We are running an application on centos with tomcat 8.0.28, java 1.8 and activemq 5.13.4 For persistency we are using Oracle 11.2
We have 3 queues with several producers and consumers. >From time to time we have the effect that messages remain in the ACTIVEMQ_MSGS table in the oracle database though these messages are consumed correctly (we get correct results). These messages never get removed from the ACTIVEQ_MSG table. Our problem occurs, when we restart activemq, because these messages are redelivered. We checked the activemq-Logfile, the enqueue/dequeue counters for the queues are identical. What is astonishing us is the following logged DELETE command: --- 2017-06-09 10:24:58,281 [ Adaptor Task-2] DEBUG JournalPersistenceAdapter - Checkpoint started. 2017-06-09 10:24:58,291 [ Adaptor Task-2] DEBUG JDBCPersistenceAdapter - Cleaning up old messages. 2017-06-09 10:24:58,291 [ Adaptor Task-2] DEBUG DefaultJDBCAdapter - Executing SQL: DELETE FROM ACTIVEMQ_MSGS WHERE (PRIORITY=? AND ID <= ( SELECT min(ACTIVEMQ_ACKS.LAST_ACKED_ID) FROM ACTIVEMQ_ACKS WHERE ACTIVEMQ_ACKS.CONTAINER=ACTIVEMQ_MSGS.CONTAINER AND ACTIVEMQ_ACKS.PRIORITY=?) ) 2017-06-09 10:24:58,300 [ Adaptor Task-2] DEBUG DefaultJDBCAdapter - Deleted 0 old message(s) at priority: 0 2017-06-09 10:24:58,300 [ Adaptor Task-2] DEBUG JDBCPersistenceAdapter - Cleanup done. 2017-06-09 10:24:58,300 [ Adaptor Task-2] DEBUG JournalPersistenceAdapter - Checkpoint done. --- Though we have entries in ACTIVEMQ_MSGS we do not have entries in the ACTIVEMQ_ACKS table, so if I understand it correctly the statement above can never remove any records. Here is our configuration file: --- <broker brokerName="KWG_BROKER" useJmx="true" xmlns="http://activemq.apache.org/schema/core";> <persistenceFactory> <journalPersistenceAdapterFactory journalLogFiles="5" dataDirectory="${activemq.data}/activemq-data" dataSource="#oracle-ds"/> </persistenceFactory> <transportConnectors> <transportConnector uri="tcp://<ourhost>:8016?transport.trace=false&transport.soTimeout=60000&keepAlive=true"/> </transportConnectors> </broker> <bean id="oracle-ds" class="org.apache.commons.dbcp2.BasicDataSource" destroy-method="close"> <property name="driverClassName" value="oracle.jdbc.OracleDriver"/> <property name="url" value="jdbc:oracle:thin:@<ourhost2>:1521/<ourdbname>"/> <property name="username" value="benutzer"/> <property name="password" value="passwort"/> <property name="poolPreparedStatements" value="true"/> </bean> --- And a code snippet: ---- public void onMessage(final Message message, final Session session) { try { _log.debug("OmqListener.On Message Beginn"); ObjectMessage om = (ObjectMessage) message; Object obj = om.getObject(); if (obj instanceof OutMessage) { OutMessage ome = (OutMessage) obj; String messtype = ome.getTyp().toString(); String oid = message.getStringProperty(Constants.OID); String erstellungsdatum = message.getStringProperty(Constants.EINSTELLUNGSDATUM); _log.debug(new StringBuffer("Message erhalten Typus=").append(messtype) .append("; OID=").append(oid) .append("; Erstellungsdatum=").append(erstellungsdatum) .append(";Logging=").append(ome.isLogging()).toString()); _ausgangsService.performOutmessage(ome, _propholder.isAntwortLoeschenAusDb()); _log.info("onMessage: Anzahl Versuche = " + ome.getAnzahlVersuche()); } _log.debug("OmqListener.On Message Ende"); } catch (Throwable e) { //auf erneute Zustellversuche(Timer!) warten _log.error("Fehler!", e); } } --- When we restart activemq, what possibilities do I have to filter out in my programme those messages that have already been delivered ? I could delete messages from ACTIVEMQ_MSGS before we restart activemq but how can I decide between old messages (already delivered) and those that still need to be delivered ? Thanks ! -- View this message in context: http://activemq.2283324.n4.nabble.com/Oracle-delivered-Messages-remain-in-ACTIVEMQ-MSGS-tp4729613.html Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.
