Hello!
I’ve been running unRAID for about two years now, and recently had a thought to use some spare parts and separate my server into two based on use. The server was used for personal photos, videos, documents, general storage, projects, AI art, media, multitude of docker containers, etc. But I was thinking, it’s a bit wasteful to run parts that I use once or twice a week or less 24/7, there is just no need for the power use and wear and tear on the components. So I was thinking to separate this into a server for storage of photos, videos and documents powered on when needed, and then a second server for the media which can be accessed 24/7.
Server 1 (photos, videos, documents, AI experiments): 1 x 16TB parity, 2 x 14TB array. I7 6700k, 16GB ram
Server 2 (media, docker): 1 x 10TB parity, 1 x 10TB and 2 x 6TB array. Cheap 2 core skylake CPU from spare parts, 8GB ram.
With some testing, server 2 only pulls about 10w while streaming media locally, which is a huge drop from the 90+ watts at idle that it was running when I had everything combined.
I was hoping to use an old laptop I have laying around for the second server instead, which has an 8 core CPU, 16GB ram, and runs at 5w idle. I have a little NVMe to SATA adapter that works well but the trouble is powering the drives reliably.
Anyways, pros of separating it out, lower power usage, less wear and tear on HDDs so I will have to replace them less frequently.
Cons, running and managing two servers.
Ideally, I’d like to run server 1 on the cheap 2 core skylake CPU (it’s only serving some files after all), server 2 on the laptop with 8 cores (but still have the issue of powering the drives), and then take the i7 6700 for a spare gaming PC for family.
Alternative would be to just combine everything back into one server and manage the shares better, have drives online only when needed, etc. But I had issues with this, and would sometimes log into the web ui to find all drives spun up even though nothing was being accessed.
Anyways, I hope all of that makes sense. Any insight or thoughts would be appreciated!
I actually tried this as my second step in trouble shooting, the first being using different ports.
In the non-omada management software, it defaults to 10G, and if the devices is on before the switch it negotiates 10G correctly and works at full speed (tested with iperf3). As soon as any of the 10G connected devices is rebooted, I’m back to 1G. To fix it, I then have to set the port to 1G with flow control on, apply changes, save config, refresh page, change to 10G with flow control off, apply, save config and it goes back to 10G again. Alternatively I can reboot their switch and it’s fine again.
In Omada its the same, fewer steps to get there but I have to sometimes do it 2-3 times before it works.
Same issue with both 10G TP-Link switches, so I’m thinking it might be the SFP. Using Intel SFP+ with FS optical cables. I’m using a DAC for the uplink from the 10G switch to my unmanaged 2.5G switch, and that doesn’t have the problem of dropping, always works max speed.