add support for SIGUSR1 to trigger flush (#7366)

docs/OUTPUTS.md

@@ -94,6 +94,19 @@ You should also add the following to your `SampleConfig()`:
   data_format = "influx"
 ```
 
+## Flushing Metrics to Outputs
+
+Metrics are flushed to outputs when any of the following events happen:
+- `flush_interval + rand(flush_jitter)` has elapsed since start or the last flush interval
+- At least `metric_batch_size` count of metrics are waiting in the buffer
+- The telegraf process has received a SIGUSR1 signal
+
+Note that if the flush takes longer than the `agent.interval` to write the metrics
+to the output, you'll see a message saying the output `did not complete within its 
+flush interval`. This may mean your output is not keeping up with the flow of metrics, 
+and you may want to look into enabling compression, reducing the size of your metrics, 
+or investigate other reasons why the writes might be taking longer than expected.
+
 [file]: https://github.com/influxdata/telegraf/tree/master/plugins/inputs/file
 [output data formats]: https://github.com/influxdata/telegraf/blob/master/docs/DATA_FORMATS_OUTPUT.md
 [SampleConfig]: https://github.com/influxdata/telegraf/wiki/SampleConfig