FLUSH
The FLUSH
command commits any pending data changes and forces RisingWave to persist updated data to storage, which guarantees subsequent reads can access the latest data.
Syntax
Purpose of FLUSH
RisingWave uses a snapshot-based mechanism for data durability. Periodically, a snapshot is triggered which flushes operator state to storage like S3. If a failure occurs, operators recover from the latest snapshot.
However, snapshots are asynchronous. By default, RisingWave does not guarantee read-after-write consistency — a write may not be visible to subsequent reads immediately.
The FLUSH
command provides a way to synchronously persist pending changes and ensure subsequent reads reflect the latest data. It blocks until the changes are durable.
When to use FLUSH
FLUSH
should be used after any statements that change data, including:
- DML like
INSERT
,UPDATE
, orDELETE
. - DDL creating objects that query the data.
In the example below, FLUSH
ensures the pending data changes from the INSERT
are persisted, and the new materialized view has been populated before querying it: