activemq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Claus Ibsen (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (AMQ-2953) Disk limits not observed when memory limits exceeded for non-persistent messaging
Date Sat, 02 Nov 2013 20:27:20 GMT

     [ https://issues.apache.org/jira/browse/AMQ-2953?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Claus Ibsen updated AMQ-2953:
-----------------------------

    Fix Version/s: NEEDS_REVIEWED

> Disk limits not observed when memory limits exceeded for non-persistent messaging
> ---------------------------------------------------------------------------------
>
>                 Key: AMQ-2953
>                 URL: https://issues.apache.org/jira/browse/AMQ-2953
>             Project: ActiveMQ
>          Issue Type: Bug
>          Components: Message Store
>    Affects Versions: 5.3.2
>            Reporter: Richard Bonneau
>             Fix For: NEEDS_REVIEWED
>
>         Attachments: activemq-JMX-low-tempUsage-lowstoreUsage-STOMP.xml
>
>
> Using the Kahadb persistence adapter.
> When producing non-persistent messages and using the <systemUsage> element to specify
memory and disk limits,
> it appears that after memory limit is reached that staging incoming messages to disk
continues to happen even
> past the specified disk limit.  More specifically if <memoryUsage> limit is exceeded,
we see messages being
> stored into files labelled as data-TopicSubscription-<n>.log.  However, even when
<tempUsage> element specifies a limit on the
> disk space to be used, messages continue to be stored there and the limit is not adhered
to.
> Attaching the configuration file used for this bug report. I simply used the producer/consumer
programs in the example folder to populate and consume large number of messages.
> Note that with ActiveMQ 5.4, similar behavior except that the location of the data files
is in data\localhost\tmp_storage and the data files are named  db-<n>.log.
> We need to have the limit(s) adhered to and then the producer should be held up until
disk or memory is freed up as expected from the description of handling non-persistent messages.
 Under the current implementations, the producer can continue to produce messages until all
available disk space is allocated. 



--
This message was sent by Atlassian JIRA
(v6.1#6144)

Mime
View raw message