Using a Server Cluster for Unraid and Plex

[Yeremyah777]

I know this topic has come about occasionally, but I have yet to see anyone actually try to do this and post the results. I am a 4k plex WHORE. Because of this, I have to transcode on the regular, and it takes a TON of processing power. Since I don't have a dedicated GPU to offload the transcode, and my servers are pretty old, I am thinking about "building" a cluster box to handle all of the processing.

 

My question is, would this be worth the time and energy to do so? Currently, I have no electric bill, and have a lot of time since I became disabled. Unfortunately, I am VERY budget limited now. In the past, I was able to obtain a decent amount of older equipment as listed below. I

 

Included devices:

1x - Dell Poweredge r410 with dual Xeon e5620 Procs and 8GB 1866 ECC-DDR3

1x - HP Proliant ml350 g5 with dual Xeon e5620 procs and 48GB 1866 ECC-DDR3 and an Nvidia gt 1030

1x - HP ProLiant DL380p Gen8 with dual Xeon E5-2650 v2 procs and 64GB 1866 ECC-DDR3 HP Smart Memory.

 

Altogether I have 17 different HDDs/SSDs

I'm not concerned with the storage solution, as I have plenty of HBAs and Raid controller cards. I DO wish I could pool all these devices together to make the processing capabilities even stronger. I am running OpnSense on the r410 right now, which is complete overkill. The ml350 is nothing more than a download Server, and the DL380 is running unRaid with plex, and some Game servers on it. Plex alone will tax the system to 80 or even 90% transcoding just one or two 4k streams. I cannot afford a new GPU; the 1030 cannot transcode, and my 1070 in my gaming rig is used constantly for streaming to Twitch, gaming, and whatever else. The gaming rig has a 5950X, and only 16GB or 3600-DDR4.

 

So between everything, I would like to have my gaming machine, and have the ability to run all 3 Servers under a cluster OS like Proxmox, and feed all of the cores/threads/RAM and storage to unRaid, so I can run OpnSense as a VM, and all of the dockers. It'd also be nice to have the processing leftover to run some experimental VMs, an IOS box, and some more docker containers. Is this a bad idea? Should I consider something else? Should I leave everything as is? Does anyone have good advice for testing to see if this would work as a cluster. I understand that mis-matching CPU eras in a cluster is not the best idea, but I'm not sure if there's a better alternative. I will be making some money over the summer, and upgrading the CPUs in the DL380p to dual Xeon E5-2697 V2 procs, but I'd like to add that processing power to the mix, instead of relying entirely on a single machine to do the transcodes when it'll still tax the crap out of them transcoding 4k to 1080p and 2k videos. I do have direct play enabled, but even then, It kicks the crap out of the system resources. Any ideas, or advice would be sincerely appreciated.

Edited June 1, 2023 by Yeremyah777

[whipdancer]

What are you watching on that it requires transcoding? It would be far cheaper to upgrade your player.

[Yeremyah777]

Sadly just a TV, phone and desktop HTPC app