Is a queue tree work without “limit at” and “max limit” speeds?
Is there any measurement script, to dinamicly set theese values, or an other mikrotik router in uplink path can signal to the main router with queue tree, to start queuing packages because the uplink router reached its uplink line capacity?
There are a main router with queue tree, and a load balancer router with two non equal WAN uplink line. I want to fine tune the QoS on the main router to max out the ISP lines capability.
No, it will not work.
Priority is ONLY used to decide what to drop when limits are reached ( NO!, it does NOT change packet order, it does NOT put higher priority packets first) - if limits are not reached, nothing will be dropped
Only theese limits are for dropping packets, no other method to the uplink router signal to the main router’s queue tree to start dropping packets?
Only if you base it on some packet header fields like DSCP or TTL
Could you show me an example?
This forum is like a ghost city 
The soliution: On main router I marked the wan targeted packets with dscp value, on the load balancer I made queue trees to each uplink, and made postrouting mangle rules, to mark outgoing packets with packet mark by their dscp value.
Only thing I dont know. I made packet marks for a priority to each ISP destination (if a packet targeted to ISP1 then it marked to prio_x_ISP1, if ISP2 then prio_x_ISP2). Is it enough one packet mark rule and label by priority, and the queue tree knows, the packet is targeted to the root interface of the tree?
Other question. I worked by this article: https://wiki.mikrotik.com/wiki/DSCP_based_QoS_with_HTB
The DSCP queue tree made tree leafs whith default queues. I modified the tree for the futureproof.
http://www.patrickdenis.biz/blog/ip-precedence-dscp-tos-lookup-table/
By the higher drop probality I used smaller queue variants of the default. For low drop priority I used queue size 10, for medium queue size 7, for high queue size 5. Is it a right strategy? Should I lower the bucket sizes too?