supagold

Thanks for the followup. I have narrowed it down to either my pihole or unifi dockers, which are the only ones that have an assigned IP. The other dockers are currently up and running with normal stability. This does seem to match the thread that JorgeB reference, the difference is that it looks like those people were adding new hardware, where mine is existing hardware that's perfectly stable under 6.8.x and only shows issues with 6.9.

Thanks for the response! Is there something in the diagnostics that's indicating this might be the cause of the crashing? I read through that thread, and I'm also using a pihole docker, so I'll disable that for the time being. (Pretty sure that's my only one with a mapped IP.) I feel like I might've read that 6.9 modified how some of the docker networking works - is there a better way I might map an IP onto a container now?

I recently decided to upgrade from 6.8.3 to 6.9.2. I first attempted this a couple weeks back, and immediately noticed some pretty severe stability issues. I'd typically end up rebooting my system for one reason or another, but it wasn't uncommon to have uptimes of several months. After upgrade, my server would typically crash within 12-24h. The symptoms would often first manifest as inability to load docker or VM tabs in the web interface, up to loss of the web interface itself. Typically I could still ping the server, though not in every case. At the time, I thought it might be related to a mining VM I was running on my 1080ti. (I thought possibly the card might be going bad.) Eventually, I rolled back to 6.8.3, and the stability issues disappeared. Recently I removed that video card to a separated machine (where it's been rock-solid, fwiw), and decided last night to attempt moving back to 6.9.2 again. The same thing is happening - I think I got about 12-14h of uptime before the system crashed. I could roll back to 6.8.3 again, but if it's possible to figure out the issue that would be great. I've attached a diagnostic. Any help that might be provided is appreciated. Thanks! unraid-diagnostics-20210422-0928.zip

Yeah, I did that. I also disabled passthrough for the onboard NIC I was testing and have it just on br0, and tried updating the virtual MAC address. Any other ideas?

OK, that did fix my issue. Thanks! However, now the VNC adapter won't connect. It comes up, then nearly instantly fails with "Failed to connect to server". I'm not sure what logs to pull to diagnose this.

Thanks for the advice @bonienl - before I make the change, can I ask what the SYSNICS configuration does?

Added to my original post.

Recently my motherboard died, and I replaced it with a new motherboard of the same model (same hardware rev, as well). I was hoping that Unraid would be able to just pick up and run without having to modify much, which it has, except that my windows VM, which was working perfectly, will no longer start the onboard ethernet adapter that I'm passing through. Windows Device Manager gives "code 10" "cannot start". This is the case both in the VM I had been using, and a from scratch install I just tried. I've noticed that Unraid (and windows) is showing my NIC as "(2)", as in this: IOMMU group 15: [8086:15b8] 00:1f.6 Ethernet controller: Intel Corporation Ethernet Connection (2) I219-V I've also seen error messages that "eth4 is missing" when the system is booting, but I haven't been able to find that in the System Log. (I've got a 4 port intel nic that is eth0-3.) This is the XML for the new win10 VM I just created: https://pastebin.com/sJtU0b9M And I have this in my syslinux config at boot: append pcie_acs_override=downstream vfio-pci.ids=8086:15b8 pci-stub.ids=10de:1aec,10de:1aed initrd=/bzroot Forgive me if I should have included more troubleshooting information - I'm fairly new at unraid. Thanks in advance for any help you might be able to provide! unraid-diagnostics-20191014-2012.zip