Pretty obvious you're missing files needed to run the miner.  Says so right in your kernel log, "bmminer not found"
What bsp says is irrelevant since you're not fully booting up.

That's what they do, yes.  Perfectly normal, Mine ON! 

Simple supply and demand.  Supply is limited, demand is growing exponentially as more and more people want to get in on the action!

Your ramp down will be about the same as the ramp up (varies with total pool hashrate), so until the full time has elapsed with your miners off you'll still have shares in the 5Nd range that will be paid when a block is hit.  When, not if!!
Mine on (when you can)!  

What psu are you using?

Yes, and be sure to disconnect the I/O ribbon cable from the boards not powered.
Mine on! 

Ah, so you definitely have open ports.  Shut 'em down and this problem should go away!
Mine On! 

Changing the password in the GUI will not solve this problem, the API login is hard-coded in the firmware and they'll just keep logging in and hijack again - probably a bot anyway.  Check for open ports on your router that are allowing incoming connections and close them down.
Or get a new router, they're cheap compared to mining for someone else!

You behind a firewall?  If not, you could be getting your miner hijacked via the API that is left WIDE OPEN for anyone to come in and change things... 

How long did you let it run? Newest version of machines and firmware are reported to take 45 minutes or longer before finishing the initial boot process and normal running parameters are set. Sounds excessive, yes, but that's the way BiteMe, I mean ShitMain, oh crap, Bitmain forked up the miner software on their machines.

Well hitting a block is based on shares, and more precisely one that solves the block, but it is only human to want to equate that to time. The actual percentage of block diff submitted to the pool is what's important, and that will increment differently versus time as the pool's hashrate varies. For instance if the pool's hashrate stays constant than there's a direct linear relationship between time and % of block diff submitted (until the next diff change however).  If the pool's hashrate varies, as it inevitably does, then time becomes less of a direct measure and actual shares submitted to the pool are the more accurate measure.  
Mine On!  

^^^ What he said!  And (safely) Mine On! 

BBQ! I'd get a fishtape and some copper and repull!

In the USA no customs until you hit $2500 I believe.  Members have ordered miners 1 at a time and mostly kept under the radar, though one had FedEx add together two shipments that arrived separately about a week apart and still charged him! 

Check for the dreaded hidden 'space' at the end of either your username or pool URL.  That got me once.  Just once! 

yes, just disconnect all the power plugs AND the I/O ribbon on one hashboard and power up.

Unfortunately it makes perfect sense.  There's Duty on incoming computer equipment.

Yeah, and with numbers like that you start to think "hack" as in an inside job...

That's what we're trying to help you figure out.  Have you disconnected the power and I/O to one of the boards and run like that?  It really sounds like your PSU isn't supplying enough power for all 3 hashboards, run it with only 2 to help find out.

Well it's obviously wrong so it must be getting confused with noise coming from the PSU's when measuring a miner versus a straight resistive load from the heater.
That's weird!  If you have a O'scope handy it'd be interesting to look at the power going to the meter when measuring a miner and see what's going on.