Holy shit.
Apparently things can go VERY wrong with these units.
So yeah, one of my miners became almost completely dead. Web interface responded but never updated, nor did the miner actually ever started.
My issue was corrupted .pyc-files inside the control box. So basically none of the numerous python scripts worked and it appears that the whole thing is run by python... along with busybox, ccminer and some web server.... Great!
This will cause the "Error: ValueError: Bad marshal data" -messages while running the .py -scripts.
Well luckily I did come up with workaround but jeez. This will also help with people who have forgotten/lost/whatever their passwords for the boxes.
I'd imagine this works the same with A5 with similar problem.
For the corrupted .pyc files (web interface responses but never gives out any data & no mining happens):
1) telnet to miner IP at port 8100
2) login: root
3) password: innot1t2
4) rm -rf /lib/python3.4/__pycache__/*.pyc
5) reboot
If you need to adjust your super secure password (works the same as A5); login like before but head to /home/www/conf and edit the 'password' file with 'vi'. Vi can be a pain to use for the first time but basically since it's just one line you need to edit... press 'i' and edit in the password you want (default being 12345678). Press esc. Type ':wq!' and hit enter...
Hope this saves someones day. My evening was ruined since I initally thought that the miner is actually dead...
PS.
No need to worry deleting the .pyc files, they are automatically recompiled back in place once you reboot and the various scripts are automatically run.