You're viewing documentation for a previous version of Scylla Java Driver. Switch to the latest stable version.
In driver 4, statement classes are immutable. All mutating methods return a new instance, so make sure you don’t accidentally ignore their result:
BoundStatement boundSelect = preparedSelect.bind(); // This doesn't work: setInt and setPageSize don't modify boundSelect in place: boundSelect.setInt("k", key); boundSelect.setPageSize(1000); session.execute(boundSelect); // Instead, reassign the statement every time: boundSelect = boundSelect.setInt("k", key).setPageSize(1000);
All of these mutating methods are annotated with
@CheckReturnValue. Some code analysis tools –
such as ErrorProne – can check correct usage at build time, and report
mistakes as compiler errors.
The driver also provides builders:
BoundStatement boundSelect = preparedSelect.boundStatementBuilder() .setInt("k", key) .setPageSize(1000) .build();
Because it’s the right abstraction to use. A completable future, as its name indicates, is a future that can be completed manually; that is not what we want to return from our API: the driver completes the futures, not the user.
CompletionStage does not expose a
get() method; one can view that as an encouragement to
use a fully asynchronous programming model (chaining callbacks instead of blocking for a result).
At any rate,
CompletionStage has a
toCompletableFuture() method. In current JDK versions, every
CompletionStage is a
CompletableFuture, so the conversion has no performance overhead.
DowngradingConsistencyRetryPolicyfrom driver 3?¶
As of driver 4.10, this retry policy was made available again as a built-in alternative to the
default retry policy: see the manual to understand how to use it.
For versions between 4.0 and 4.9 inclusive, there is no built-in equivalent of driver 3
That retry policy was indeed removed in driver 4.0.0. The main motivation is that this behavior should be the application’s concern, not the driver’s. APIs provided by the driver should instead encourage idiomatic use of a distributed system like Apache Cassandra, and a downgrading policy works against this. It suggests that an anti-pattern such as “try to read at QUORUM, but fall back to ONE if that fails” is a good idea in general use cases, when in reality it provides no better consistency guarantees than working directly at ONE, but with higher latencies.
However, we recognize that there are use cases where downgrading is good – for instance, a dashboard application would present the latest information by reading at QUORUM, but it’s acceptable for it to display stale information by reading at ONE sometimes.
Thanks to JAVA-2900, an equivalent retry policy with downgrading behavior was re-introduced in driver 4.10. Nonetheless, we urge users to avoid using it unless strictly required, and instead, carefully choose upfront the consistency level that works best for their use cases. Even if there is a legitimate reason to downgrade and retry, that should be preferably handled by the application code. An example of downgrading retries implemented at application level can be found in the driver examples repository.
In driver 3, it was possible to configure the load balancing policy to automatically failover to a remote datacenter, when the local datacenter is down.
This ability is considered a misfeature and has been removed from driver 4.0 onwards.
However, due to popular demand, cross-datacenter failover has been brought back to driver 4 in version 4.10.0.
If you are using a driver version >= 4.10.0, read the manual to understand how to enable this feature; for driver versions < 4.10.0, this feature is simply not available.
The driver now uses Java 8’s improved date and time API. CQL type
timestamp is mapped to
java.time.Instant, and the corresponding getter and setter are
See Temporal types for more details.
If you benchmark DDL queries such as
session.execute("CREATE TABLE ..."), you will observe a
noticeably higher latency than driver 3 (about 1 second).
This is because those queries are now debounced: the driver adds a short wait in an attempt to group multiple schema changes into a single metadata refresh. If you want to mitigate this, you can either adjust the debouncing settings, or group your schema updates while temporarily disabling the metadata; see the performance page.
This only applies to DDL queries; DML statements (
INSERT…) are not debounced.
On this page
DowngradingConsistencyRetryPolicyfrom driver 3?