I think you are seeing a bug that I resolved in the past few weeks - https://issues.apache.org/jira/browse/ARTEMIS-4418 the openwire client prefetched messages can get out of sequence and the delivery count can get updated in error. I think 2.31.0 will help. an alternative approach would be to use the spring-boot-starter-artemis to use a different protocol. AMQP or CORE will behave differently, the issues is with the openwire protocol implementation see: https://mvnrepository.com/artifact/org.springframework.boot/spring-boot-starter-artemis/3.1.4
On Tue, 3 Oct 2023 at 13:35, Christian Esposito <christian.espos...@iet.it> wrote: > Hello, > > We are using ActiveMQ Artemis to enable multiple web applications of the > same type (developed with Spring Boot) to send messages with each other to > keep data up to date. When working in a development environment such as > Eclipse/IntelliJ IDEA, we are able to have these instances communicate with > each other without issues. However, we encounter issues when we deploy the > application on Wildfly in both development and production machines. > > In both environments, there is a server hosting the Artemis broker, and > there are other servers, each hosting various instances of the application. > The problem we encounter is the loss of some messages, which end up in the > DLQ (Dead Letter Queue). This issue particularly occurs as network traffic > increases, especially on queues that accumulate hundreds or sometimes > thousands of messages. Upon examining the Artemis logs, this is what is > reported for each lost message: > WARN [org.apache.activemq.artemis.core.server] AMQ222149: Message > Reference[8970196914]:RELIABLE:CoreMessage[messageID=8970196914,durable=true,userID=1adf1f6d-119a-11ee-93f5-005056bbcfcf,priority=4, > timestamp=Fri Jun 23 09:46:46 CEST 2023,expiration=0, durable=true, > address=I01-or,size=35909,properties=TypedProperties[__AMQ_CID=ID:A12.sg.it-34045-1681291484686-0:1,_AMQ_GROUP_SEQUENCE=0,__HDR_BROKER_IN_TIME=1687506406853,_AMQ_ROUTING_TYPE=1,__HDR_ARRIVAL=0,__HDR_COMMAND_ID=192161375,__HDR_PRODUCER_ID=[0000 > 0040 7B01 002C 4944 3A41 4947 4F52 2D31 322E 736F 6769 6E2E 6974 2D33 ... > 3132 3931 3438 3436 3836 2D31 3A31 0000 0000 0000 0001 0000 0000 0298 > 3F1A),_AMQ_DUPL_ID=ID:A12.sg.it-34045-1681291484686-1:1:43532058:1:1,__HDR_MESSAGE_ID=[0000 > 0053 6E00 017B 0100 2C49 443A 4149 474F 522D 3132 2E73 6F67 696E 2E69 ... > 00 0000 0000 0100 0000 0002 983F 1A00 0000 0000 0000 0100 0000 0000 0000 > 00),__HDR_DROPPABLE=false]]@1995702679 has reached maximum delivery > attempts, sending it to Dead Letter Address DLQ from I01-or > > As indicated by the logs, the maximum number of attempts to recover the > message has been reached. However, upon analyzing the application logs, > there are no errors or any information indicating an attempt by the > application to retrieve the message (for each received message, logs are > generated at the time of message reception). Additionally, there are no > messages indicating connection issues to the Artemis broker. > > We attempted to replicate the same scenario locally by using a Docker > container for Artemis and a virtual machine to deploy the application (with > the same OS as the development and production machines). When deploying the > application on Wildfly we encounter a similar issue as described earlier. > However, if we try to run the application from the virtual machine but in a > development environment (IntelliJ IDEA), the issue does not occur (we > performed the same operations in both cases and used the same configuration > to connect to the Artemis broker). > > Regarding the versions used, we provide them below: > > * ActiveMQ Artemis, version 2.16.0. We also tried using a more recent > version of Artemis (2.30.0), but we have the same problem. > * Web Application: Spring Boot, Java 8, Tomcat web server. > * Database: PostgreSQL version 10.10. > * Wildfly 18.0.0. > As for the configuration parameters used for Artemis, the broker.xml > configuration file has not been modified. The configuration takes place > within the application. We are using the starter > ‘spring-boot-starter-activemq’ with the following settings: > > * Artemis Mode: native. > * Connection Factory - trust al packages: true. > * Connection Factory – connection response timeout: 999999. > * Connection Factory – close timeout: 999999. > * Connection Factory – send Timeout: 999999. > * Connection Factory – use async send: true. > * Redelivery Policy - initial redelivery delay: 1000. > * Redelivery Policy - back off multiplier: 3. > * Redelivery Policy - max redeliveries: 7. > * Redelivery Policy - max redelivery delay: 60000. > * Default JMS Container Factory – recovery interval: 5000. > > What could be the cause of the problem, and how can we resolve it? > > Thank you, > > Christian Esposito >
