itimpi

That is sensible if you are happy for the files to end up on the main array. And thus protected against drive failure. Some people prefer that they never go to the array to avoid spinning up array drives unnecessarily and thus use different settings. It is a personal preference so there is no ‘right’ answer - it all depends on the behavior you want to achieve.

The problem will be that you have explicitly named /mnt/cache in the ftp settings, but the Use Cache setting for the share is set to “Yes” which tells Mover to move it to the array when it runs. If you do not want Mover to touch it set it to Use Cache=Only (or Prefer if you want Mover to move files back from the array to the cache). An alternative would be to set up ftp to use /mnt/user instead of /mnt/cache as that would find files regardless of whether they were on the cache or array.

Do you mean multiple arrays or multiple cache pools - they are not the same thing.

Do the following: Take a screenshot of your current disk assignments as a Trial license will not let you transfer the disk configuration to a new USB disk (you can with a non-Trial version). Use the USB Creator tool to write a new USB stick as if you were starting from scratch. Copy across the config folder contents from the old USB to the new USB stick overwriting existing contents of that folder. Do not copy the license key file or the super.dat file that contains the old assignments. The config folder contains all the settings and configuration information you have previously set up. Apply for a new Trial license linked to the new USB stick as the license file is stick specific. Assign the disks exactly as you had them previously and start the array. Make sure you do not accidentally assign a data disk as a parity disk as that would destroy its contents. UnRAID recognises disks it has previously used and leaves their data intact so you should now be up running exactly as you were on the old USB stick. BTW: it is a good idea after doing all this to click on the flash device on the Main tab and use the option to make a backup of the Flash contents. This can then be used in the future if you ever need to create a new usb stick for any reason. It is a good idea (although not mandatory) to do this after any major configuration change.

It sounds as I the USB stick may be on the verge of failing as going read-Only is a typical failure mode. If you decide to replace it go for a USB2 version instead of a USB3 one as they tend to be more reliable. You do not need anything larger than a 8TB version although a higher capacity version may not cost more. As it runs from RAM Unraid gets negligible performance benefit from USB3. i also note that you have the Parity Check Tuning plugin installed with both the option to pause/resume if disks overheat set and also debug logging set. This means that you will get a message in the syslog every 5 minutes as a monitor task runs. If you are worried about these then turn of the debug logging option.

You create it yourself it is called User Scripts for a reason As an example I have one that contains: mkdir /mnt/disks/DJW_NVME logger -tgo "Created /mnt/disks/DJW_NVME for VM storage" mount /dev/disk/by-id/nvme-INTEL_SSDPEKNW020T8_BTNH90420CG32P0C-part1 /mnt/disks/DJW_NVME to mount a SSD I use for VMs. Note the use of the by-id way of identifying the drive so that it still works if the /dev/sdX type id changes for any reason.

Turn off the Automount in UD. the scripts will be ones run via the User Scripts plugin (which is the first thing I normally install from the Apps tab).

The easiest way to achieve this is to not have the mounting handled by UD but have do it via a User Script set to run on array start. For completeness you may want a complementary script to handle unmounting on array stopping.

Anything on the host that is not under /mnt or /boot is automatically in RAM.

I do not think we know this to be the case ! Limetech could be trying feverishly to identify the cause and therefore how to fix it and we would not necessarily know about it.

The file integrity plugin will allow you to pinpoint exactly which file has the problem. It also allows you to do it quickly without doing binary compares of files. Note that corruption can occur for a wide variety of reasons. This is different to ‘bitrot’ which is a special case where a file reads data back incorrectly but does not realise the read is bad.

It is easy to detect that something is wrong. What is NOT possible with parity is detecting which drive is the problem one (and thus which one to change to correct the error).

UnRAID is simply reporting what the BIOS is telling it. There is no limitation of that sort in Unraid itself.

@testdasiIn your list of options you left out 1a) USB 2 external ports? They always seem to work better than any type of USB3 ports. You also did not mention that USB3 drives seem to have a much higher failure rate. My suspicion is that they tend to run hot which is not good if they are going to run 24x7.

I think the No and Only options should specifically state that mover will take to action. It can be inferred from the current wording but could be more explicit. in the "No" case it should perhaps be emphasised that if files that logically belong to a share end up on the cache (which can happen with downloader type apps) they will not be moved to the array

Not sure of the impact on performance. You can use the "Disable Security Mitigations" plugin to turn them off.

Yes - you need Use Cache = Prefer. The Setting affects: where new files get created what action (if any) mover should take The GUI built-in Help makes it clear how each of the settings affect the above two points. They are not necessarily what many new users expect from the text of the possible values.

The correct forum will be the one you get by clicking on the containers icon in the Dockers tab to get the context sensitive menu and selecting the Support option.

Not necessarily if you are simply going to use it for a purpose that does not involve User Shares

Did you make sure the share was set Use Cache=Prefer? You might want to enable mover logging to see if that give any more information.

I seem to get a hash key mismatch on every file which is a hard link (at the User Share level) to another file in the same User Share. Is this expected behaviour? Is there anything I can do to avoid getting this error? The links are in a sub-folder of a user share - do I have to use the Custom folder option to exclude the sub-folders containing these hard links. If so can I have a folder name of the form FILMS/!3D.

I wonder if there should be a similar warning displayed on the shares page if you set Use Cache to anything other than “No”and the cache option is not enabled under Settings->Global Share Settings?

Any time you write to any array disk both parity drives will also be written to, and as you say with your configuration that is half your drives. With so few drives I would not think having two parity drives is cost effective. Unless you plan to add more drives later you might be better off using the parity2 drive as either an additional data disk, or as a disk to be used for backups.

Did you make sure that the docker and VM services were stopped (rather than simply no containers or VMs running)? If the services were running they would have files open that mover will therefore not move.

Turning on the help in the GUI should make it clear why this happened. Basically 'Only' was the wrong setting to use - you need to use 'Prefer' to get the behaviour you want.