order

This paper mainly studies Sharing-jdbc's SQLExecution Hook.

SQLExecutionHook

incubator-shardingsphere-4.0.0-RC1/sharding-core/sharding-core-execute/src/main/java/org/apache/shardingsphere/core/execute/hook/SQLExecutionHook.java

public interface SQLExecutionHook {
    
    /**
     * Handle when SQL execution started.
     * 
     * @param routeUnit route unit to be executed
     * @param dataSourceMetaData data source meta data
     * @param isTrunkThread is execution in trunk thread
     * @param shardingExecuteDataMap sharding execute data map
     */
    void start(RouteUnit routeUnit, DataSourceMetaData dataSourceMetaData, boolean isTrunkThread, Map<String, Object> shardingExecuteDataMap);
    
    /**
     * Handle when SQL execution finished success.
     */
    void finishSuccess();
    
    /**
     * Handle when SQL execution finished failure.
     *
     * @param cause failure cause
     */
    void finishFailure(Exception cause);
}

• The SQL Execution Hook interface defines start, finishSuccess, finishFailure methods

SPISQLExecutionHook

incubator-shardingsphere-4.0.0-RC1/sharding-core/sharding-core-execute/src/main/java/org/apache/shardingsphere/core/execute/hook/SPISQLExecutionHook.java

public final class SPISQLExecutionHook implements SQLExecutionHook {
    
    private final Collection<SQLExecutionHook> sqlExecutionHooks = NewInstanceServiceLoader.newServiceInstances(SQLExecutionHook.class);
    
    static {
        NewInstanceServiceLoader.register(SQLExecutionHook.class);
    }
    
    @Override
    public void start(final RouteUnit routeUnit, final DataSourceMetaData dataSourceMetaData, final boolean isTrunkThread, final Map<String, Object> shardingExecuteDataMap) {
        for (SQLExecutionHook each : sqlExecutionHooks) {
            each.start(routeUnit, dataSourceMetaData, isTrunkThread, shardingExecuteDataMap);
        }
    }
    
    @Override
    public void finishSuccess() {
        for (SQLExecutionHook each : sqlExecutionHooks) {
            each.finishSuccess();
        }
    }
    
    @Override
    public void finishFailure(final Exception cause) {
        for (SQLExecutionHook each : sqlExecutionHooks) {
            each.finishFailure(cause);
        }
    }
}

• SPISQL Execution Hook implements the SQL Execution Hook interface; it registers the SQL Execution Hook using New Instance Service Loader; the set of SQL Execution Hooks is created by New Instance Service Loader. New Service Instances; the start method traverses the SQL Execution Hooks and executes its start method; the find Success method traverses the SQL Execution Hooks and executes its find Success method; The rule traverses sqlExecution Hooks and executes its finishFailure method

OpenTracingSQLExecutionHook

incubator-shardingsphere-4.0.0-RC1/sharding-opentracing/src/main/java/org/apache/shardingsphere/opentracing/hook/OpenTracingSQLExecutionHook.java

public final class OpenTracingSQLExecutionHook implements SQLExecutionHook {
    
    private static final String OPERATION_NAME = "/" + ShardingTags.COMPONENT_NAME + "/executeSQL/";
    
    private ActiveSpan activeSpan;
    
    private Span span;
    
    @Override
    public void start(final RouteUnit routeUnit, final DataSourceMetaData dataSourceMetaData, final boolean isTrunkThread, final Map<String, Object> shardingExecuteDataMap) {
        if (!isTrunkThread) {
            activeSpan = ((ActiveSpan.Continuation) shardingExecuteDataMap.get(OpenTracingRootInvokeHook.ACTIVE_SPAN_CONTINUATION)).activate();
        }
        span = ShardingTracer.get().buildSpan(OPERATION_NAME)
                .withTag(Tags.COMPONENT.getKey(), ShardingTags.COMPONENT_NAME)
                .withTag(Tags.SPAN_KIND.getKey(), Tags.SPAN_KIND_CLIENT)
                .withTag(Tags.PEER_HOSTNAME.getKey(), dataSourceMetaData.getHostName())
                .withTag(Tags.PEER_PORT.getKey(), dataSourceMetaData.getPort())
                .withTag(Tags.DB_TYPE.getKey(), "sql")
                .withTag(Tags.DB_INSTANCE.getKey(), routeUnit.getDataSourceName())
                .withTag(Tags.DB_STATEMENT.getKey(), routeUnit.getSqlUnit().getSql())
                .withTag(ShardingTags.DB_BIND_VARIABLES.getKey(), toString(routeUnit.getSqlUnit().getParameters())).startManual();
        
    }
    
    private String toString(final List<Object> parameterSets) {
        return parameterSets.isEmpty() ? "" : String.format("[%s]", Joiner.on(", ").join(parameterSets));
    }
    
    @Override
    public void finishSuccess() {
        span.finish();
        if (null != activeSpan) {
            activeSpan.deactivate();
        }
    }
    
    @Override
    public void finishFailure(final Exception cause) {
        ShardingErrorSpan.setError(span, cause);
        span.finish();
        if (null != activeSpan) {
            activeSpan.deactivate();
        }
    }
}

• OpenTracing SQL Execution Hook implements the interface of SQL Execution Hook. Its start method creates and starts span, activeSpan; finishSuccess and finishFailure methods both execute span.finish() and activeSpan.deactivate(), but finishFailure marks the exception information of span.

SQLExecuteCallback

incubator-shardingsphere-4.0.0-RC1/sharding-core/sharding-core-execute/src/main/java/org/apache/shardingsphere/core/execute/sql/execute/SQLExecuteCallback.java

@RequiredArgsConstructor
public abstract class SQLExecuteCallback<T> implements ShardingGroupExecuteCallback<StatementExecuteUnit, T> {
    
    private final DatabaseType databaseType;
    
    private final boolean isExceptionThrown;
    
    @Override
    public final Collection<T> execute(final Collection<StatementExecuteUnit> statementExecuteUnits, final boolean isTrunkThread,
                                       final Map<String, Object> shardingExecuteDataMap) throws SQLException {
        Collection<T> result = new LinkedList<>();
        for (StatementExecuteUnit each : statementExecuteUnits) {
            result.add(execute0(each, isTrunkThread, shardingExecuteDataMap));
        }
        return result;
    }
    
    private T execute0(final StatementExecuteUnit statementExecuteUnit, final boolean isTrunkThread, final Map<String, Object> shardingExecuteDataMap) throws SQLException {
        ExecutorExceptionHandler.setExceptionThrown(isExceptionThrown);
        DataSourceMetaData dataSourceMetaData = DataSourceMetaDataFactory.newInstance(databaseType, statementExecuteUnit.getStatement().getConnection().getMetaData().getURL());
        SQLExecutionHook sqlExecutionHook = new SPISQLExecutionHook();
        try {
            sqlExecutionHook.start(statementExecuteUnit.getRouteUnit(), dataSourceMetaData, isTrunkThread, shardingExecuteDataMap);
            T result = executeSQL(statementExecuteUnit.getRouteUnit(), statementExecuteUnit.getStatement(), statementExecuteUnit.getConnectionMode());
            sqlExecutionHook.finishSuccess();
            return result;
        } catch (final SQLException ex) {
            sqlExecutionHook.finishFailure(ex);
            ExecutorExceptionHandler.handleException(ex);
            return null;
        }
    }
    
    protected abstract T executeSQL(RouteUnit routeUnit, Statement statement, ConnectionMode connectionMode) throws SQLException;
}

• The execute0 method of SQLExecuteCallback creates SPISQLExecutionHook before execution, then calls the sqlExecutionHook.start method, executes the sqlExecutionHook.finishSuccess method after execution, and executes the sqlExecutionHook.finishFailure method after capture of SQLException.

Summary

The SQL Execution Hook interface defines start, finishSuccess, finishFailure methods; SPISQLExecution Hook implements the SQL Execution Hook interface; it registers the SQL Execution Hook using New Instance Service Loader; the sqlExecution Hooks collection is created by New Instance Service Loader. new Service Instances; the start method traverses the SQL Execution Hooks and executes its start method; Then traverse sqlExecution Hooks to execute its finishSuccess method; the finishFailure method traverses sqlExecution Hooks to execute its finishFailure method

doc

• SQLExecutionHook