Package io.trino.benchto.driver.execution

Class ExecutionSynchronizer

java.lang.Object

io.trino.benchto.driver.execution.ExecutionSynchronizer

@Component
public class ExecutionSynchronizer
extends Object

This class is responsible for synchronizing threads in driver if graphite metrics collection is enabled. Graphite collects metrics with predefined resolution, ex. 10 s.

After query/benchmark is finished we should wait at least 2 resolutions before we execute next query/benchmark, so runs does not interfere with each other.

Graphite metrics loading should be delayed at least 1 resolution to make sure that last probe was stored in graphite.

Constructor Summary

Constructors

Constructor	Description
ExecutionSynchronizer()

Method Summary

Modifier and Type	Method	Description
void	awaitAfterBenchmarkExecutionAndBeforeResultReport(Benchmark benchmark)	If metrics collection is enabled and we are doing concurrent benchmark, we should wait between benchmarks, so measurements are accurate.
void	awaitAfterQueryExecutionAndBeforeResultReport(QueryExecutionResult queryExecutionResult)	If metrics collection is enabled and we are doing serial benchmark, we should wait between queries, so measurements are accurate.
<T> CompletableFuture<T>	execute(java.time.Instant when, Callable<T> callable)	Executes callable when time comes.
void	shutdown()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

ExecutionSynchronizer

public ExecutionSynchronizer()

Method Details

shutdown

@PreDestroy
public void shutdown()
              throws InterruptedException

Throws:

InterruptedException

awaitAfterQueryExecutionAndBeforeResultReport

public void awaitAfterQueryExecutionAndBeforeResultReport(QueryExecutionResult queryExecutionResult)

If metrics collection is enabled and we are doing serial benchmark, we should wait between queries, so measurements are accurate.

awaitAfterBenchmarkExecutionAndBeforeResultReport

public void awaitAfterBenchmarkExecutionAndBeforeResultReport(Benchmark benchmark)

If metrics collection is enabled and we are doing concurrent benchmark, we should wait between benchmarks, so measurements are accurate.

execute

public <T> CompletableFuture<T> execute(java.time.Instant when,
 Callable<T> callable)

Executes callable when time comes. The callable gets executed immediately, without offloading to a backghround thread, if execution time requested has already passed.