You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@activemq.apache.org by "Gary Tully (JIRA)" <ji...@apache.org> on 2017/03/03 11:28:46 UTC

[jira] [Resolved] (AMQ-3233) warnOnProducerFlowControl is never reset to 'true' after the first INFO message is logged.

     [ https://issues.apache.org/jira/browse/AMQ-3233?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Gary Tully resolved AMQ-3233.
-----------------------------
       Resolution: Fixed
    Fix Version/s:     (was: NEEDS_REVIEW)
                   5.15.0

strange that this was out there for so long!
in any event, the policy entry blockedProducerWarningInterval needed to be respected such that the logging rate can be regulated. The default value of 30s   means the behaviour is not unlike a single shot as it is now but in the case of long term blocking the warning will repeat as new producers are blocked.

note the advisoryForFastProducers gives a notification every time.
Also the jmx support for detecting flow control is improved by AMQ-6614

> warnOnProducerFlowControl is never reset to 'true' after the first INFO message is logged.
> ------------------------------------------------------------------------------------------
>
>                 Key: AMQ-3233
>                 URL: https://issues.apache.org/jira/browse/AMQ-3233
>             Project: ActiveMQ
>          Issue Type: Bug
>          Components: Broker
>    Affects Versions: 5.4.2
>         Environment: All
>            Reporter: Adrian Trenaman
>            Assignee: Gary Tully
>             Fix For: 5.15.0
>
>
> Looking at the code in Queue.java (around line 543), you can see that the warnOnProducerFlowControl variable is set to false when we log the INFO message "Usage Manager Memory Limit (...) reached on ....". However, it is never set to true again. So, if we consider operations users trawling through large logs over a period of time, the may see the first warning on Monday, but then see no further warnings of this behavior for the rest of the week. Surely we should be resetting the warnOnProducerFlowControl variable to true, so that subsequent flow control situations are correctly logged?



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)