RX Drops since ROS 7

Since ROS 7, near as I can remember, we are seeing constant counts of RX Drops on ports that have any traffic at all. The more traffic the more drops. These seem to be real because they affect traffic. We see these on all ten RB4011s in the network. Most are 7.12 but I upgraded a couple to 7.15 and the problem persisted. There is nothing in the logs.
Layer two network. The CPU usage is below 5%. Memory and disk usage is also very low. The max traffic we see, on 1GB or 100MB interfaces, full dup, is about 25Mb. Even on one that is receiving less than 2Mb of traffic, RX Drop counts increase with every counter update. We also see RX Code Errors and RX FCS errors counting up, but not as fast as drops.

We did not see these drops with ROS6. The network was very stable.

Is there something that needs adjusted when you go to ROS 7? Buffer sizes?
Any suggestions appreciated.

Jim

Update, looking at every interface it appears those connected to another router or to a SAF microwave are stable. It’s the ones connected to a Ubiquity AF24 microwave at 1Gb that are counting the RX Drops and other errors. Setting these interfaces to 1Gb, 100Mb, auto or fixed does not stop the errors.
Jim