You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@activemq.apache.org by "Gary Tully (JIRA)" <ji...@apache.org> on 2011/04/27 18:12:03 UTC
[jira] [Commented] (AMQ-3299) Prioritizing of messages isn't
applied to all messages during test.
[ https://issues.apache.org/jira/browse/AMQ-3299?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13025868#comment-13025868 ]
Gary Tully commented on AMQ-3299:
---------------------------------
You need to enable priority dispatch via the broker config (destination policy) for a destination.
Also note that the default store, kahaDB supports high (>4), default(4) and low(<4),so three priorities only. The JDBC store supports 0-9.
Have a look at the org.apache.activemq.store.MessagePriorityTest for the configuration. There are two variants, JDBC and KahaDB that show the difference in implementation.
eg:
http://svn.apache.org/viewvc/activemq/trunk/activemq-core/src/test/java/org/apache/activemq/store/jdbc/JDBCMessagePriorityTest.java?view=log
http://svn.apache.org/viewvc/activemq/trunk/activemq-core/src/test/java/org/apache/activemq/store/kahadb/KahaDBMessagePriorityTest.java?view=log
> Prioritizing of messages isn't applied to all messages during test.
> -------------------------------------------------------------------
>
> Key: AMQ-3299
> URL: https://issues.apache.org/jira/browse/AMQ-3299
> Project: ActiveMQ
> Issue Type: Bug
> Components: Broker
> Affects Versions: 5.4.2
> Environment: Operating System = Windows XP version 5.1 running on x86
> Java; VM; Vendor = 1.6.0_23
> Runtime = Java HotSpot(TM) Client VM 19.0-b09
> Reporter: Robert Liguori
> Attachments: priority_consumer_log.txt, priority_publisher_log.txt
>
>
> Considering this link: http://activemq.apache.org/activemq-performance-module-users-manual.html and sample test performance benchmarks of 1 million messages every 5 mins, I created and ran a priority test with 333 messages being sent every second.
> Problem: It appears as if the priority algorithm is not being applied (or even attempted) to many of the messages in the middle of the run. Consider this output and examine the message order. Note: the number following the #s represent the message and order-number sent by the producer. In this piece of the consumer log, these numbers are sequentially ordered meaning that the producer did not order them (based on priority) before sending them out:
> ...
> 09:59:58 INFO MESSAGE PRI-4 #20951 Count-19953
> 09:59:58 INFO MESSAGE PRI-9 #20952 Count-19954
> 09:59:58 INFO MESSAGE PRI-2 #20953 Count-19955
> 09:59:58 INFO MESSAGE PRI-5 #20954 Count-19956
> 09:59:58 INFO MESSAGE PRI-5 #20955 Count-19957
> 09:59:58 INFO MESSAGE PRI-9 #20956 Count-19958
> 09:59:58 INFO MESSAGE PRI-7 #20957 Count-19959
> 09:59:58 INFO MESSAGE PRI-0 #10648 Count-19960
> 09:59:58 INFO MESSAGE PRI-3 #20959 Count-19961
> 09:59:58 INFO MESSAGE PRI-2 #20960 Count-19962
> 09:59:58 INFO MESSAGE PRI-8 #20961 Count-19963
> 09:59:58 INFO MESSAGE PRI-4 #20962 Count-19964
> 09:59:58 INFO MESSAGE PRI-7 #20963 Count-19965
> 09:59:58 INFO MESSAGE PRI-3 #20964 Count-19966
> 09:59:58 INFO MESSAGE PRI-9 #20965 Count-19967
> 09:59:58 INFO MESSAGE PRI-3 #20966 Count-19968
> 09:59:58 INFO MESSAGE PRI-7 #20967 Count-19969
> 09:59:58 INFO MESSAGE PRI-1 #20968 Count-19970
> 09:59:58 INFO MESSAGE PRI-8 #20969 Count-19971
> 09:59:58 INFO MESSAGE PRI-1 #20970 Count-19972
> 09:59:58 INFO MESSAGE PRI-8 #20971 Count-19973
> 09:59:58 INFO MESSAGE PRI-9 #20972 Count-19974
> 09:59:58 INFO MESSAGE PRI-2 #20973 Count-19975
> 09:59:58 INFO MESSAGE PRI-4 #20974 Count-19976
> 09:59:58 INFO MESSAGE PRI-6 #20975 Count-19977
> 09:59:58 INFO MESSAGE PRI-8 #20976 Count-19978
> 09:59:58 INFO MESSAGE PRI-2 #20977 Count-19979
> 09:59:58 INFO MESSAGE PRI-5 #20978 Count-19980
> 09:59:58 INFO MESSAGE PRI-6 #20979 Count-19981
> 09:59:58 INFO MESSAGE PRI-5 #20980 Count-19982
> 09:59:58 INFO MESSAGE PRI-1 #20981 Count-19983
> 09:59:58 INFO MESSAGE PRI-4 #20982 Count-19984
> 09:59:58 INFO MESSAGE PRI-5 #20983 Count-19985
> 09:59:58 INFO MESSAGE PRI-2 #20984 Count-19986
> 09:59:58 INFO MESSAGE PRI-6 #20985 Count-19987
> 09:59:58 INFO MESSAGE PRI-8 #20986 Count-19988
> 09:59:58 INFO MESSAGE PRI-7 #20987 Count-19989
> 09:59:58 INFO MESSAGE PRI-3 #20988 Count-19990
> 09:59:58 INFO MESSAGE PRI-7 #20989 Count-19991
> 09:59:58 INFO MESSAGE PRI-3 #20990 Count-19992
> 09:59:58 INFO MESSAGE PRI-4 #20991 Count-19993
> 09:59:58 INFO MESSAGE PRI-5 #20992 Count-19994
> 09:59:58 INFO MESSAGE PRI-6 #20993 Count-19995
> 09:59:58 INFO MESSAGE PRI-1 #20994 Count-19996
> 09:59:58 INFO MESSAGE PRI-0 #10667 Count-19997
> 09:59:58 INFO MESSAGE PRI-7 #20996 Count-19998
> 09:59:58 INFO MESSAGE PRI-3 #20997 Count-19999
> 09:59:58 INFO MESSAGE PRI-8 #20998 Count-20000
> 09:59:58 INFO MESSAGE PRI-8 #20999 Count-20001
> ...
> I am separately attaching fuller consumer and producer logs. Clear ordering of messages is noticeable at the beginning and end of the test in the consumer log. Actually, there appears to the sending the high priority messages, followed by the sending of expedited messages intermixed with aged normal priority messages, followed by no ordering, followed by low priority messages with the value of '0'. The question is... why did the message ordering stop?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira