Pandemic

I'm looking for help setting up docker in the most ideal way. I've made other posts but they died out and rather than trying to modify what I have I'll just ask for the best way and do that. I've got an array of about 10 disks, one cache drive which is a 10k rpm raptor 150gb, and (for now) a 512gb NVME. I want to set up docker to run plex and ollama containers. Because ollama (and the UI) will be running LLMs I'm hoping to set this up to run as fast as possible. I need to figure out how to remove the raptor drive from the mix, I'll figure that out later. I'll run the 512gb NVME as the cache drive, I expect. Is a pool a mix of multiple drives (ex: two NVME drives?) I think running docker from the NVME will be best, keeping it outside the array. What should the docker size allocation be? I had it set to about 300gb just to keep space on reserve for the future but I'm not sure if that's necessary. Let me know your suggestions so I can try to understand how to future-proof this setup.

I was just coming in to mention the Plex container is currently using: 12.79GiB / 31.19GiB while there has been no transcoding activity for at least 8 hours. df -h /tmp/ shows: root@name:~# df -h /tmp/ Filesystem Size Used Avail Use% Mounted on rootfs 16G 1.1G 15G 7% / df -h /plex_tmp/ shows: root@Arc:~# df -h /plex_tmp/ Filesystem Size Used Avail Use% Mounted on rootfs 16G 1.1G 15G 7% / Both look fine but the docker tab in unraid is showing more consumption and over the 4gb I thought had been set.

Good call. I think I had been using it like that for a number of years but I changed it to /plex_tmp and used: --mount type=tmpfs,destination=/plex_tmp,tmpfs-size=4000000000 I'll report back with results. thanks!

I've noticed Plex is using 14.79GiB / 31.19GiB this morning when the last play was a few hours ago. This is the highest usage I've seen. /data/mnt/user/Media Library/ /config/mnt/user/Docker/Plex-Media-Server /transcode/tmp --mount type=tmpfs,destination=/tmp,tmpfs-size=4000000000 I thought the above would limit RAM usage for transcoding to 4gb. I'm just confused as to what's happening to cause almost 15gb usage in Plex from what should have been direct play over three hours ago.

I've been running plex on a quicsync proc for years now and never did this. I did notice 1-2 transcodes using a chunk of the proc so it was likely never utilizing quicsync. Thanks for putting this together!

If I'm alright with making sure the NVME disk is mounted each time, whats the benefit of running docker from a pool or array rather than the disk itself? edit: I have a 10,000 RPM raptor drive that's 150gb as cache right now. If I use the NVME 512gb drive to replace the cache drive and act as the docker directory, would that be the ideal way to set this up?

I don't think i understood what I was doing when I set it up years ago. I'm finding the common question from others is why this is so different than default. I think it'd make the most sense to purge everything and try to rebuild but I'd like to explore how best to set it up in the future. I run media from Plex, I run Storj and share a disk outside the array, and I'm hoping to run ollama UI. I'm hoping I may be able to share the GPU between plex and ollama when one is not in use.

I'm spinning my wheels and I don't know how to dig my way out here. Originally I set up Docker to use Disk 7 of the array. I want to use Ollama UI along with a GPU to host a local LLM and thought moving docker to a SSD would be a good idea. I copied the docker-xfs.img as well as the 'Docker and VM' folder. Docker settings show the following locations: Docker vDisk location: /mnt/disks/BE77072214FF00210169/docker-xfs.img Default appdata storage location: /mnt/user/Docker and VM/ My confusion is that I have 'docker and vm' on Disk7 still (never removed yet), as well as the SSD (makes sense) as well as /user and /user0 I don't understand what containers are using what information. When I moved the docker-xfs.img and 'docker and vm' folder from Disk 7 to the SSD, Plex reverted to default and I lost my play history, etc. I don't understand why that happened. I need to know if this is worth fixing/organizing or if I should scrub everything and start over from the SSD. Is the /user folder a part of the array? If so, is there a way I could make Docker seperate from the array in an effort to allow the drives to spin down and possibly save power? Will utilization of the 512gb SSD make Ollama and local LLMs that much faster? I copied the contents of Disk 7 'Docker and VM' folder to the /user 'docker and vm' folder and I'm getting this error with plex: Stopping Plex Media Server. Starting Plex Media Server. Error: Unable to set up server: sqlite3_statement_backend::loadOne: attempt to write a readonly database (N4soci10soci_errorE) Storj has an error too saying: Error: Error opening database on storagenode: database: orders opening file "config/storage/orders.db" failed: unable to open database file: no such file or directory Storj and Plex had been working fine after the move. Storj started the error last week and Plex started this error after I copied the 'docker and vm' contents from Disk 7 to the /user directory in an attempt to restore historic data. Thanks for the help. arc-diagnostics-20240819-0953.zip

The Docker and VM was set up, originally, on disk 7 and, yeah, the parity and disk 7 always spun. I moved everything over to an SSD and allocated 300gb to it because I'm planning on installing the ollama UI along with a few models and wanted to make sure I had enough space. Now that disk 7 is not associated with docker, I'm still finding the parity and disk 1 are almost always spinning for some reason. Disk 1 is where the bulk of media is stored so maybe plex is keeping it spinning, I'm not sure. Why did I change away from standard defaults? I'm not sure. I may have set it up not understanding what I was doing or maybe there was a reason for it. I'd set it up from scratch but I have a storj instance running and I seem to remember it was a pain to get working. I'd also like to figure out why my plex server had to be set up from scratch causing my watch history to be lost.

I've attached the diagnostics file to help clarify. arc-diagnostics-20240731-1121.zip

Good point, sorry about that. arc-diagnostics-20240731-1121.zip

Most of the array is not in use so the drives can spin down. I moved docker to the SSD so the HDDs should be able to spin down. The thing is, parity and disk 1 (the two 12tb drives) are constantly up. Is it possible to see what's accessing them to see if they can be off-loaded? I was thinking this may help save a bit of power but, so far, it seems to be a lot of work for nothing. I appreciate the help.

Appdata is under mnt/user/docker but this directory didn't change, I don't think. I have an application support folder in the docker folder that I copied to the SSD. I copied the docker folder from the HDD to the SSD and pointed docker to the new location. Appdata location did not change.

Good to know. Where would this appdata normally be kept? I see under /mnt/user/appdata but there is nothing there for plex.

I wanted to check here to see if there's a way I can move my plex history. I copied the docker-xfs.img file as well as the docker folder from the drive on the array to the SSD. I would assume that would have moved the play history as well.... but it hasn't. Any ideas?

It is enough for now assuming the 3 onboard SATA ports + 8 through the expansion card are stable (so far they are) but I'm trying to plan for the future and/or have the most stable system running. I've got 2 more empty ports on the motherboard as well.

I'm running two Supermicro 8-Port SAS/SATA Card - (AOC-SASLP-MV8) for a 15 HDD bay Unraid server at the moment. I want to put a GPU in the PCIEx16 slot so I only have room for one of these cards. I've got 11 HDDs right now so I'm using 8 ports on the SAS card plus 3 SATA ports on the motherboard. I an probably continue like this for a few years, to be honest. I'm thinking about how best to go forward. I only really use about 2-3 drives at any given time. So my options are to look around for two 8 port SAS HBA thats PCIE x1 (if it exists) or a 16 port SAS HBA on PCIE x4 or x8. Gen 3 PCIE x1 seems to have a max throughput of 1GB/s which should? be fine for 2-3 drives at full R/W speeds. I doubt there would be a quality chipset or if this even exists. Any suggestions are welcome. I'm not sure that running through the 3 SATA ports on the motherboard + 8 ports on the current expansion card is a good config.

For those who come after me with this question, the answer is Docker: https://hub.docker.com/r/ollama/ollama

I'm interested in using ollama UI on my unraid box. I picked up a 3060 12gb and plan on installing ollama either as a docker container or as a VM. I'm not sure which is the best path, to be honest.

I'm having a QUIC issue where it constantly goes to misconfigured. I've checked the network gear and port forwarding is correct. Sometimes when I restart the Storj docker it works for minutes or hours then it goes out again. I'm not sure exactly what's happening but when I look in the docker tab I notice the Port Mappings (App to Host) "unknown IP:28967 <-> local ip:28967" I assume the unknown IP should be my WAN IP?

I've filled the first disk. How can I add a second disk to this node? What config type?

Yeah - I missed that part. I deleted the preclear log, formatted, and now it's showing up. Thanks.

should the disk (sdl) show up in mnt/disks/? Right now I've precleared the disk (again) and it seems to be mounted but nothing shows under mnt/disks/

I have not - I'll give that a try. Thanks

yeah I've done that.