我非常確定空閒連接沒有被重新使用或者我正在泄漏連接。我有一個從文件使用者開始的簡單路線。文件使用者使用文本文件。拿起文件後,我檢查一個表以確保這不是重複的文件。DBCP空閒連接不在駱駝路由中重複使用
然後,我將消息正文從文件轉換爲字符串。然後,我將文件拆分並通過一條路徑運行各個部分,具體取決於它是哪種類型的記錄。這些路由中的每一條最終都會將此記錄插入運行在MySQL上的服務器上的登臺表中。
下面是路線的簡化版本。
<bean id="myPool" class="java.util.concurrent.Executors" factory-method="newFixedThreadPool">
<argument index="0" value="8"/>
</bean>
<camelContext trace="false" handleFault="true" errorHandlerRef="redeliveryErrorHandler" id="FileETLProcess" xmlns="http://camel.apache.org/schema/blueprint">
<errorHandler type="DefaultErrorHandler" useOriginalMessage="true" id="redeliveryErrorHandler">
<redeliveryPolicy maximumRedeliveries="3" redeliveryDelay="25" backOffMultiplier="2" useExponentialBackOff="false" retryAttemptedLogLevel="INFO"/>
</errorHandler>
<onException useOriginalMessage="true">
<exception>java.lang.Exception</exception>
<handled>
<constant>true</constant>
</handled>
<log message="STARTING ERROR GENERAL HANDLER"/>
<bean ref="GeneralError"/>
<to uri="smtp://[email protected]&[email protected]&subject=GENERAL ERROR: A File Could Not Be Imported&contentType=text/html"/>
<to uri="file:d:/Inbox/.badFile?fileName=${file:onlyname.noext}_GENERALERROR_${date:now:yyyyMMddHHmmss}.${file:name.ext}"/>
</onException>
<route id="ExtractFileRoute">
<from uri="file:d:/Inbox?delay=10000&move=.donebackup/${date:now:yyyyMMdd}/${file:onlyname.noext}_DONE_${date:now:yyyyMMddHHmmss}.${file:name.ext}&readLock=changed&include=.*.dl&maxMessagesPerPoll=0&sortBy=${file:length}"/>
<bean ref="FileCheck"/>
<choice>
<when>
<simple>${header.ACCEPTEDFILE} == 'YES'</simple>
<log message="File Import Route Started At:${date:now:yyyy-MM-dd HH:mm:ss}"/>
<convertBodyTo type="java.lang.String"/>
<log message="Converted File To String:${date:now:yyyy-MM-dd HH:mm:ss} handing data to File To DB route."/>
<split parallelProcessing="true" executorServiceRef="myPool" streaming="true" shareUnitOfWork="true">
<tokenize token="\n"></tokenize>
<setHeader headerName="SPLITFINISHED">
<simple>${property.CamelSplitComplete}</simple>
</setHeader>
<setHeader headerName="SPLITNUMBER">
<simple>${property.CamelSplitIndex}</simple>
</setHeader>
<bean ref="EnrichHeader"/>
<choice>
<when>
<simple>${header.RECORDTYPE} == 'HEADER'</simple>
<doTry>
<unmarshal ref="bindyHeader"/>
<bean ref="HeaderPersist"/>
<choice>
<when>
<simple>${property.CamelSplitComplete} == true</simple>
<to uri="direct:auxrecordsmove"/>
</when>
</choice>
<doCatch>
<exception>java.lang.Exception</exception>
<handled>
<constant>true</constant>
</handled>
<bean ref="RecordErrorReport"/>
<choice>
<when>
<simple>${property.CamelSplitComplete} == true</simple>
<to uri="direct:auxrecordsmove"/>
</when>
</choice>
</doCatch>
</doTry>
</when>
<when>
<simple>${header.RECORDTYPE} == 'A'</simple>
<doTry>
<unmarshal ref="bindyAccount"/>
<bean ref="AccountPersist"/>
<choice>
<when>
<simple>${property.CamelSplitComplete} == true</simple>
<to uri="direct:auxrecordsmove"/>
</when>
</choice>
<doCatch>
<exception>java.lang.Exception</exception>
<handled>
<constant>true</constant>
</handled>
<bean ref="RecordErrorReport"/>
<choice>
<when>
<simple>${property.CamelSplitComplete} == true</simple>
<to uri="direct:auxrecordsmove"/>
</when>
</choice>
</doCatch>
</doTry>
</when>
<when>
<simple>${header.RECORDTYPE} == 'C'</simple>
<doTry>
<unmarshal ref="bindyComaker"/>
<bean ref="CoMakerPersist"/>
<choice>
<when>
<simple>${property.CamelSplitComplete} == true</simple>
<to uri="direct:auxrecordsmove"/>
</when>
</choice>
<doCatch>
<exception>java.lang.Exception</exception>
<handled>
<constant>true</constant>
</handled>
<bean ref="RecordErrorReport"/>
<choice>
<when>
<simple>${property.CamelSplitComplete} == true</simple>
<to uri="direct:auxrecordsmove"/>
</when>
</choice>
</doCatch>
</doTry>
</when>
Some other beans here........
<when>
<simple>${property.CamelSplitComplete} == true</simple>
<to uri="direct:auxrecordsmove"/>
</when>
<otherwise>
<to uri="smtp://[email protected]&[email protected]&subject=URGENT:UNKOWN RECORD TYPE FOUND IN FILE"/>
<choice>
<when>
<simple>${property.CamelSplitComplete} == true</simple>
<to uri="direct:auxrecordsmove"/>
</when>
</choice>
</otherwise>
</choice>
</split>
</when>
<otherwise>
<to uri="file:d:/RMSInbox/.badFile?fileName=${file:onlyname.noext}_POSSIBLE_DUPLICATE_ERROR_${date:now:yyyyMMddHHmmss}.${file:name.ext}"/>
<bean ref="FileErrorReport"/>
<to uri="smtp://[email protected]&[email protected]&subject=ERROR: A File Could Not Be Imported&contentType=text/html"/>
</otherwise>
</choice>
</route>
因此,這條路線上的每條消息最終都會碰到一個將它插入數據庫的bean。因此,我增加DBCP的依賴關係,然後在我的OSGi的XML藍圖聲明,如下所示:
<bean id="dataSource" class="org.apache.commons.dbcp.BasicDataSource" destroy-method="close">
<property name="driverClassName" value="com.mysql.jdbc.Driver"/>
<property name="url" value="jdbc:mysql://Canttouchthis:3306/ETLDB"/>
<property name="username" value="ETLUser"/>
<property name="password" value="password"/>
<property name="initialSize" value="2"/>
<property name="maxActive" value="16"/>
<property name="maxIdle" value="16"/>
<property name="minIdle" value="2"/>
<property name="timeBetweenEvictionRunsMillis" value="180000"/>
<property name="minEvictableIdleTimeMillis" value="180000"/>
<property name="testOnBorrow" value="true"/>
<property name="testWhileIdle" value="true"/>
<property name="testOnReturn" value="true"/>
<property name="validationQuery" value="SELECT 1"/>
<property name="maxWait" value="10000"/>
<property name="removeAbandoned" value="true"/>
<property name="logAbandoned" value="false"/>
<property name="removeAbandonedTimeout" value="300"/>
</bean>
我也宣佈我的豆子,將做處理這樣的:
<bean id="AccountPersist" class="com.foo.NewAccount.AccountInformationToDatabase">
<property name="dataSource" ref="dataSource"/>
</bean>
現在拆分時在文件已經完成我想確保記錄匹配。基本上文件有帳戶記錄和一些輔助信息。所以我在分離完成時檢查路由,然後一旦文件完全在暫存表中,我在MySQL中執行一些額外的完整性檢查。
這第二條路看起來是這樣的:
<route trace="false" id="MoveMatchingAuxRecordsFromStage">
<from uri="direct:auxrecordsmove"/>
<log message="File Import Route Ended At:${date:now:yyyy-MM-dd HH:mm:ss}"/>
<log message="ETL Route Start AT: ${date:now:yyyy-MM-dd HH:mm:ss}"/>
<log message="Moving Matching Comaker records at: ${date:now:yyyy-MM-dd HH:mm:ss}"/>
<bean ref="CoMakerETL"/>
<log message="Matching Comaker records move finised at: ${date:now:yyyy-MM-dd HH:mm:ss}"/>
<log message="Moving Matching Credit History records at: ${date:now:yyyy-MM-dd HH:mm:ss}"/>
<bean ref="CreditHistoryETL"/>
<log message="Matching Credit History records move finised at: ${date:now:yyyy-MM-dd HH:mm:ss}"/>
<log message="Moving Matching Extra Detail records at: ${date:now:yyyy-MM-dd HH:mm:ss}"/>
<bean ref="ExtraDetailInformationETL"/>
<log message="Matching Extra Detail records move finised at: ${date:now:yyyy-MM-dd HH:mm:ss}"/>
<log message="Moving Legal Information records at: ${date:now:yyyy-MM-dd HH:mm:ss}"/>
<bean ref="LegalInformationETL"/>
<log message="Matching Legal Information records move finised at: ${date:now:yyyy-MM-dd HH:mm:ss}"/>
<log message="ETL Route Finished At ${date:now:yyyy-MM-dd HH:mm:ss}"/>
</route>
所以在我的測試一切正常,我可以導入安靜有效像這樣的文件。當我在文件夾中放置超過5個文件時,我的問題就開始了。基本上,我觀察MySQL將連接池擴展到最大大小,然後不重新使用連接。
所以我們打16個併發連接他們去已經在4,5,6文件加載的地方,然後我突然收到以下錯誤後的幾個文件,睡覺:
Cannot get a connection, pool error Timeout waiting for idle object
或者,正如它出現在日誌
[ pool-3-thread-35] oveMatchingAuxRecordsFromStage INFO Matching Extra Detail records move finised at: 2013-07-26 17:41:59
[ pool-3-thread-35] oveMatchingAuxRecordsFromStage INFO Moving Legal Information records at: 2013-07-26 17:41:59
[ pool-3-thread-35] DefaultErrorHandler INFO Failed delivery for (MessageId: ID-IMS-WS2013-001-52799-1374824474993-0-2693 on ExchangeId: ID-IMS-WS2013- 001-52799-1374824474993-0-3230). On delivery attempt: 0 caught: java.lang.Exception: Cannot get a connection, pool error Timeout waiting for idle object
[thread #0 - file://d:/RMSInbox] ExtractRMSNewAccountFileRoute INFO STARTING ERROR GENERAL HANDLER
在MySQL的最大連接數已經被推到一個巨大的512。我曾嘗試過各種池大小,線程選項等
作爲的問題我感興趣的是我的所有JDBC代碼遵循這種結構。它沒有複雜的SQL剛插入語句....
public class RecordParserErrorReporter {
private static final String SQL_INSERT="INSERT INTO `ETL`.`ETLLog` "+
" ("+
" `ETL_log_text`, "+
" `filename`, "+
" `record_number`, "+
" `error_message`) "+
" VALUES "+
" ("+
" ?, "+
" ?, "+
" ?, "+
" ? "+
"); ";
private BasicDataSource dataSource;
public BasicDataSource getDataSource() {
return dataSource;
}
public void setDataSource(BasicDataSource dataSource) {
this.dataSource = dataSource;
}
public void HandleError
(
@Body String msgBody
, @Headers Map hdr
, Exchange exch
)
{
Connection conn = null;
PreparedStatement stmt=null;
try
{
Exception e = exch.getProperty(Exchange.EXCEPTION_CAUGHT,Exception.class);
conn= dataSource.getConnection();
stmt =conn.prepareStatement(SQL_INSERT);
stmt.setString(1, msgBody);
stmt.setString(2, (String)hdr.get("CamelFileName"));
stmt.setInt(3, (Integer)hdr.get("SPLITNUMBER"));
stmt.setString(4, e.getMessage());
stmt.executeUpdate();
}
catch (Exception e)
{
System.out.println(e.getMessage());
}
finally
{
try
{
if (stmt!=null)
{
stmt.close();
}
if (conn!=null)
{
conn.close();
conn= null;
}
}
catch(SQLException e)
{
System.out.println(e.getMessage());
}
}
}
}
如何追查爲什麼我的連接不被重用?或者如果我泄漏連接,我該如何跟蹤?
更新:
我設置文件消費者消費的每一個投票1個文件與投票之間的第二延遲。我可以看到它爲每次路由啓動創建一個新的連接池,然後不重用先前創建的池。顯示對於我排隊的每個文件都會創建一個新的連接池。不完全是我想要的。它是否正確。
我還列舉了這看起來見下面的屏幕截圖:
更新2:
這不是在春天DM但OSGI藍圖運行。我相當確信數據源不止一次被實例化。
更新3:
好確定,因此DataSource不獲取實例化不止一次。閒置連接根本不用。我確實發現了一些有趣的東西,所以我懷疑這可能與我所看到的更多信息有關:鏈接:http://fusesource.com/forums/thread.jspa?threadID=4659&tstart=15
根據視圖的數量來判斷這個問題越來越明顯了。