Pandemic

I'm not quite sure how to do the license transfer.

Thanks for that, I'm going to try to update now. It makes sense that the nvidia driver needs to load sooner than the cache drives. When copying to the new drive, could I just use a backup? Plex still isn't starting. I'm not sure that's related.

I expected the nvidia driver would live in the cache drive. The 2gb USB has been getting more and more difficult to use but I need to look into how to swap from my decade old license paired with the drive sn.

arc-diagnostics-20260226-1305.zip

I just updated to unraid 7.2.4 and I'm not able to install the nvidia drivers. It runs, seems to install fine, but it doesn't seem to stick. Essentially, ollama, plex, and a few other containers aren't running. I think it's because the nvidia driver isn't there. -----------------Downloading Nvidia Driver Package v590.48.01------------------ ----------This could take some time, please don't close this window!------------ --------------Can't download Nvidia Driver Package v590.48.01----------------- Removing package: nvidia-driver-2026.01.16 plugin: run failed: '/bin/bash' returned 1 Executing hook script: CA_postHook Clearing Community Applications plugin cache Executing hook script: post_plugin_checks Plugin installed

I found a 6.1.7 zip at 50mb or so and removed it. I'm not sure that I can clean anything else out: boot device shows 416030720 free but upgrade needs 828727261 Yeah, 7.1.4 is 790mb. I'll have to follow manual to install from a different device.

I'm sorry, boot device shows 358719488 free but needs 828727261 .... that's way less than what the upgrade needs. Would a manual upgrade somehow avoid this?

plugin: installing: unRAIDServer.plg Executing hook script: pre_plugin_checks plugin: downloading: unRAIDServer.plg ... done plugin: downloading: unRAIDServer-7.1.4-x86_64.zip ... done plugin: downloading: unRAIDServer-7.1.4-x86_64.md5 ... done boot device shows 358719488 free but upgrade needs 828727261 plugin: run failed: '/bin/bash' returned 1 Executing hook script: post_plugin_checksI'm using an old original 2gb flash drive but it also shows more free space than what the upgrade needs.

I'm now running 2x1tb nvme drives in a cache pool. I've got the docker img file on cche but I can't seem to add appdata on cache. What's the best method here?

Interesting, thanks for that information. I'll try to look into the best method of repairing the flash drive. Where are the FSCK files? Do you have any recommendations on repairing? When I got Unraid (like 15 years ago now) I got two usb drives. The second one is largely unused so I could always swap them out. I also see that these drives are "Ineligible for self-replacement".

I just tried the box on a different browser - worked. Cleared box's cache in firefox - seems to be working now. I'd be curious to know if anyone recognizes any issues in the diagnostics though

I've been having some odd issues with my unraid box so I updated to 6.12.15. After the update I get: 400 Bad Request Request Header Or Cookie Too Large I was getting the same thing when I opened a share called 'appdata' that I don't remember setting up. The other day I had an issue where SSH was disabled and the machine name was changed to tower. That's about the third time it's happened in 6 months. This time, my 512gb m.2 ssd cache drive just disappeared and won't come back up after reboots. No idea what's going on with this box. tower-diagnostics-20250121-1422.zip

Right, so something is screwy with unraid. I'm going to try to take it apart and redo a bunch of settings.

I tried backing up the appdata and none of it seems to be taking effect. I thought the docker.img was essentially temp data and the appdata was persistent... idk, I'm having to set up a new plex server from scratch and this is the second or third time in a matter of six months. Irritating.

ughhh, lol. Just brutal. I have to set everything back up. It's like the docker-xfs.img file just disappeared but the space was still reserved.

Managed to format the 2tb nvme cache drive to XFS so at least it's showing up now.

I'm starting to lose my mind on this one. I've attached the diagnostics. My old cache was a 512gb nvme and it was working just fine. As I said in the original post, one morning docker had failed. I know that the nvme heatsync had a bit of horizontal pressue on it from a neighboring card so I figured it was a connection issue. I've now got it in the slot without any cooler. Zero pressure. I've had it boot and recognize the 512 a few times but it always drops. SMART shows no issue (fast test). I'm now trying to assign a 2tb nvme to the cache but I'm getting "unmountable: unsupported or no file system." The file system shows as XFS... tower-diagnostics-20250121-1422.zip

I'm working through an issue with Docker again. This morning I checked in and docker was not running. I had a bit of an issue with one nvme slot being a bit too close to a PCIE slot and the NVME heatsink was touching the heatsink of a PCIE. I figure it was causing a connection issue even though it's been fine for the past month. Anyway, I'm using a different slot and I took the heatsink off. The cache drive is showing up but the directory (/mnt/cache/docker-xfs.img) has a triagle next to it and says path does not exist. Now, there is a large chunk of the nvme drive that's being used. I expect it's the previous .img file but I don't know how to find it. It's not showing in WinSCP. I think my only option at this point is to format and set it up again. I'm backing up the /user folder to try to keep my persistent data. I wanted to check here in case you have any other options.

I wonder if an unplanned power down could have caused that in the course of testing the PSU. Any idea why 'watch -n sensors' isn't working when it was previously?

Odd. The machine name was changed from it's previous name to 'tower'. I assume this is default naming?

I just finished installing a new PSU and cpu cooler. When I booted it, for some reason SSH went from being enabled to disabled. When I enable it, it seems to constantly want to load from tower.local rather than the IP (which I've always used). The tower.local shows the page larger for some reason so it's not my favorite. After enabling SSH and logging in, I'm not able to run commands I had been running yesterday. watch -n 0.5 nvidia-smi (actually now working) watch -n sensors My first question is how a hardware mod could change ssh settings. I did move the USB boot drive to a different port but I wouldn't think that would be the culprit. Next question is how to fix the issue of not being able to run watch -n sensors. When I run it I get: watch: failed to parse argument: 'sensors': invalid argument tower-diagnostics-20241126-1342.zip

Sorry, what should the destination be? I have it set to mnt/disk7 but maybe i should choose cache?

I removed my cache drive and added a new one. Since then I noticed a new share "ab_20240823_073553" has shown up. It seems to be todays date? Does anyone know where that came from and what it's doing?

For sure but the docker tab showing 12gb was different than both /tmp and /plex_tmp was showing. Is it possible the container was using extra memory not allocated to transcoding?

I used appdata backup (I think I did it right) and then stopped the array, selected no disk for cache, started the array, stopped the array, and then selected the 512gb NVME. I kept the 300gb docker size just so I don't have to redo that aspect. I know it's way more than I need but with some LLMs being 20+ gigs I figured I'd try to future proof. Can I just reduce the size without having to rebuild everything? Funny thing though, as I'm moving data to the UNRAID box, my speeds are about 10-20MB/sec slower than they were with the raptor drive. Everything I'm seeing suggests NVME would have a faster write speed so I'm not sure what's happening. edit: this may be an issue originating from the box containing the files to transfer to the unraid box. That's about the only logical conclusion I can think of.