activemq-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ARTEMIS-201) Log warning if server can crash on OutOfMemory due to "misconfiguration"
Date Mon, 23 Nov 2015 20:29:11 GMT

    [ https://issues.apache.org/jira/browse/ARTEMIS-201?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15022928#comment-15022928
] 

ASF GitHub Bot commented on ARTEMIS-201:
----------------------------------------

Github user jbertram commented on the pull request:

    https://github.com/apache/activemq-artemis/pull/245#issuecomment-159054551
  
    The scenario you outline (X addresses with Y max-size-bytes and Z heap-size that would
require N addresses to go into paging mode before an OOME occurs) is just one possible scenario.
 
    The scenario I outlined previously is equally valid (1 address with Y max-size-bytes and
Y/2 heap-size) and no paging would be involved before an OOME occurred so your suggested message
wouldn't make sense.
    
    As far as using getting scared of this WARN message when they have many shallow destinations
that never receive enough messages to reach max-size-bytes I think they should be scared because
their broker is in a vulnerable position.  If something goes wrong and they start getting
increased message production then they'll go OOME due to their sub-optimal configuration.
    
    Whether or not it should be a WARN or an INFO is debatable, IMO. Being vulnerable to an
OOME if someone dumps a bunch of messages on the server seems fairly serious to me.


> Log warning if server can crash on OutOfMemory due to "misconfiguration"
> ------------------------------------------------------------------------
>
>                 Key: ARTEMIS-201
>                 URL: https://issues.apache.org/jira/browse/ARTEMIS-201
>             Project: ActiveMQ Artemis
>          Issue Type: New Feature
>          Components: Broker
>    Affects Versions: 1.0.0
>            Reporter: Miroslav Novak
>            Assignee: Justin Bertram
>             Fix For: 1.1.1
>
>
> Imagine situation where server is started with 3000 destinations and max-size-bytes is
set to 10MB. This would mean that JVM would have to be started with at least 30GB of memory
to prevent OOM in case that all destinations get filled up. (PAGE mode is not a solution in
this case as it starts once destination exceeds 10MB in memory)
> Purpose of this jira is to provide check which would print warning in case that such
OOM can happen. This check would be executed during start of server and then with adding any
destination at runtime.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message