[
https://issues.apache.org/jira/browse/ARTEMIS-1326?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16116705#comment-16116705
]
ASF GitHub Bot commented on ARTEMIS-1326:
-----------------------------------------
Github user franz1981 commented on the issue:
https://github.com/apache/activemq-artemis/pull/1444
@clebertsuconic
Probably we use the same term for different things: what do you mean by
burst and complete?
For what I mean if you start calling addBytes several times until timeout
is not finished yet and you have a flush method (months test observer) that do
not sleep at least timeout time, it is normal that you will have your callbacks
called in short time.
What is important is that the buffer can't allow to flush new data until a
new timeout will be expired: it is like the notion of credits for the flow
control.
If you finish the credits immediately you can't have new ones until the
timeout will finish and it is exactly what is the definition of IOPS for the
disks.
It is not a matter of MAPPED or not...but a matter of a test that is not
emulations any correct disk behaviour: asynchronously o not.
> testTimeOnTimedBuffer incorrect timeout count
> ---------------------------------------------
>
> Key: ARTEMIS-1326
> URL: https://issues.apache.org/jira/browse/ARTEMIS-1326
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Reporter: Francesco Nigro
> Assignee: Francesco Nigro
>
> The test isn't evaluating the timeout expiration while writing in batches
> into the TimedBuffer
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
