CCR2004-16G-2S+ Unstable Upload/Download After ~20 Days (Firewall & QoS Insights)

I’m running a CCR2004-16G-2S+ on RouterOS 7.18.2. After about 20 days of uptime, WAN performance degrades—upload/download become unstable, and only a router reboot temporarily fixes it.

/queue tree
add name=ACKQueue  packet-mark=ACKTraffic    parent=ether16 priority=1
add limit-at=50M   max-limit=250M          name=DNSQueue   packet-mark=DNSTraffic parent=ether16 priority=2
add limit-at=700M  max-limit=1G            name=HTTPQueue  packet-mark=HTTPTraffic parent=ether16 priority=3 queue=pcq-download-default
add limit-at=500M  max-limit=1G            name=BulkQueue  packet-mark=BulkTraffic parent=ether16 queue=pcq-download-default

/ip firewall filter
add action=drop   chain=input   connection-state=invalid comment="Drop invalid"
/ip firewall filter
add action=drop   chain=forward connection-state=invalid comment="Drop invalid"
/ip firewall filter
add action=drop   chain=forward connection-limit=64,32 connection-state=new in-interface=ether16 protocol=tcp comment="Limit new WAN TCP"
/ip firewall filter
add action=accept chain=forward connection-state=new in-interface=ether16 protocol=tcp comment="Allow new WAN TCP"

/ip firewall mangle
add action=mark-packet chain=prerouting comment="Mark ACK"    new-packet-mark=ACKTraffic packet-size=0-123 passthrough=no protocol=tcp tcp-flags=ack
add action=mark-packet chain=prerouting comment="Mark DNS"    new-packet-mark=DNSTraffic dst-port=53 passthrough=no protocol=udp
add action=mark-packet chain=prerouting comment="Mark HTTPS"  new-packet-mark=HTTPTraffic dst-port=80,443 passthrough=no protocol=tcp
add action=mark-packet chain=prerouting comment="Mark Bulk"   new-packet-mark=BulkTraffic passthrough=no

Has anyone seen Queue Tree performance degrade after long uptimes? I have tried with and without fasttrack enabled, both scenarios fail.
Is there a way to “refresh”/reset QoS without a full reboot (scheduler script, connection flush)?
What should I monitor or clear?