JonathanM

The docker system is designed to economize where possible, if multiple containers use the same layers, it keeps track and shares them. Those pieces are inside the docker image file.

This is one of a few reasons why USB is not recommended for parity array members. Unraid drops a disk when a write to it fails, so chances are high what is physically on the drive doesn't match what is emulated by parity. Normally I'd recommend simply rebuilding to the same drive, but without diagnostics, there is no way to know if that's the best course of action right now. Attach diagnostics zip file to your next post in this thread if you want better advice.

I recommend looking up spaceinvader one's videos on youtube, he goes into VM's and many other things with Unraid that are important to understand.

Yes, but Unraid uses the flash for licensing and configuration changes. It must be present at all times in the server. The Unraid USB is not an installation that installs the OS onto internal media, it IS the OS, and it loads into RAM to run each time the server is rebooted.

That really sounds like the PSU voltage is sagging, could be either lack of capacity or too many splitters / faulty splitters / poor connections. I personally attempt to power 4 or fewer drives from each set of wires coming from the PSU. Also, is the HBA getting good airflow across its heatsink? The designers assume servers will have forced air across all card slots, but most desktop cases rely on the cards to have their own fans if needed.

Exactly. Normally PC fans exhaust on the side with the motor mounting spokes, the fan in the top right of this picture appears to be blowing out.

DON'T go wild and install everything in the nerdpack. Have a specific reason and use for anything you select. Some tools create issues just by being installed. That's one reason why they aren't just all installed in the normal version of Unraid.

Link from the page you posted.

All offline licensing will need to be handled by emailing Limetech.

https://linuxize.com/post/how-to-use-linux-screen/

If you start screen before you start mc you can log back in and resume the screen session and pick back up exactly where you left off.

Depends on whether or not you started a screen session before you ran mc.

Unfortunately the author appears to be mia. @Squid, maybe this plugin should have a note added?

I don't know if the situation has changed, but years ago the secret sauce was never fully understood. It was thought that a weighted average of changes over time, so drastic changes from one boot to the next would trigger, but those same changes done over weeks would not. Now with the advent of ms logins, activation seems tied there as well. https://support.microsoft.com/en-us/windows/activate-windows-c39005d4-95ee-b91e-b399-2820fda32227

I could be wrong but it looks to me that some of your fans are blowing the wrong way.

If the container is referencing or writing to any files that have already been moved to the array, the array will stay spun up. Cache: yes means NEW files written to the share will be on the cache, then when mover runs they will be transferred to the array if they aren't being held open by something. Cache: prefer will attempt to move the entire contents of the share to the cache drive, assuming there is enough free space. Once again, any files that are open will fail to move.

Did you finish the balance operation to remove the drive? Until the BTRFS filesystem has purged the drive from the pool, it will remain linked no matter where you assign it or don't, as long as it's physically connected and recognized. If you don't mind losing the data on both devices you can run blkdiscard /dev/??? on both and they will no longer have the BTRFS pool data.

Depends on your hardware. If everything is compatible and working properly, stopping the array should be enough. However, it's much safer to power down, and it doesn't really take that much more time. Your call, but I'd power down, even if I was sure my hardware could handle a hot swap.

When was the last time you replaced the key? You are allowed 1 automatic transfer per year, otherwise you need to contact Limetech directly.

root has never been allowed network share access. It's only allowed for the webGUI or SSH console.

Reverse your logic and it's pretty easy. echo "_waiting 5 minutes" sleep 300 echo "_starting Container X" /usr/bin/docker start (container X) echo "_starting Container Y" /usr/bin/docker start (container Y) echo "_starting Container Z" /usr/bin/docker start (container Z) echo "_modifying hosts" ?command to add hosts entry? Disable auto start for the containers, set the script to start at array start, it waits 5 minutes, starts the container(s), then does your hosts thing. Sorry, no help there. Why does it have to be limited to those containers, and why only after start? Couldn't you just do the magic in your router, or in Unraid itself?

The logical reason is that Unraid doesn't run from a disk or any permanent media. The OS stays totally in RAM. Any additions to the OS must be carefully weighed against the penalties of requiring extra RAM for a feature very few people would use, and that have viable alternatives.

When dedicating cores to VM's, be cautious of limiting the hosts performance too much. Keep in mind that the "motherboard" your VM is running on is a software construct running on the host. If you slow down your VM's "motherboard", you will cripple the VM the same as if you didn't give it enough RAM. It's a better idea to approach resource allocation by giving the VM only enough to do the job properly, when your benchmarks in the VM cease to improve with added resources, back off 1 tick and leave it. Every resource you can leave for the host will go towards keeping the back end processes smooth for the VM. That is especially true of RAM, when you allocate it to the VM, the host can't use it for caching to speed up the VM's I/O. I see many people complain that they gave the VM every last bit of CPU and RAM they could, and it STILL hesitates and stutters. Many times that is because the host has run out of resources to feed the VM what it needs.

Which filesystem? Diagnostics might shed some light.