No pci detected (No 10 gbe card)

[Albahttiti]

Hello I have the X540-T2 Anyfabric card and after upgrading to 6.5 I keep getting this during the kernel load

"pci 0000:08:00.0: [firmware bug]: disabling VPD access (can't determine size of non-standard VPD format)"               - Picture attached.

 

Also the unraid doesn't seem to be detecting any pci network cards.

I also realized the if the power is interrupted to the server, upon booting it finds the network card but the array wont start (clicking start array does nothing)

no network diagnostic:

tower-diagnostics-20180331-1328.zip

 

network showing diagnostics:

tower-diagnostics-20180330-2232.zip

 

maybe @limetech can verify if Lenovo ThinkServer X540-T2 AnyFabric 10 Gb 2-Port Base-T Ethernet Adapter by Intel is supported in the 6.5 and if you can help with the matter.

BTW I only have an Asus RT-N16 router and the second port is connected to a windows 10 pc with intel x540-T2 card (I don't have a network switch)

Edited March 31, 2018 by Albahttiti
added details

[thestraycat]

I'm having issues with the X540-T2 as well on 6.5.1RC2 .... NIC is dissapearing from UNRAID and needing a reboot (sometimes a few)

[JorgeB]

11 hours ago, Albahttiti said:

Also the unraid doesn't seem to be detecting any pci network cards.

On the non network diags the NICs are not being detected by Linux, i.e., it's a hardware problem.

[thestraycat]

I'm testing another 10gb X540T2 in my server tonight... I'm wondering whether overheating is coming into play or not? In my case the cards are recognized and by 6.5.1rc2 .. and then

stop and come back after a few reboots. Tests with the 2nd card should uncover more.

[Albahttiti]

@johnnie.black I dont think so since Ubuntu is able to detect the network.

Edited April 1, 2018 by Albahttiti

[John_M]

Have you tried it in a different slot or, failing that, re-seating it?

 

[Albahttiti]

I did re-seating but this is an anyfabric device so it cannot be connected to regular pci slots.

[John_M]

What sort of socket does it use then? If it isn't PCIe it might well not be supported.

 

[Albahttiti]

It is motherboard socket that if I understand this correctly it uses the PCI Express 3.0 x8 interface

 

[John_M]

Hmmm. If you unplug it, are your other PCIe devices detected?

 

[Albahttiti]

I dont have any other pci devices 

[John_M]

Really? No onboard gigabit Ethernet, for example?

[Albahttiti]

@John_M it only have the management port no other network ports.

[JorgeB]

8 hours ago, Albahttiti said:

@johnnie.black I dont think so since Ubuntu is able to detect the network.

 

Quote

00:1f.3 SMBus [0c05]: Intel Corporation C610/X99 series chipset SMBus Controller [8086:8d22] (rev 05)
    Subsystem: Intel Corporation C610/X99 series chipset SMBus Controller [8086:7270]
    Kernel driver in use: i801_smbus
    Kernel modules: i2c_i801
01:00.0 Ethernet controller [0200]: Intel Corporation Ethernet Controller 10-Gigabit X540-AT2 [8086:1528] (rev 01)
    Subsystem: Lenovo ThinkServer X540-T2 AnyFabric [17aa:1073]
    Kernel driver in use: ixgbe
    Kernel modules: ixgbe
01:00.1 Ethernet controller [0200]: Intel Corporation Ethernet Controller 10-Gigabit X540-AT2 [8086:1528] (rev 01)
    Subsystem: Lenovo ThinkServer X540-T2 AnyFabric [17aa:1073]
    Kernel driver in use: ixgbe
    Kernel modules: ixgbe
04:00.0 RAID bus controller [0104]: LSI Logic / Symbios Logic MegaRAID SAS-3 3108 [Invader] [1000:005d] (rev 02)
    Subsystem: Lenovo ThinkServer RAID 720ix [17aa:1053]
    Kernel driver in use: megaraid_sas
    Kernel modules: megaraid_sas

 

NICs detected and driver loaded
 

Quote

00:1f.3 SMBus [0c05]: Intel Corporation C610/X99 series chipset SMBus Controller [8086:8d22] (rev 05)
    Subsystem: Intel Corporation C610/X99 series chipset SMBus Controller [8086:7270]
    Kernel driver in use: i801_smbus
    Kernel modules: i2c_i801
04:00.0 RAID bus controller [0104]: LSI Logic / Symbios Logic MegaRAID SAS-3 3108 [Invader] [1000:005d] (rev 02)
    Subsystem: Lenovo ThinkServer RAID 720ix [17aa:1053]
    Kernel driver in use: megaraid_sas
    Kernel modules: megaraid_sas

 

 

NICs not detected = hardware problem, possibly an intermittent problem, since sometimes they do work, same might happen with Ubuntu

 

 

[Albahttiti]

anyway I ordered a new pcie x540-T2 regular card to see if that solve the issue.

[limetech]

On 4/1/2018 at 3:59 PM, Albahttiti said:

I dont think so since Ubuntu is able to detect the network.

 

From  your diags (thanks for posting those btw), there is this line:

Mar 30 22:05:36 Tower kernel: ixgbe: Intel(R) 10 Gigabit PCI Express Network Driver - version 5.1.0-k

I'd be curious what version your Ubuntu install is using.

 

Here's why I'm asking.  There are two "ixgbe" drivers: one in-tree linux kernel driver, and one available from the Intel download site.  The Intel download is currently at version 5.3.6.

 

This is something which is fairly common in the linux world, that there exists an in-tree kernel driver for a device and also a vendor-maintained driver.  Most of the time one is based on the other.  Obviously this is a maddening situation: why can't a vendor just maintain the in-tree kernel driver?  I posed that question to a kernel developer who works on the in-tree ixgbe driver and this is their reply:

 

Quote

That's a good question, and it doesn't get talked about much.  The same team at Intel in fact *does* support both of the drivers - I know because I used to be a part of that team.  Yes, there are differences in the two drivers, because different customers have different needs.  The in-kernel driver must follow all the rules of a standard Linux driver, so has some restrictions placed on what it can offer.  The out-of-tree driver has extra features and some configuration variables that are not allowed in the Linux kernel, but which make the driver a little more flexible for some customers' needs.  Some customers are restricted to straight Linux kernel and the drivers it offers, and some customers are able to use alternative drivers.

There are efforts to keep the drivers in sync with bug fixes and changes in the Linux kernel, but they're not always perfectly aligned.  As for the version numbering, those are to be taken with a few grains of salt...  they are not really kept as closely in line as they should be, but then it also takes a little longer to get changes into the kernel.

 

The problem we have is that if we use a vendor-supplied driver, often (but not always), it will not compile without error on a new linux minor kernel release.  This presents us with a dilemma: we either are stuck on the current kernel or we  have to revert back to in-tree driver.