Hi all,
I’m quite new to the fantastic Mikrotik product family.
Now have to setup this Scenario running RSTP:
Any hints or examples for the best practice RSTP setup are very welcome, the different setup guides are quite confusing being new to this product and topic.
Thank you guys! 
Not much to take care of. One major thing is to set priority on bridge of switch you want to use as “master” in STP hiearchy to value, lower than default (which is 0x8000) … 0x2000 would be a safe value. This way you won’t see topology changes if some switch changes its MAC address (if priorities are the same on all switches, then lowest MAC value wins the master election). And even more importantly: active/backup link state is determined in relation to master switch … and you want to have this in optimal way according to actual topology (and not according to switch MAC addresses).
Another thing is to use port-cost-mode=long … in principle one should use same port-cost-mode on all devices in same STP area (and some older? devices don’t support the “long” mode). The mode “long” became new default since a recent v7 version … because the previously default “short” doesn’t properly support link speeds faster than 1Gbps. This setting is not available in ROS v6 (the only mode is “short”), so make sure all of your switches run v7 stable (7.15.something right now).
As to which link is active and which is backup: bridge ports have path costs … and by default these are related to link speed (and can, thus, be dynamic if link speed changes). STP algorithms then use working links with least path cost towards master as active, other links are used as backup. You can affect this by setting port cost manually … the higher value, the lower probability for that link to be active (if other candidates for being active are working).
Which all means that you want to make core switch the most likely one to become master (i.e. set priority to lowest value there) … the “ring” link between core switch 2 and core switch 3 will be considered as backup (since it’s further away from core switch 1 than the other two links. Ditto for connections from device switches … since they are directly connected to core switch 1, that link will become active (and the link, passing via core switch 2, will be backup due to higher cumulative path cost).