trurl

OK. Media is all on the array now, but your cache is still mostly full. Do you have any VMs? I now see what the main culprit actually is though. Your docker image is 80G. 20G should be more than enough, and if it's not you have one or more of your docker applications misconfigured. I see you're currently using only 13G so that isn't unreasonable. Why did you set it to 80G? Have you had problems filling docker image? Making it larger won't fix the problem of filling docker image, it will just make it take longer to fill. Go to Settings - Docker, disable and then delete the docker image from that same page. Change it to 20G and enable to recreate. Then go to the Apps page and use the Previous Apps feature. It will let you reinstall your dockers exactly as they were. Then post new diagnostics.

How is it supposed to "look" at the "current file size"? What does that even mean when the file doesn't exist on Unraid yet?

Are you using a slave Access Mode for the mapping?

Unraid doesn't care about the port or the sdX designation, it keeps track of disks by serial number. The sdX designations can change from one boot to another regardless of port, though it often will not change unless you add or remove disks.

That still leaves out some details. What are you drag and drop from? Windows File Explorer? Another OS instead of Windows? Another application besides File Explorer? We have users that could "drag and drop" using any of these possibilities. You must have changed something at some point, maybe before those diagnostics, since High-water is default and to get a share set to Most Free you must have set it that way.

Actually that post spells it out a little better than the one I gave: Click on the icon for your Plex on the Dashboard or Docker page and select Edit. That will take you to the Edit Container page (same as the Add Container page you used to setup plex docker). Beside your mapping for /mnt:/data is an Edit button. Click that button and change Access Mode to RW/Slave.

Your cache is too small to use for much except appdata and other docker/VM related shares (domains, system). Stop writing to your server until you get everything else moved from cache. domains and system are all on cache where they belong so they are looking good. Don't change anything about them. appdata is cache-prefer, which is how it should normally be, except it has apparently overflowed and has files all over the array now. Set appdata to cache-only for now so it will be ignored by mover until we get other things moved off cache. You have a share anonymized as M---a, probably Media, which I'm sure is the main culprit. You currently have that set to cache-no, which mover ignores, and it has files on cache. Probably you set it to cache-no trying to fix your problem. Set that share to cache-yes. Then run Mover (Array Operation - Move Now). Wait for it to complete then post new diagnostics so we can check your progress. There are going to be some additional steps required to get things as they should be.

I hope you are planning to recruit moderators from each of the supported languages.

You have a lot of shares, all with no Minimum Free, and a lot of mostly full disks. This is likely to cause issues with overfilling disks. The general recommendation is to set Minimum Free for a share to larger than the largest file you expect to write to the share. Unraid has no way to know how large a file will become when it chooses a disk to write. If a disk has less than Minimum, it will choose another. Cache also has a Minimum setting in Global Share Settings. Other than that, it isn't clear what the specific issue is. You might try deleting and recreating docker image in Settings - Docker. Then the Previous Apps feature on the Apps page will reinstall your dockers just as they were.

Splitting works fine, it is merging that bites us in the butt.

Here is a FAQ with more details on the nuances of the Use cache setting: https://forums.unraid.net/topic/46802-faq-for-unraid-v6/?do=findComment&comment=537383

Virtualizing Unraid is not supported, but we do have a subforum for users who attempt to virtualize Unraid can support each other. https://forums.unraid.net/forum/46-virtualizing-unraid/ Do you intend to run Unraid virtually, or are you just trying it out that way?

Yes. Mover ignores cache-no and cache-only shares. It moves cache-yes shares from cache to array, and cache-prefer shares from array to cache. Cache-yes/prefer shares can overflow to the array if cache goes below its Minimum Free setting in Global Share Settings. Also, mover can't move open files, and it won't move duplicates. appdata, domains, and system shares are cache-prefer by default, so your dockers/VMs performance won't be impacted by the slower array, and so they won't keep array disks spinning, since these files are usually open files. Those shares all look correct according to the diagnostics you posted assuming you haven't changed them. That is good because it can take a few steps to get them moved to cache since their files are open. If you want to move other shares from cache to array you must set them to cache-yes. Then if you don't want new files for those shares written to cache, you can set them to cache-no after all the files have been moved.

Go to Tools - Diagnostics and attach the complete Diagnostics ZIP file to your NEXT post. Sounds like maybe your issues are larger than just this container, so I may split to its own thread.

Go to Tools - Diagnostics and attach the complete Diagnostics ZIP file to your NEXT post.

That is not the recommended solution, and we still don't know what was going on. Prefer is the correct setting for those shares because you want them to stay on cache for better performance of Dockers/VMs and so those won't keep array disks spinning. And Prefer will only move from array to cache and after it has finished moving all it can it won't do anything else. And next time mover runs it won't move those again since they are already where they belong. Setting these to cache yes really doesn't do anything to fix anything, it will just put those on the array where your Docker/VM performance will be affected by the slower parity array, and they will keep array disks spinning. It is true that after mover has finished moving it won't move those again, but that isn't any different than the behavior with Prefer, it is just a different (and not ideal) final destination for the shares. Possibly you were using those user shares for some other purpose than intended, such as using appdata for downloads or something. It would be better to try to figure out the real problem (and the real solution) and so better understand how to avoid causing these problems in the future.

Unraid is a stripped down, customized version of slackware linux, and doesn't contain everything required to build drivers or packages. And installing drivers yourself even if you had one already built has problems of its own, since the OS is unpacked fresh from the archives on flash into RAM at each boot, and the OS runs completely in RAM (think of it as firmware). So no changes you make to the OS survive a reboot. Are you sure that NIC isn't already supported? Intel NICs are typically on the recommended list.

Your question seems to be related to Split Level. Split Level overrides Minimum Free so if your Split Level is trying to keep those TV Shows together then it will keep writing to the same disk despite Minimum Free. So, you can set a large Minimum to get the whole TV Show on the same disk. Once the disk goes below Minimum the TV Show will keep going to that disk according to Split Level, but other things in the share will go to other disks. Note that Split Level can make you overfill a disk if everything in the split will not fit.

You definitely should upgrade Unraid. Not guaranteed any current plugin or docker will work with that old version, and it is very difficult to support since nobody that might help is running it.

I noticed in your diagnostics you had some remote shares mounted with Unassigned Devices. Does plex use these? Your only mapping that would reach these (as well as the rest of your disks and user shares local to the Unraid server) is /mnt:/data But, Unassigned Devices need to be mapped with the slave option, which you don't have in your /mnt mapping.

This suggests it has some problem writing to flash when shutting down. Here is how unclean shutdowns are detected:

The screenshot often doesn't show some important details. Post your docker run command as explained in this very first link in the Docker FAQ:

No. You can limit a user share to a specific disk so when that disk runs out of space that would be the limit.

Except when the link gets broken (my bad) and the plugin isn't updated for the new link. The UD support thread is pinned near the top of the Plugin Support subforum.