Open msaspence opened 8 years ago
This is an interesting one. We've tried to mitigate the impact on the running application to the extent that we either get the metrics, or the metrics get dropped. This could be an interesting-to-have, but it also might require more code than we'd like. There may be more valuable features to squeeze in for a similar byte addition.
This doesn't just give us barometers request time but since we immediately respond with 200 it gives us latency too
If we track barometer's own request times and add it to the next payload we can get some idea of how much impact those requests are likely to have on the host application's requests.
This is implemented in Bucky we can probably lift it out of there.