Pulling around 200W on average.
Looks like home assistant
For storage redundancy RAID 5 is not recommended, particularly as you get to high capacity drives (think >8TB). I think the rating to consider is URE (unrecoverable read error, usually 1 in 10^14 bits read).
Once a drive inevitably fails and you are forced to resilver the array to avoid data loss. During the resilver the healthy disks are running at 100%, reading every bit of data they have to complete the parity calculation and determine what data is missing. The chances of encountering a URE on another drive is a near certainty at high capacities as the total number of bits read exceeds the URE rating. As result the resilver would fail and the array would be lost.
RAID 6 as a minimum (2 drive redundancy), although a popular option now (and the layout I use) is mirrored vdevs.
Edit: Consider TrueNAS for NAS software. I have been using it for 10 years and it is absolutely rock solid. 25TB usable storage across 4x mirrored vdevs. I run it as a VM inside Proxmox with 4 logical cores on a 10 year old Xeon with 16GB RAM for the VM (I run ECC as was recommended at the time, but whether it’s still considered necessary I’m not certain).
I would also recommend getting an LSI HBA (host bus adapter) like the 9207-8i flashed to IT mode (it must not be in raid mode, let TrueNAS manage the disks directly). This simplifies passing through all the disks to a VM.
This is what I do as well. I have a public DNS record for my internal reverse proxy IP (no need to expose my public IP and associate it with my domain). I let NPM reach out to the DNS provider to complete verification challenge using an account token, NPM can then get a valid cert from Let’s Encrypt and nothing is exposed. All inbound traffic on 80/443 remains blocked as normal.
I use the Amplipi from Micro-Nova for whole-home audio and I love it. It’s local, open source and has a Home Assistant integration.
The main unit has 6 zones, but expansions units can be added. I think it supports up to 4 simultaneous streams. We use 2x AirPlay streams, and a turn table connected via RCA, but many other options are supported. They detail it all on their website and GitHub repo.
You mention frigate specifically. Were you running this on the system when the drive failed, or is this a future endeavour?
I bring this up because I also use frigate, and for some time I was running with a misconfigured docker compose that drove my SSD wearout to 40% in a matter of months.
Make sure that the tmpfs is configured per the frigate documentation and example config. If misconfigured like mine was, all of that IO is on disk. I believe the ramdisk is used for temp storage of camera streams, until an event occurs and the corresponding clip is committed to disk.
Good luck!