Squid

Shouldn't make a difference.  Ensure that disk down down is still set correctly (change it to 30, then reset it back to 15 and hit apply)
 
Failing all else, post your diagnostics

The share system does not support that.  It will support separate shares with varying permissions for the users on them.

A rebuild doesn't care if the drive is empty or if it's full.  It's simply a sector by sector calculation needed to recreate the drive.  
 
The alternative to a rebuild of the drive is to do a New Config and assign the drives appropriately, but in that case the parity needs to be rebuilt to reflect the actual sector contents of the drive in question.
 
IE:  No real way around the rebuild.

Yeah, you've got physical 2 nics, but 3 are listed in the config (must be a ghost from a previous configuration or something)
 
I've only got 1 (and no easy method for adding another without destroying the WAF while she's watching movies), but you can try making a change to Network settings and reverting it and finally applying to fix you up.
 
Worst case you delete those 2 files and reconfigure accordingly.

Settings - Management Settings, use SSL to be either Auto or Yes and get a certificate via the button at the bottom.
 
Browsers automatically append :80 as it's the default if you don't specify a port.  If you've changed it, append the port

Both your shares screenshot and the diagnostics show that the system isn't what the individual shares may say.  (They may be showing that because there is no "cache" drive and that's what it's defaulting to).  Make a change (any change) and revert it then apply and the continue with my directions.
 
And yes by disable all services I mean to disable the docker service, disable the VM service in settings, not just stop the individual apps.  appdata for sure contains files on the array due to how this was working for you and it is going to significantly affect your performance.

Nothing to worry about.  Related to a current issue under active investigation / resolution by the MyServers team

Anything outside of the array is a "Cache Pool".  Previously you would accomplish something like what you want via mounting the drive in Unassigned Devices.  But, it's easier to let the OS itself manage it as a cache pool.
 
No physical disconnecting is required.  You're just rearranging how the system thinks the drives are assigned.
 
Your new config allows you to remove the drive from the array (parity will need to be rebuilt).  And you hit add pool to add another cache pool that is specific to the security drive.
 
Writes etc to that drive won't involve parity at all, and you can still access it via the user shares, so no adjustment to any VM's or Docker containers is required.  Just the shares that exist on that drive you set to be "Cache Only" and to use that new cache pool.

It's something with the script that he's got running to set stuff up.  No idea.  Post in the support thread.

sdd (disk 1) seems to have connection problems and is suffering from device resets slowing everything down.  Reseat the cabling

Look in your download client for an option to set permissions (advanced view?)  Then change it to 0777

Can't see where it all started, but a ton of write errors to the cache pool.  But, the one drive (...3042) isn't showing as being online in the SMART reports, so you might want to check cabling etc (or maybe it just died). 
 
@JorgeB will be able to assist in any recovery operation

Does your VM's attached monitor (you're obviously doing a GPU passthrough) normally display the bootup sequence from the BIOS of the server itself?
 
Since you had to go into the BIOS to set the date and time, it implies your battery (probably a CR2032) is shot and the primary GPU you're using for the server (I'm assuming you have 2 -> one for passthrough, and the other (iGPU / cheapie card) is used for Unraid.  You've got to set the primary video card to be iGPU / el cheapie.  Then we'll be able to make some more progress

My first thought when I did the text was "FIRE FIRE FIRE BEES!", but decided to settle on what is there now.
 
Surprisingly, the webGUI itself (and the underlying emhttp) required memory, none was available so Linux killed off some Plex on Jan 3 (which was based upon what it needed / was last used was the best candidate).  Looks like you possibly noticed that and restarted the array since
 
Currently, nothing is any unusual amount of memory so its possible that on 1/3 something was just going nuts (without really diving into every file) with download / transcoding to RAM, etc etc.  
 
Suffice it to say that the system operated as intended and killed off what was required for it to stay running as the linux programmers designed.  It doesn't appear to be a common issue for you, so you're probably going to be safe from the bees (or anaphylactic shock)

Unlike "typical" RAID solutions, Unraid does not work any better (or worse) with matching sets of drives, so nothing to worry about 

Quick google search said to avoid the card if possible, but what @itimpi was referring to was the internal power connections coming from the powersupply.  All too often people don't consider that every molex splitter is 
Another point of failure terrible quality to begin with, regardless if you paid $1 or $20 for them And also quite often do not consider the power draw on individual rails coming from the power supply vs the amount of components plugged into them
 
Another possibility depending upon how you're connecting the drives is the SATA connected to the drives.  Reseat.
 
Not saying that any of this is your particular problem.

01:00.0 RAID bus controller [0104]: Broadcom / LSI MegaRAID SAS 2008 [Falcon] [1000:0073] (rev 03) Subsystem: Dell PERC H310 [1028:1f78] Kernel modules: megaraid_sas You've got the controller set up in IR mode.  You still have to tell the controller in IT's bios (CTRL-S IIRC) to either switch to JBOD or create 4x RAID pools composed of a single drive each.
 
Since it's a Perc H310, you'll actually have by far the best results on Unraid (or any NAS) for that matter by cross flashing it to be the generic LSI HBA
 
Lots of information on that in the internet (just search H310 crossflash to IT), and a rather long thread here https://forums.unraid.net/topic/12114-lsi-controller-fw-updates-irit-modes/
 

You've got to reseat the cabling to both ST10000VN0004-1ZD101  ZA29CF7E (disk 10),  and ST10000VN0004-1ZD101   ZA2AJ2XJ (disk 11)
 
Sata is finicky at times and it doesn't take much for continual resets to happen on a slightly misaligned cable (or a bad cable)

Restore the USB from a backup (or manually download the OS version and save all of the bz* files in the zip to the flash drive).  Worse case scenario you would have to set up a new boot device

Actually, that display is correct at the time the diagnostics were taken.  The syslog has a total size of 0 bytes.  Have you by chance deleted /var/log/syslog (or doing anything with it)

Jan 16 07:52:26 Potti root: nginx: [emerg] cannot load certificate "/etc/ssl/certs/unraid_bundle.pem": PEM_read_bio_X509_AUX() failed (SSL: error:0909006C:PEM routines:get_name:no start line:Expecting: TRUSTED CERTIFICATE)  
 I suspect that this is the root of your problems.
 
Delete the contents of /config/ssh and /config/ssl on the flash drive and reboot
 
This could be flash drive corruption, as the system isn't reading the key file correctly (or at all), but that shouldn't affect being able to hit the UI
 

Most people use the Parity Check Tuning plugin so that the scheduled checks run in batches during off hours (ie: mine run between midnight (when I go to sleep) and 4 am (when the wife gets up) and resumes the next day until its completed).  But, yes since there's competing bandwidth depending upon how much bandwidth is required by the media, pauses etc may result

Stop the VM service and docker service in Settings (VM Settings and Docker Settings).  Then
mv /mnt/cache/system/libvirt.img /mnt/cache/system/libvirt.img.new mv /mnt/cache/system/docker.img /mnt/cache/system/docker.img.new Restart the VM service
 
Do the VMs show back up again?  If it does (and it should in theory), then 
rm /mnt/cache/system/libvirt.img.new rm /mnt/cache/system/docker.img.new  
Followed by stopping the each of the services again.  Then change the system share's settings to use Cache: prefer, then Main, run Mover.  After its done, restart each of the services.
 
 
Other comments.  The usual setting for use Cache on the appdata share is prefer, for far better performance (and make a backup via the appdata backup plugin)

MCE was issued during processor initialization and affects some combinations of hardware and is safe to ignore

There's no link being detected on any of your NICs.  You sure the cable is plugged in? (Both ends).  Any lights on the NIC or the switch?