jfoxwoosh

Can you post screen captures of your unraid's "Networking Setting" page? Also, did you try traceroute from your server to devices on private vlan? You want to compare the path taken by: 1. device_on_private to server_on_core 2. server_on_core to device_on_private

Run your unraid in safe mode (this disables docker/plugin/VM) and without any external USB device connected. See if lockup happens again. If not, then you can start turning on additional services one by one.

If you can use domain name to access your services from within your home, I think you already have something like NAT reflection enabled (or perhaps split DNS)? However, since you said you can still get access as long as you are on the core vlan through domain name, I am guessing you might have asymmetric routing happening. You might want to start using tracert (on windows) and traceroute (Linux) from your each vlan to check the routing path from your server to device and vice-versa. If they take different path, then your firewall will most likely block the tcp connection from establishing.

Wow. Your problem sounded exactly like mine. I had this as well, and only until last Dec, i got everything sorted. When you say you cannot access nextcloud from private vlan, are you trying to access it through your subdomain.domain.com or by IPaddress:Port?

you have 2 GPUs (does the server also have a iGPU?), and the 1660 is vfio binded. Is your monitor connected to the 1030? If your monitor is connected to the 1660, then it will stay at screen and won't go any further which is perfectly normal. Which GPU is placed in the primary pcie slot. Does your motherboard has option to select choosing the primary GPU?

Give this a read. It explains why UEFI mode introduce difficulty to GPU passthrough when there is only 1 GPU in the system. There is a potential work around by adding "video=efifb:off" to syslinux config.

My UnRaid server (LAN 192.168.1.0/24) and VM (VLAN10 192.168.10.0/24) are on different subnets. I am trying to understand why the UnRaid's VM cannot access home assistant docker's web UI (or any other docker service with a web UI) through the server's IP+Port directly. For example, the VM cannot access the UI by server's IP:Port (e.g. 192.168.1.100:8123 for home assistant). The VM can access unraid's web gui just fine, and both server and VM can ping each other with no problem. My work around for now is to access it through reverse proxy (w/NAT reflection) so the VM traffic will go out to WAN and come back.

This is very similar to what I had before. See here. What solved it for me was to change the cpu idle current setting in the bios.

have you tried changing your usb flash drive? I had something similar like this before and found out my flash drive was dying.

msi b550 gaming plus

I am out of ideas. Did you by any chance jump through major updates like from 6.9.x directly to 6.11.x? I wonder if this is detrimental or not. All I can think of is to start over with a fresh trial install and test the passthrough, unfortunately.

Even with a brand new VM? Have you try creating a Linux VM?

The first big decision for you is whether to run HA as docker or VM. If VM, then you can do exactly what you do currently (I am assuming you are running full HomeAssistant OS right now). You can simply export and then import your instance. If docker, then you will have to understand there is no add-on store on HomeAssistant-Core, and many services will be run as a separate dockers. The transition is quite easy (I used to run full HA on rasp pi), just put all the dockers on the same docker network, and everything can then be configured in the UI just like before.

Depending on your MB and having a 2nd GPU in the system, you will need make some changes if you want to try UEFI boot. I have to add video=efifb:off myself.

I have to use VBIOS, or it won't work for me. You can simply download it from techpower up. I will stick with legacy boot. It will give you less issues.

I have two VMs, Windows 11 and Linux Mint 21, using my RX 6600XT and encounter no problem. My server is on 6.11.2. Is your MB configured to use legacy or UEFI boot? Did you vfio bind your GPU? Are you using a VBIOS? Have you tried creating a new VM from scratch and see if that work just for troubleshooting purpose? Don't passthrough any device other than the GPU.

If you intend to run HA as docker, then you have to run Frigate as a separate docker since the home-assistant-core does not have the add-on store. You can use HACS to integrate HA and Frigate dockers.

Do you have any CPU related overclocking or undervolting enabled?

@thibfighterDid you set up plex docker to transcode to RAM? If so, try disabling that first.

I switched over to frigate NVR because its better integration with Home Assistant. I can set up automation based on objects and zones. I also use Home Assistant app as my way of accessing the camera feeds from outside the house, which answer your question on not using the reolink app. I have been toying with compreface facial recognition (which also integrate nicely with frigate) lately, and this is when I found out reolink's camera image quality is a bit sub-par. I read some discussion on Home Assistant subreddit suggesting dahua, hikvision, or amcrest instead. The storage space used really depends on how you setup the NVR. For example, I set it to only record when "human" object is detected in my drive way, and the recordings are only kept for 14 days. I have about 350GB of recording data right now for your reference.

Please watch this video by spaceinvaderone. Your isp could be at fault here. Instead using http validation, you should switch over to dns method.

Have you tried changing the hw acceleration option to VA-API?

@PlanetDyna take a look at the Complex Networks section of this guide. If unraid wiregurad setting UI is telling you to setup static route, then you will have to make some changes accordingly.

I am running 2 reolink poe camera (RLC-520) with my system. I used to run the shinobi NVR and recently converted to frigate NVR. Other than the image quality not being great, I have no complaint with the camera. Watch this series videos by spaceinvader1. It will give you idea for where to start.

I have identified the cause of the MCE error message. I need to set "typical idle current" in bios for the CPU as described in this post by JorgeB. Thanks! I remember having this set correctly when the server was built initially, however, after some MB bios updates later, I forgot to change it back when all bios settings got reset. It has been 30+ days now, and the server hasn't given any error message like before.