[
https://issues.apache.org/jira/browse/ARTEMIS-1025?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15951652#comment-15951652
]
ASF GitHub Bot commented on ARTEMIS-1025:
-----------------------------------------
Github user clebertsuconic commented on the issue:
https://github.com/apache/activemq-artemis/pull/1119
I see what you're doing now... if !isWritable, it think it should perform a
flush and block the thread...
We should always avoid actually calling write while isWritable...
I think this needs some simplification though...
i - just call a simple writeAndflush if beyond the limit... you may add a
log.debug (**debug**) statement if it needed to flush for the lack of flow
control from our codebase. We can use a fixed timeout here.. and log.warn if
missed. We have a callback timeout already.. we should just reuse the same.
(the less properties to be configured the better).
ii - that means no more crazy properties.
We may talk on monday about this.. but this is getting shape for me now..
we just need to clear the PR before we merge it.
> OutOfDirectMemoryError raised from Netty
> ----------------------------------------
>
> Key: ARTEMIS-1025
> URL: https://issues.apache.org/jira/browse/ARTEMIS-1025
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Components: Broker
> Reporter: Francesco Nigro
> Assignee: Francesco Nigro
>
> If you send and receive a lot of messages in short time to Artemis via Netty
> connector, the OutOfDirectMemoryError exception is thrown from the client.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
