Unraid, two NVMe drives, and the LSI card connected to a DAS with 8 HDDs. I can easily pass through the NVMe drives to the VM, but when I try to pass through the LSI card, the system reboots. I tried covering pins 5-8, but that didn’t help. Any ideas?
 

Reduskaa

New Member
Sep 1, 2022
3
1
3
you have to check iommu groups: proxmox wiki pci passtrough
If that is not correct you can passtrough disks only but you will loose smart data and not be able to spin down disks in unraid.
Thank you, I’ve been using Proxmox for a long time, and the LSI controller has a dedicated group. I need to pass through the controller itself, as passing through the disks is inefficient and dangerous
 
  • Like
Reactions: P00r

Koenraad

New Member
Aug 4, 2024
3
1
1
My best guess is that your network interfaces got renamed. At least that would be the most common cause I have run into. Check ip a versus /etc/network/interfaces and make sure the NIC you want is what is on the bridge.
My network connectivity is back.
Indeed, it was a renaming of the NIC that was the problem. I fixed /etc/network/interfaces and all is fine now.
Many thanks.
P.S. taping off the SMBUS pins didn't do anything I could see.
 
  • Like
Reactions: Patrick

Prognosis5101

New Member
Aug 31, 2024
5
3
3
I have been dealing with the lockup issue for a month or so.
13900H
48GB Crucial T48G56C46S5 x2
Microcode 0x4121
Proxmox 8.2.4


I have asked for the beta BIOS, but I have not received it yet.

My work around so far is to shutdown the server 2x a day with a cron job:
00 23 * * * /sbin/shutdown -P +5

I have the server plugged into a smart plug with energy. I set it to power off between 23:04 and 23:15 only if the power levels dip below 5 watts for 2 minutes. After 5 minutes it powers back on. (Set the BIOS to always power on on power restoral)

I have found that a simple reboot had no effect on stability. It would continue to freeze whenever. In order for the system to be relatively stable, I HAD to power it off for a bit.

I have set the power limit PL1 to 45000, and the PL2 to 50000. I am hoping that capping the power will have some positive effect.
 
Last edited:
  • Like
Reactions: minisfckr-01

Nigdzie

New Member
Aug 28, 2024
5
4
3
My network connectivity is back.
Indeed, it was a renaming of the NIC that was the problem. I fixed /etc/network/interfaces and all is fine now.
Many thanks.
P.S. taping off the SMBUS pins didn't do anything I could see.
this is know bug ;)

for me this was a pain .... after making fix /etc/udev/rules.d/70-persistent-net.rules all is fine
 

BlueChris

Active Member
Jul 18, 2021
152
53
28
53
Athens-Greece
this is know bug ;)

for me this was a pain .... after making fix /etc/udev/rules.d/70-persistent-net.rules all is fine
Sorry for asking but i dont know how to do that, i mean i see info all over the internet from many years before but can you point to something certain and concrete on this?
thx
chris
 

bachle

New Member
Sep 3, 2024
1
0
1
Hi,

I am trying to connect 1080ti to MS-01 machine directly through a PCIe riser cable. I have an external PSU that are powering the 1080ti.
However, it does not seems to work. Both windows and fedora failed to recognize the 1080ti.
I have another eGPU, and if I plug my PCIe cable into eGPU, then the 1080ti works fine.

Do you have any idea that why my MS-01 does not register device in the PCIe slot?

Thanks,
Bach
 

Nigdzie

New Member
Aug 28, 2024
5
4
3
Sorry for asking but i dont know how to do that, i mean i see info all over the internet from many years before but can you point to something certain and concrete on this?
thx
chris
example of my usage MAC is random so you have to put own MAC's base on EIF and to get mac just use 'ip a' command:
Here is the udev rules file that will assign the correct static network interface names as per your requirements:

Code:
# Assigning static network interface names
SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="AA:BB:CC:DD:EE:01", NAME="enp87s0"
SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="AA:BB:CC:DD:EE:02", NAME="enp90s0"
SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="AA:BB:CC:DD:EE:03", NAME="enp2s0f0np0"
SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="AA:BB:CC:DD:EE:04", NAME="enp2s0f1np1"
Instructions:

1. Open the file: Open the `/etc/udev/rules.d/70-persistent-net.rules` file using a text editor:

Code:
nano /etc/udev/rules.d/70-persistent-net.rules
2. Paste the rules: Paste the code above into the file.

3. Save the file: Save the changes and close the editor (in `nano`, use `CTRL+O`, then `CTRL+X`).

4. Restart udev:

Code:
systemctl restart udev
5. Restart networking (optional, if the new names are not applied automatically):

Code:
systemctl restart networking
After performing these steps, your network interfaces should receive the correct, static names as per your requirements.
 

welcometors

New Member
Jan 16, 2024
21
9
3
Hi All

I am building my own case for better cooling and I want to install an PCIe card for 2.5 inch sas/sata SSD (max 8).

Which card will be better? Will LSI 9300-8i fit?

Also I see a lot of Cheap Micron U.2 SSDs on eBay. They are even cheaper than intel SATA SSDs. How can I use them instead of SATA?
 

Fazio

Member
Dec 20, 2022
29
25
13
I have been dealing with the lockup issue for a month or so.
13900H
48GB Crucial T48G56C46S5 x2
Microcode 0x4121
Proxmox 8.2.4


I have asked for the beta BIOS, but I have not received it yet.

My work around so far is to shutdown the server 2x a day with a cron job:
00 23 * * * /sbin/shutdown -P +5

I have the server plugged into a smart plug with energy. I set it to power off between 23:04 and 23:15 only if the power levels dip below 5 watts for 2 minutes. After 5 minutes it powers back on. (Set the BIOS to always power on on power restoral)

I have found that a simple reboot had no effect on stability. It would continue to freeze whenever. In order for the system to be relatively stable, I HAD to power it off for a bit.

I have set the power limit PL1 to 45000, and the PL2 to 50000. I am hoping that capping the power will have some positive effect.
I have the same setup as yours and facing the same issues since 2 months! I had to buy a smart plug to remotely reboot the host. Last crash happened this morning on PVE 8.2.4:

Code:
Sep 06 05:24:52 PVE1 kernel: BUG: unable to handle page fault for address: 00006c8bfbb76498
Sep 06 05:24:52 PVE1 kernel: #PF: supervisor write access in kernel mode
Sep 06 05:24:52 PVE1 kernel: #PF: error_code(0x0002) - not-present page
Sep 06 05:24:52 PVE1 kernel: PGD 0 P4D 0
Sep 06 05:24:52 PVE1 kernel: Oops: 0002 [#1] PREEMPT SMP NOPTI
Sep 06 05:24:52 PVE1 kernel: CPU: 12 PID: 513 Comm: dp_sync_taskq Tainted: P     U     OE      6.8.8-4-pve #1
Sep 06 05:24:52 PVE1 kernel: Hardware name: Micro Computer (HK) Tech Limited Venus Series/AHWSA, BIOS AHWSA.1.22 03/12/2024

---
Sep 06 08:40:28 PVE1 kernel: microcode: Current revision: 0x00004121
I haven't changed any power limit so far. Please let me know in case you find any solution!
 

anewsome

Active Member
Mar 15, 2024
112
107
43
After a few months of frustration and BIOS updates, I've finally seen a bit of stability in the 5 node MS01 cluster. I'm currently on the BIOS version 1.24, with the option to reduce the memory speed (which I've done). Since v1.24 BIOS update, I haven't seen any of the old crash behavior. I did update to the latest microcode weeks ago, but the Proxmox version remains the older 8.1.11. I've been reluctant to upgrade and introduce more variables in tracking down cause of the crashes.

There was a power outage that outlasted my UPS reserve a few weeks ago. If not for the power outage, I'd have at least a month of uptime across the 5 nodes.

I don't know when the 1.24 version will be publicly available, or maybe it already is. When it does become available, I'd recommend installing it and reducing the memory speed for more stable operation.
 

Sartorius

New Member
Sep 6, 2024
1
0
1
Newbie to the forum but long time reader of STH. I picked up a 12600H barebones on Amazon Prime day for Homelab work and perhaps as a PFsense router for my multi-gig Google fiber. After reading wadup's post on page 30 about replacing the thermal paste, I was able to get the foam cover off in one piece. Here's the view of the screws to remove the heatsink for replacing the thermal compound. I've got some Thermal Grizzly Hydronaut on the way. I don't see any resistors or anything else around the CPU. Do I still need to apply some nail polish or another insulating liquid around the CPU in case of liquid metal compound leakage?

Is there any reason to reapply the foam cover to the heatsink? Does it keep much heat from escaping back into the case?

Many thanks! This forum is incredibly helpful.
 

Attachments

bbusse

New Member
Sep 7, 2024
3
2
3
After a few months of frustration and BIOS updates, I've finally seen a bit of stability in the 5 node MS01 cluster. I'm currently on the BIOS version 1.24, with the option to reduce the memory speed (which I've done). Since v1.24 BIOS update, I haven't seen any of the old crash behavior. I did update to the latest microcode weeks ago, but the Proxmox version remains the older 8.1.11. I've been reluctant to upgrade and introduce more variables in tracking down cause of the crashes.

There was a power outage that outlasted my UPS reserve a few weeks ago. If not for the power outage, I'd have at least a month of uptime across the 5 nodes.

I don't know when the 1.24 version will be publicly available, or maybe it already is. When it does become available, I'd recommend installing it and reducing the memory speed for more stable operation.
What hardware specs are you running? I've been through several forums looking at these issues and still considering purchasing. Are you running 96gb or 64gb of ram, and which kit? 12900H or 13900H ? I plan on buying a couple of these to replace my aging Dell R620's running ESXi and figured i'd get some MS-01's and run Proxmox due to Broadcom F'ing everything up for home users ;)
 

PigLover

Moderator
Jan 26, 2011
3,215
1,571
113
Planning to use Samsung pm9a3. Will this work out of the box or do I need to change any voltage settings in bios?

View attachment 38824
Hard to tell from the picture and I can't find any info on the product link from Ali - but it appears to be a full-height PCIe card. Based on the dimensions of a 2.5inch SSD I don't think you could fit the drive on anything but a full-height card. MS-01 only has room for half-height cards.

So, no, it is not compatible (unless it really is a 1/2 height card!).
 
  • Like
Reactions: P00r

bbusse

New Member
Sep 7, 2024
3
2
3
Hard to tell from the picture and I can't find any info on the product link from Ali - but it appears to be a full-height PCIe card. Based on the dimensions of a 2.5inch SSD I don't think you could fit the drive on anything but a full-height card. MS-01 only has room for half-height cards.

So, no, it is not compatible (unless it really is a 1/2 height card!).
The specs of the product do say the dimensions, which are 132x83x18mm. Back plate is 120mm edge to edge.

Per the googles, Low-profile PCI cards have a reduced bracket height of 79.2 mm. So yeah, it's too big.