[ 
https://issues.apache.org/jira/browse/ARTEMIS-4971?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17883941#comment-17883941
 ] 

Clebert Suconic edited comment on ARTEMIS-4971 at 9/23/24 5:04 PM:
-------------------------------------------------------------------

[~jpbriquet]I have minimized the number of ocurrencies the unack would happen. 


Since there are cases where it's ok to miss an ACK, I am adding a configuration 
option to log.warn when missed.

Say if you have consumers on the target queue, you could have a legal case of a 
missed ack.


was (Author: clebertsuconic):
[~jpbriquet]I have minimized the number of ocurrencies the unack would happen. 
But in case it happens I will now log.warn


although there are a few cases where it's expected to not ACK.. like after a 
failure.. etc... 


But that's what a warn is... a WARN.. it could be ok in certain cases.

> AckManager "giving up" processing of Acks in warning log level
> --------------------------------------------------------------
>
>                 Key: ARTEMIS-4971
>                 URL: https://issues.apache.org/jira/browse/ARTEMIS-4971
>             Project: ActiveMQ Artemis
>          Issue Type: Bug
>          Components: AMQP, Broker
>    Affects Versions: 2.35.0, 2.36.0
>            Reporter: Jean-Pascal Briquet
>            Assignee: Clebert Suconic
>            Priority: Minor
>             Fix For: 2.38.0
>
>
> Within the AMQP AckManager, the following log message, currently at DEBUG 
> level, should be increased to WARN level:
> {code:java}
> logger.debug("Retried {} {} times, giving up on the entry now", retry, 
> retry.getPageAttempts());{code}
> Setting this log to WARN level would allow operators to be aware of two 
> things:
> - The AckManager may require configuration tuning of queue/page attempts. 
> This can easily occur with the default configuration when handling large 
> messages.
> - Missed message replication, which is crucial when replication is required, 
> allowing operators to investigate the source of the problem.
> The downside is that this change may trigger a large number of logs when 
> enabled on Artemis nodes containing messages created before the mirroring 
> configuration activation.
> This might not be desirable for everyone, so it could be beneficial to have 
> it enabled as an optional parameter, configurable via a system environment 
> variable, similar to the fine-tuning options available for AckManager.
> Ideally, the log should as well contain the queue name, which would allow to 
> track down the source of messages.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@activemq.apache.org
For additional commands, e-mail: issues-h...@activemq.apache.org
For further information, visit: https://activemq.apache.org/contact


Reply via email to