Package com.google.cloud.spanner

Interface TransactionContext

All Superinterfaces:

AutoCloseable, ReadContext

public interface TransactionContext
extends ReadContext

Context for a single attempt of a locking read-write transaction. This type of transaction is the only way to write data into Cloud Spanner; DatabaseClient.write(Iterable) and DatabaseClient.writeAtLeastOnce(Iterable) use transactions internally. These transactions rely on pessimistic locking and, if necessary, two-phase commit. Locking read-write transactions may abort, requiring the application to retry. However, the interface exposed by TransactionRunner eliminates the need for applications to write retry loops explicitly.

Locking transactions may be used to atomically read-modify-write data anywhere in a database. This type of transaction is externally consistent.

Clients should attempt to minimize the amount of time a transaction is active. Faster transactions commit with higher probability and cause less contention. Cloud Spanner attempts to keep read locks active as long as the transaction continues to do reads, and the transaction has not been terminated by returning from a TransactionRunner.TransactionCallable. Long periods of inactivity at the client may cause Cloud Spanner to release a transaction's locks and abort it.

Reads performed within a transaction acquire locks on the data being read. Writes can only be done at commit time, after all reads have been completed.

Conceptually, a read-write transaction consists of zero or more reads or SQL queries followed by a commit.

Semantics

Cloud Spanner can commit the transaction if all read locks it acquired are still valid at commit time, and it is able to acquire write locks for all writes. Cloud Spanner can abort the transaction for any reason. If a commit attempt returns ABORTED, Cloud Spanner guarantees that the transaction has not modified any user data in Cloud Spanner.

Unless the transaction commits, Cloud Spanner makes no guarantees about how long the transaction's locks were held for. It is an error to use Cloud Spanner locks for any sort of mutual exclusion other than between Cloud Spanner transactions themselves.

Retrying Aborted Transactions

When a transaction aborts, the application can choose to retry the whole transaction again. To maximize the chances of successfully committing the retry, the client should execute the retry in the same session as the original attempt. The original session's lock priority increases with each consecutive abort, meaning that each attempt has a slightly better chance of success than the previous.

Under some circumstances (e.g., many transactions attempting to modify the same row(s)), a transaction can abort many times in a short period before successfully committing. Thus, it is not a good idea to cap the number of retries a transaction can attempt; instead, it is better to limit the total amount of wall time spent retrying.

Application code does not need to retry explicitly; TransactionRunner will automatically retry a transaction if an attempt results in an abort.

Idle Transactions

A transaction is considered idle if it has no outstanding reads or SQL queries and has not started a read or SQL query within the last 10 seconds. Idle transactions can be aborted by Cloud Spanner so that they don't hold on to locks indefinitely. In that case, the commit will fail with error ABORTED.

If this behavior is undesirable, periodically executing a simple SQL query in the transaction (e.g., SELECT 1) prevents the transaction from becoming idle.

See Also:

• DatabaseClient.readWriteTransaction(TransactionOption...)
• TransactionRunner

Nested Class Summary

Nested classes/interfaces inherited from interface com.google.cloud.spanner.ReadContext

ReadContext.QueryAnalyzeMode

Method Summary

Modifier and Type	Method	Description
default ResultSetStats	analyzeUpdate(Statement statement, ReadContext.QueryAnalyzeMode analyzeMode, Options.UpdateOption... options)	Deprecated. Use analyzeUpdateStatement(Statement, QueryAnalyzeMode, UpdateOption...) instead to get both statement plan and parameter metadata
default ResultSet	analyzeUpdateStatement(Statement statement, ReadContext.QueryAnalyzeMode analyzeMode, Options.UpdateOption... options)	Analyzes a DML statement and returns query plan and statement parameter metadata and optionally execution statistics information.
long[]	batchUpdate(Iterable<Statement> statements, Options.UpdateOption... options)	Executes a list of DML statements in a single request.
com.google.api.core.ApiFuture<long[]>	batchUpdateAsync(Iterable<Statement> statements, Options.UpdateOption... options)	Same as batchUpdate(Iterable, UpdateOption...), but is guaranteed to be non-blocking.
void	buffer(Mutation mutation)	Buffers a single mutation to be applied if the transaction commits successfully.
void	buffer(Iterable<Mutation> mutations)	Buffers mutations to be applied if the transaction commits successfully.
default com.google.api.core.ApiFuture<Void>	bufferAsync(Mutation mutation)	Same as buffer(Mutation), but is guaranteed to be non-blocking.
default com.google.api.core.ApiFuture<Void>	bufferAsync(Iterable<Mutation> mutations)	Same as buffer(Iterable), but is guaranteed to be non-blocking.
long	executeUpdate(Statement statement, Options.UpdateOption... options)	Executes the DML statement (which can be a simple DML statement or DML statement with a returning clause) and returns the number of rows modified.
com.google.api.core.ApiFuture<Long>	executeUpdateAsync(Statement statement, Options.UpdateOption... options)	Same as executeUpdate(Statement,UpdateOption...), but is guaranteed to be non-blocking.

Methods inherited from interface com.google.cloud.spanner.ReadContext

analyzeQuery, close, executeQuery, executeQueryAsync, read, readAsync, readRow, readRowAsync, readRowUsingIndex, readRowUsingIndexAsync, readUsingIndex, readUsingIndexAsync

Method Details

buffer

void buffer(Mutation mutation)

Buffers a single mutation to be applied if the transaction commits successfully. The effects of this mutation will not be visible to subsequent operations in the transaction. All buffered mutations will be applied atomically.

bufferAsync

default com.google.api.core.ApiFuture<Void> bufferAsync(Mutation mutation)

Same as buffer(Mutation), but is guaranteed to be non-blocking.

buffer

void buffer(Iterable<Mutation> mutations)

Buffers mutations to be applied if the transaction commits successfully. The effects of the mutations will not be visible to subsequent operations in the transaction. All buffered mutations will be applied atomically.

bufferAsync

default com.google.api.core.ApiFuture<Void> bufferAsync(Iterable<Mutation> mutations)

Same as buffer(Iterable), but is guaranteed to be non-blocking.

executeUpdate

long executeUpdate(Statement statement,
 Options.UpdateOption... options)

Executes the DML statement (which can be a simple DML statement or DML statement with a returning clause) and returns the number of rows modified. For non-DML statements, it will result in an IllegalArgumentException. The effects of the DML statement will be visible to subsequent operations in the transaction.

executeUpdateAsync

com.google.api.core.ApiFuture<Long> executeUpdateAsync(Statement statement,
 Options.UpdateOption... options)

Same as executeUpdate(Statement,UpdateOption...), but is guaranteed to be non-blocking. If multiple asynchronous update statements are submitted to the same read/write transaction, the statements are guaranteed to be submitted to Cloud Spanner in the order that they were submitted in the client. This does however not guarantee that an asynchronous update statement will see the results of all previously submitted statements, as the execution of the statements can be parallel. If you rely on the results of a previous statement, you should block until the result of that statement is known and has been returned to the client.

analyzeUpdate

@Deprecated
default ResultSetStats analyzeUpdate(Statement statement,
 ReadContext.QueryAnalyzeMode analyzeMode,
 Options.UpdateOption... options)

Deprecated.

Use analyzeUpdateStatement(Statement, QueryAnalyzeMode, UpdateOption...) instead to get both statement plan and parameter metadata

Analyzes a DML statement and returns query plan and/or execution statistics information.

ReadContext.QueryAnalyzeMode.PLAN only returns the plan for the statement. ReadContext.QueryAnalyzeMode.PROFILE executes the DML statement, returns the modified row count and execution statistics, and the effects of the DML statement will be visible to subsequent operations in the transaction.

analyzeUpdateStatement

default ResultSet analyzeUpdateStatement(Statement statement,
 ReadContext.QueryAnalyzeMode analyzeMode,
 Options.UpdateOption... options)

Analyzes a DML statement and returns query plan and statement parameter metadata and optionally execution statistics information.

ReadContext.QueryAnalyzeMode.PLAN only returns the plan and parameter metadata for the statement. ReadContext.QueryAnalyzeMode.PROFILE executes the DML statement, returns the modified row count and execution statistics, and the effects of the DML statement will be visible to subsequent operations in the transaction.

batchUpdate

long[] batchUpdate(Iterable<Statement> statements,
 Options.UpdateOption... options)

Executes a list of DML statements in a single request. The statements will be executed in order and the semantics is the same as if each statement is executed by executeUpdate in a loop. This method returns an array of long integers, each representing the number of rows modified by each statement.

If an individual statement fails, execution stops and a SpannerBatchUpdateException is returned, which includes the error and the number of rows affected by the statements that are run prior to the error.

For example, if statements contains 3 statements, and the 2nd one is not a valid DML. This method throws a SpannerBatchUpdateException that contains the error message from the 2nd statement, and an array of length 1 that contains the number of rows modified by the 1st statement. The 3rd statement will not run.

batchUpdateAsync

com.google.api.core.ApiFuture<long[]> batchUpdateAsync(Iterable<Statement> statements,
 Options.UpdateOption... options)

Same as batchUpdate(Iterable, UpdateOption...), but is guaranteed to be non-blocking. If multiple asynchronous update statements are submitted to the same read/write transaction, the statements are guaranteed to be submitted to Cloud Spanner in the order that they were submitted in the client. This does however not guarantee that an asynchronous update statement will see the results of all previously submitted statements, as the execution of the statements can be parallel. If you rely on the results of a previous statement, you should block until the result of that statement is known and has been returned to the client.