Hello.
I'm trying to make a PCP field to propagate through CRS-125-24G without success.
PCP is set with bridge rule on CCR, then travels with VLAN to the CRS. I use mirroring to sniff out packets.
I can see the PCP field set on packets entering the switch on port 1, then they leave with the field unset (0) on the other.
I have set egress vlan translation on out port with pcp propagation set.
I have set egress PCP propagation for exit port
I have set ingress PCP propagation for input port.
VLANs work as expected, just the missing pcp field is the problem.
ROS v 6.29.1

I don't think that any bridging is an acceptabble solution. It's a switch, we should be switching.

config as follows (cut down to most important parts) :
port 1 is the trunk where tagged (with prio set) traffic comes in from router
on port 21 there is the AP that i want to receive tagged and prioritized traffic

/interface ethernet
set [ find default-name=ether1 ] name=ether1-master-downlink
set [ find default-name=ether2 ] master-port=ether1-master-downlink name=ether2-slave-uplink1
.. to all ports

/interface ethernet switch vlan
add ports="ether1-master-downlink,ether2-slave-uplink1,ether3-slave-uplink2,ether5-slave-srv1,ether6-slave-srv2,ether7-slave-wicedyrektor,ether8-slave-ksiegow\
osc,ether9-slave-qnap,ether10-slave-pracownicy,ether11-slave-pracownicy,ether12-slave-pracownicy,ether18-slave-AP,ether19-slave-AP,ether21-slave-APs3,swit\
ch1-cpu" vlan-id=10
add ports="ether1-master-downlink,ether3-slave-uplink2,ether4-slave-uplink3,ether13-slave-AP-mgmt,ether18-slave-AP,ether19-slave-AP,ether20-slave-AP,ether21-s\
lave-APs3,ether22-slave-APs7,ether23-slave-APs10,ether24-slave-APs14,switch1-cpu" vlan-id=50
add ports=ether1-master-downlink,ether14-slave-goscie,ether19-slave-AP,ether21-slave-APs3,switch1-cpu vlan-id=30
add ports=ether1-master-downlink,ether19-slave-AP vlan-id=20

/interface ethernet switch egress-vlan-tag
add tagged-ports=ether1-master-downlink,ether3-slave-uplink2,ether19-slave-AP,ether21-slave-APs3,switch1-cpu vlan-id=10
add tagged-ports=ether1-master-downlink,ether2-slave-uplink1,ether3-slave-uplink2,ether4-slave-uplink3,switch1-cpu vlan-id=50
add tagged-ports=ether1-master-downlink,ether19-slave-AP,ether21-slave-APs3,switch1-cpu vlan-id=30
add tagged-ports=ether1-master-downlink,ether19-slave-AP vlan-id=20

/interface ethernet switch egress-vlan-translation
add customer-vid=50 customer-vlan-format=tagged new-customer-vid=0 pcp-propagation=yes ports=\
ether13-slave-AP-mgmt,ether18-slave-AP,ether19-slave-AP,ether20-slave-AP,ether21-slave-APs3,ether22-slave-APs7,ether23-slave-APs10,ether24-slave-APs14
add customer-vid=30 customer-vlan-format=tagged pcp-propagation=yes ports=ether21-slave-APs3
add customer-vid=10 customer-vlan-format=tagged pcp-propagation=yes ports=ether21-slave-APs3

/interface ethernet switch port
set 0 egress-pcp-propagation=yes pcp-propagation-for-initial-pcp=yes qos-scheme-precedence=\
pcp-based,sa-based,da-based,dscp-based,protocol-based,vlan-based,pcp-based
set 20 egress-mirror-to=mirror0 egress-pcp-propagation=yes ingress-mirror-to=mirror0 pcp-propagation-for-initial-pcp=yes qos-scheme-precedence=\
pcp-based,sa-based,da-based,dscp-based,protocol-based,vlan-based,pcp-based vlan-type=edge-port

/interface ethernet switch vlan
add ports="ether1-master-downlink,ether2-slave-uplink1,ether3-slave-uplink2,ether5-slave-srv1,ether6-slave-srv2,ether7-slave-wicedyrektor,ether8-slave-ksiegow\
osc,ether9-slave-qnap,ether10-slave-pracownicy,ether11-slave-pracownicy,ether12-slave-pracownicy,ether18-slave-AP,ether19-slave-AP,ether21-slave-APs3,swit\
ch1-cpu" vlan-id=10
add ports="ether1-master-downlink,ether3-slave-uplink2,ether4-slave-uplink3,ether13-slave-AP-mgmt,ether18-slave-AP,ether19-slave-AP,ether20-slave-AP,ether21-s\
lave-APs3,ether22-slave-APs7,ether23-slave-APs10,ether24-slave-APs14,switch1-cpu" vlan-id=50
add ports=ether1-master-downlink,ether14-slave-goscie,ether19-slave-AP,ether21-slave-APs3,switch1-cpu vlan-id=30
add ports=ether1-master-downlink,ether19-slave-AP vlan-id=20

maybe setting dscp field and using qos DSCP map???

Hello, thanks for replying.
Your solution sounds more like a dirty workaround. Looking at how qos is convoluted in crs, I would like not to do that
Missing from my first post are :
-A plea for help. Please help.
-Overview of what I wan to accomplish, so that maybe someone can propose a different solution. Here goes :
Critical points are the APs - they're the bottleneck. They are Unifi LR APs. They broadcast a couple of VLANs to different WLANs. Since all the networks work on one radio channel, they compete. Some networks are more important than others, so should be prioritized. Herein steps the PCP.

Can the PCP be set directly on the egress of the interface ?

Hello, thanks for replying.
Your solution sounds more like a dirty workaround. Looking at how qos is convoluted in crs, I would like not to do that
Missing from my first post are :
-A plea for help. Please help.
-Overview of what I wan to accomplish, so that maybe someone can propose a different solution. Here goes :
Critical points are the APs - they're the bottleneck. They are Unifi LR APs. They broadcast a couple of VLANs to different WLANs. Since all the networks work on one radio channel, they compete. Some networks are more important than others, so should be prioritized. Herein steps the PCP.

Can the PCP be set directly on the egress of the interface ?

thats the advantage of layer 3 qos marking (dscp), the marking survives across the network unless explicitly remarking at some point, but layer 2 marking can be lost at any point and its hard to track

if im not wrong unifi suppose to do wireless qos (wmm) based on DSCP field

Hello.
Tkanks Chechito for Your suggestion. You're right about DSCP being honored by Unifi APs and it being transported with the IP header throuought the network. I've used this approach and confirmed the field is set on packets bound to APs. I have yet to witness it's (the prioritization on wifi) effectiveness in the field. In my lab the bandwidth was divided pretty equally when i tested it, but maybe the load was not sufficient, there was no packet loss anywhere.
Anyways, the original issue about PCP not being propagated is still unsolved.