The Sad State Of My NAS
It's been nearly a month since the DreadNAS went down, and it's been painful to say the least. I've ran into many moments of frustration since it's been down. My co-workers couldn't download some files I host and complained. I couldn't access my Plex server, so I had to spin up a temporary Plex on my second server. My music, accessible via Navidrome, was also down. Worst of all, the wife complained about not being able to archive her files. She was about to buy an external hard drive for storage... gross.
Frustrated, I decided to poke my head in and see if I can get it back up and running. It started with a quick search on the web. "5950x freezing" was close to what my search was, and I ran into a few people on Reddit complaining about c-states causing freezing very similar to what I was experiencing. Firing up my JetKVM, I booted up the machine for the first time in a long time. In the BIOS, I found a number of things misconfigured. I quickly made corrections, although none of it would cause the freezing I was seeing. After disabling c-state control, I saved and reset the machine. Maybe this will resolve my problem, and frankly, it doesn't sound out of the question.
Getting back into Proxmox was fun. I had a lot of cleanup to do since reinstalling it from scratch. Docker wasn't working, since something about my custom config I had migrated over from the backup wasn't quite sitting right. Long story short, I reinstalled docker. It's up and running again, although I had to reconfigure everything in the daemon.json and fix the storage driver. I also had to install the Nvidia drivers and the Nvidia Container Toolkit to get it working with Docker. I was able to get hardware transcoding to work both with Jellyfin and Plex, which is HUGE! I did just buy the Plex Pass on Black Friday. I suppose they deserve some cash for the many years it's served me, even if I don't like where they are going. Jellyfin may be the future, but Plex is still worlds easier for friends and family.
With Docker working, I brought up every container listed on my Uptime Monitoring Dashboard, along with a bunch of other essentials. With that complete, we are in a good place right now with the configuration... I did have to configure my 10g interface to use an MTU size of 1500 vs the 9000 I had set for proper 10g ethernet... For some reason, the interface kept flapping every 3 minutes or so, causing huge connection issues. Changing the MTU size helped stability, although it annoys me to no end that it isn't set properly. My first instinct is to buy a new 10g NIC off ebay, however, the Threadripper Motherboard I was planning to buy has a 10g interface built-in. I'll just wait and suffer with 1500 MTU for now.
So where does that leave us? Not sure yet, since the machine hasn't seen uptime longer than 2 hours yet. I've been tweaking it and rebooting it, and so far no freezing- but I wouldn't be surprised if right after writing this article, it freezes yet again. Ohh well- this was a last ditch effort to save money, since I don't have the cash to spend on a new server at the moment- not to mention the lack of available R-DIMMs even if I wanted to buy now. I'll update this article if it does go down, but in the meantime I'll be keeping my fingers crossed.