size of buffer used by firehose to store events.
Maximum number of events to send to Druid in one HTTP request.
Maximum number of events to send to Druid in one HTTP request. Larger batches will be broken up.
Druid indexing tasks will shut down this long after the windowPeriod has elapsed.
Druid indexing tasks will shut down this long after the windowPeriod has elapsed. The purpose of this extra delay is to allow time to receive the last few events that are valid from our perspective. Otherwise, we could think an event is just barely on-time, but the index task may not be available to receive it.
When we create new Druid indexing tasks, wait this long for the task to appear before complaining that it cannot be found.
If a push to Druid fails for some apparently-transient reason, retry for this long before complaining that the events could not be pushed.
If an indexing service overlord call fails for some apparently-transient reason, retry for this long before giving up.
True if we should add a random suffix to Druid task IDs.
True if we should add a random suffix to Druid task IDs. This is useful for testing, since it allows us to re-submit tasks that would otherwise conflict with each other. But for the same reason, it's risky in production, since it allows us to re-submit tasks that conflict with each other.