In addition to reading this guide, run the Elasticsearch Health Check-Up. Detect problems and improve performance by analyzing your shard sizes, threadpools, memory, snapshots, disk watermarks and many more.
Free tool that requires no installation with +1000 users.
What is an Elasticsearch flush?
How are Elasticsearch documents indexed?
As new documents are indexed, the operations are recorded on disk in the translog and stored in memory in a buffer. When the index refreshes, the documents in the buffer are added to a new Lucene segment which is also held in memory.
Flushing is the process that stores the in-memory segments onto disk, and simultaneously closes the translog generation and starts with a new blank translog generation.
Flushing is a process which is carried out behind the scenes by Elasticsearch using heuristics to optimise the tradeoffs between memory usage and disk writes. andFlushing normally requires no intervention by users, although a flush api is available.
In the event of a node crashing or restarting, then Elasticsearch will retrieve and flush any operations that were stored in the translog prior to the crash in order to ensure that data is not lost.