trurl

You really should just take some time to determine what is truly important enough that you can't afford to lose it. Most likely it won't require a lot of capacity. Then just back that up to disks outside the array (those disks in the drawer after you weed out the bad ones). That is probably going to be a better approach than what you are considering.

Most likely just filesystem overhead and no actual files.

Do you have a backup of your flash drive?

You still seem to be missing my point. It is entirely possible that your mappings are correct, but the application within the container is configured to write to a path that isn't mapped. And all those other things you said you did to fix this don't have anything to do with this problem. Is your docker image actually growing when no dockers are running? I assume not. So one of your applications is writing to a path that isn't mapped. Which docker(s) are you actually running when docker image fills? To give another example. Since you mentioned deluge earlier, and I gave an example using Transmission. These are both torrent applications. If you have the deluge container with /Download in the container mapped to /mnt/user/Download on the host, but within the settings of deluge itself, you tell it to write to /download, then it is going to write into the docker image. /Download and /download are not the same thing. And /Download is mapped, but /download isn't. Anything written to a path that isn't mapped is in the docker image.

Something still missing. This thread was not "starred" as one I had posted in. And in fact, nothing I posted during the "outage" shows up in my "Activity", though the posts are in the forum. Looking at the other users on this thread I see their posts on this thread are missing from their Activity also. Brit, for example, has things in his Activity where he reacted to posts in this thread, but his posts in this thread are not part of his Activity. Maybe we will have to live with it, but I have no easy way to find any thread I may have responded to this past day. ***EDIT*** This post itself does show in my Activity, but the previous post in my Activity was from before the problem happened.

Working for me.

The "same issue" referred to is this thread: https://forums.unraid.net/topic/87378-server-fails-to-start-after-update Since the specific details are likely to be very different, and we should let that other thread be dedicated to supporting its OP, I went ahead and split this hijack into its own thread. I will ask some of those same questions from that thread though:

Just for completeness, there is also a share anonymized as t--t. That share is cache-no and has files only on disk1.

I don't see a share named SuperFastTest. There is a share anonymized as F------t. It is set to cache-yes and has files on cache and disk1. Possibly you wrote the test file to that share before setting it to cache-yes, so the file went to disk1, and any attempt to replace it or read it would be to the existing file on disk1. There is another share anonymized as S------------e, could this be the one you mean? It is cache-only and has files only on cache. See if you can clarify your description and perhaps repeat the test taking this information into account.

What is the name of the share?

Hotswap isn't really that useful in Unraid, since you can't replace a disk without stopping the array anyway.

Probably best if you don't attempt to use USB for your Unraid disks. Unreliable connections will be a constant cause of disabled disks and parity failures. Maybe if you wanted to run without parity you could get away with it.

The reason it's not news is because all of that is already explained in the thread for the plugin. So anyone using that plugin should already know that.

I should have split this off immediately since it was a hijack and since it didn't really serve @asopala where it was anyway. I don't think there is anything in that other thread that is really relevant to this one but I link it here anyway: https://forums.unraid.net/topic/87378-server-fails-to-start-after-update So now we can start by asking some of the same questions we asked the OP in that other thread:

Filling docker image is about the settings within an application that causes it to write to a path that isn't mapped. If you use Transmission for example, and within the Transmission application you set it to downloading torrents to /download, but the container mapping is /Download, then that is going to write into the docker image, because Linux is case-sensitive and /download and /Download are different paths. Or, if you tell it to write to download instead of /download, then that is going to be in the image also, since it is not an absolute path. Taking them one at a time as suggested is the way to get this figured out. Which dockers do you use?