TL; DR I would recommend only running v6.10.x on a server with a Brodcom NIC that uses the tg3 driver if VT-d/IOMMU is disable or it might in some cases cause serious stability issues, including possible filesystem corruption.
Another update since this is an important issue, there's a new case with an IBM/Lenovo X3100 M5 server, this server uses the same NIC driver as the HP so this appears to confirm the problem is the NIC/NIC driver when IOMMU is enable.
Known problematic NICs:
HP Microserver Gen8:
03:00.0 Ethernet controller [0200]: Broadcom Inc. and subsidiaries NetXtreme BCM5720 Gigabit Ethernet PCIe [14e4:165f]
DeviceName: NIC Port 1
Subsystem: Hewlett-Packard Company NC332i Adapter [103c:2133]
Kernel driver in use: tg3
IBM/Lenovo X3100 M5:
06:00.0 Ethernet controller [0200]: Broadcom Inc. and subsidiaries NetXtreme BCM5717 Gigabit Ethernet PCIe [14e4:1655] (rev 10)
DeviceName: Broadcom 5717
Subsystem: IBM NetXtreme BCM5717 Gigabit Ethernet PCIe [1014:0490]
Kernel driver in use: tg3
HP ProLiant ML350p Gen8
02:00.0 Ethernet controller [0200]: Broadcom Inc. and subsidiaries NetXtreme BCM5719 Gigabit Ethernet PCIe [14e4:1657] (rev 01)
DeviceName: NIC Port 1
Subsystem: Hewlett-Packard Company NetXtreme BCM5719 Gigabit Ethernet PCIe [103c:3372]
Kernel driver in use: tg3
This driver supports many different NICs, unclear for now if all are affected or just some, also unclear if AMD based servers with AMD-Vi/IOMMU enable are affected, but for now I would recommend only running v6.10.x on a server with a Brodcom NIC that uses this driver if VT-d/IOMMU is disable or it might in some cases cause serious stability issues, including possible filesystem corruption.
When there is a problem with one of these NICs and VT-d you should see multiple errors similar to below in the logs not long after booting, usually before a couple of hours of uptime:
May 21 15:53:05 Tower kernel: DMAR: ERROR: DMA PTE for vPFN 0xb0780 already set (to b0780003 not 28dc74801)
May 21 15:53:05 Tower kernel: ------------[ cut here ]------------
May 21 15:53:05 Tower kernel: WARNING: CPU: 1 PID: 557 at drivers/iommu/intel/iommu.c:2408 __domain_mapping+0x2e5/0x390
If you see that stop using the server and disable VT-d/IOMMU ASAP, there's no need to disable VT-x/HVM, i.e., you can still run VMs (but without VT-d/IOMMU can't passthrough any device to one).
For Intel CPUs VT-d can usually be disabled in the BIOS, alternatively you can add intel_iommu=off to the syslinux.cfg append line, on the main GUI page click on flash and scroll down to "Syslinux Configuration", then add it to the default boot option, the one in green) :
In either case confirm it's really disabled, you can do that by clicking on "system information", top right of the GUI:
Original post here:
https://forums.unraid.net/topic/123620-unraid-os-version-6100-available/?do=findComment&comment=1128822