FLP reports a lot of "RecordKeyMissing" errors in its metrics (this can be viewed in the health dashboard). More specifically, the "Bytes" field is missing when it is 0 (I'm not sure why it can be 0 in the first place, but that's a different thing)

The error metric shows this is happening like 6K times/minute on a default netobserv deployment (ocp 4.15, 6-nodes cluster)

1. Deploy latest netobserv from main, enabling Packet Drops feature
2. Check the health dashboard for FLP errors
3.

lot of "RecordKeyMissing" errors

no error