When writing batches of points to Influx (via the InfluxDB::batchOf mechanism) over UDP it is easy to accidentally exceed the maximum UDP packet size. This causes an exception to be raised in UDP::send which ultimately results in the loss of the entire batch of points.
I propose that InfluxDB::flushBatch should be aware of the largest message that can be transmitted over its Transport and sends the Points in the largest batch possible.
When writing batches of points to Influx (via the
InfluxDB::batchOfmechanism) over UDP it is easy to accidentally exceed the maximum UDP packet size. This causes an exception to be raised inUDP::sendwhich ultimately results in the loss of the entire batch of points.I propose that
InfluxDB::flushBatchshould be aware of the largest message that can be transmitted over itsTransportand sends the Points in the largest batch possible.