hubble: Introduce metric for lost events #15112
Labels
kind/feature
This introduces new functionality.
pinned
These issues are not marked stale by our issue bot.
sig/hubble
Impacts hubble server or relay
Hubble currently emits
LostEvents
whenever it detects that an internal buffer overflowed. These are currently stored in the ring buffer and sent down downstream via theGetFlows
API, but this has not proven to be very useful:It is probably more useful to expose this information as Prometheus metrics and rate limitted warning messages. If they are exposed as part of the API, the message sent to the client should be a single aggregated statistic of how many events were lost during the request, rather than sending individual messages.
The text was updated successfully, but these errors were encountered: