joey911 Posted February 18 Share Posted February 18 Howdy all. Background: I built my new system apparently looking at the old style documentation (prefer cache) and I must have hit things mid-cycle because I overlooked cache pools; I just saw cache disk (singular). I'm looking for some thoughts on how to improve my setup since I just installed Plex. I also want to pick your brain for how I might make this setup a bit better. Hardware: 40 core Xenon CPUs 128gig DDR 4 memory 8 Bay tower 1TB SSD cache drive 2 Parity Drives 2 highly populated 4TB drives from my previus Unraid server 3 new 2TB enterprise disks barely populated 8 Bay USB 3 extension tower 8 barely populated disks of various sizes Nvidia card for transcoding Problem 1: Any time I make any changes or updates to Docker applications, my system freezes. Nothing responds until Docker is done. Problem 2: Docker apps are generally slow Plex is VERY slow. Calibre is slow. Channels DVR is not slow but I plan to give this up in favour of Plex. What I've done: Moved Plex transcoding to the cache drive Moved Timemachine backups to the cache drive (to relieve the array from constant usage) Other Information: Because I only have 1 SSD cache drive, I did not move system and appdata because if the cache drive fails, I'm toast. I wish there was a way to back it up to the array in case it fails so I don't lose anything. Maybe there is and I haven't found it yet? (See that question below) Because of the previous line, I also didn't bypass fuse by specifying /mnt/cache. The original intent of the server has been for NAS purposes. Because of this, I highly value data redundancy. The introduction of DVR workloads is new but it's been successful with the family so I'm going to continue this. I have some VMs running on the cache disk and I'm worried about being able to recover if I lose the cache (see further down). Things that I've been considering: Adding another SSD and add it to the cache pool enabling me to move over system and appdata I'd have to evacuate and remove a lower sized disk from the USB array and then move one of the disks from the tower. I'd hate to move one of the high quality new disks but that would be the easiest. This choice is a bit of a pain due to the tower but I can do this. Remove 2 or 3 of the new enterprise disks from the array and make them a new cache pool and move over system and appdata. Things I don't understand and could use some guidance on: Is there a way to backup the cache pool such that if it dies the system will use the array? i.e. a "copier" vs a "mover" Is data replicated across the cache pool so if I have two cache pool disks and I lose one the system will continue to function? I really appreciate any guidance you can provide me. Thanks! Quote Link to comment
JonathanM Posted February 19 Share Posted February 19 Appdata Backup plugin will make backups to the array if desired. Single device pools are allowed to be XFS, with no option to mirror, multi device pools must be BTRFS or ZFS. So, if your current pool is XFS you would need to back up the data before adding a device and changing to format to allow a mirror. If it's already BTRFS or ZFS you can set up a mirror without erasing it. Data redundancy is NOT backup, it's high availability. Subtle distinction with massive repercussions. Much data loss is deletion or overwriting, whether accidental user error or malicious. A single copy of your data is vulnerable, you must have backups elsewhere of critical data. Many of us run second servers for backup purposes. 1 Quote Link to comment
joey911 Posted February 19 Author Share Posted February 19 38 minutes ago, JonathanM said: Data redundancy is NOT backup Thanks, yeah I misspoke there. I am using the NAS aspect for backups. I don't need HA. I have appdata installed and backing up my VMs only. My single cache drive is BTRFS. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.