ken-ji

You can assign each docker with a static IP using the Fixed IP address (optional) field. Then you make static assignments to the router/DNS server. FYI router-level DHCP has no effect on dockers. that's why the custom docker network IP range needs to be carefully selected to prevent collisions with the LAN IPs

This is weird. Other than the proxynet bridge, did you create any other bridges by cli? because the bridges in your network interfaces (from the diags) named br-* do not line up with those the docker is reporting... only docker networks of type bridge can be seen with the ifconfig -a Maybe @bonienl has some idea...

I don't see anything wrong. except there are 4 docker custom networks defined. docker0 and three br-* networks. Can you run docker network list on the terminal?

Post you diagnostics file. I'm guessing there's something wrong in your config that's causing the docker to treat the br0 network as a custom definition and possibly purging it. You might also want to provide the screenshot of your docker network settings 'Settings | Docker'

AFAIK, this can be done but has a bunch of depends. Solutions in my preferred order. Easiest, but potentially most complicated for the average Joe - is to have a DNS service in your LAN which would serve your custom DNS entries. Some consumer routers, and practically all pro/expert routers like cisco/mikrotik/*-wrt/pfsense have this functionality. Pi-Hole is also another option from what I understand. depending on the clients you have at home windows PCs can look for names from a WINS server - so a linux box/vm ; a windows DC can provide this service along with a means to register the new names macs can look for names with the Bonjour service (not too familiar with it as I don't directly have macs, and use the DNS approach) which is something like mDNS; linux box/VM can support this using avahi everything else would be limited to the LAN DNS solution.

Mikrotik: Its only a suggestion... but this is a router. so you actually replace your router with it. There is a caveat, it can be hard to setup. Other suggestions: pfsense firewall appliance (you can use a PC for this) H200 and H700 - they are SAS/SATA RAID controllers so they feature the same cable connections. Key point here is that Unraid should ideally be using HBA controllers (pure JBOD only). RAID controllers (even with JBOD) are not preferred, because they can interfere with the access to the disks (migrating the system later might still need the RAID card) - You should search the forums, but I think the H200 can be crossed flash to work as an HBA (IT mode)

That was the manufacturer's product page A link (Amazon) https://www.amazon.com/Mikrotik-RB750Gr3-5-port-Ethernet-Gigabit/dp/B01MSUMVUB

Auto start is exactly what it says - the docker will auto start on array start. What you want is to add --restart unless-stopped to the Extra Parameters option

👍Don't know why, but my eyes don't feel strained staring at white pages... Dark pages just make me groan in frustration...

Well, I can only state based on my experience Emby does work fairly well as a docker with intel iGPU (Kaby-Lake, Coffee-Lake in the 6.6 series) to do accelerated transcoding. of course there's a single stream limit I think for iGPU HW decoding...

Apparently my motherboard has that particular LAN controller. So I guess it works.

Sorry if I wasn't clear, but iGPU works, dGPUs doesn't for now.

That should work. but you will need to make sure your VM is not NAT'd (connect it to eth*/br*/bond* interfaces not virbr0) It should be made clear to you (as you might be confused/misunderstanding based on your statement) VMs in unRAID are plain-jane KVM VMs and dockers are another beast altogether. That said, I'm doing some GRE tunnels too for site-to-site VPNs, but they're all on the router level ( https://mikrotik.com/product/RB750Gr3 ) - which incidentally might be a better architecture for your needs.

AFAIK, the standard internal SATA data and power connectors are keyed and designed for hot plugging. the power(or was it ground) leads will connect first to mitigate voltage drop/current inrush. But yeah, its better to have hotswap bays so you add/remove the drives in one motion and less likely to cause physical damage

Just to resolve some questions you might have. GPU based transcoding is limited to VMs in unRAID, because there are no GPU drivers on the Host to pass the GPU to the plex/emby docker. (I prefer emby too as there's no need to phone home and that stuff) Personally, as a single/dual stream user, I just use a Pentium G4620 and enable the iGPU to assist with the transcoding in Emby.