Currently, when writing to cuci_user within a minute of the last write we debounce, and within an hour we debounce 90% of the time (writing 10% of the time).
This makes testing difficult as I cannot always distinguish between a write which failed due to a bug and a write which failed because it was one of the 90%.
My suggestion:
- remove the probabilistic debounce within an hour
- have the one minute debounce turned in to a config variable, which we default to 1 hour
I am assuming that, if we are OK with only 10% of events being recorded within an hour, then we are OK with none of them being recorded.
Open question: should the debounce time for cuci_temp_edit be controlled by the same config variable?