[
https://issues.apache.org/activemq/browse/SM-1601?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Freeman Fang updated SM-1601:
-----------------------------
Fix Version/s: 3.2.3
Description:
The servicemix cxf binding component tests print a LOT of output to the
console. This slows down the test build/run significantly. Besides from the
excessive amount of output to the console being not very helpful (because of
the quantity), the build slow-down can actually cause test failures. Trying to
debug a CxfBcRMSequenceTest on aix which passed when launched via mvn on the
command line, but failed when launched from a TeamCity build, it seems that the
excessive logging which was then also pushed from the TeamCity buildagent to
the TeamCity server as the build progressed, caused the individual test cases
in the CxfBcRmSequenceTest to take any where from 2 to 12 minutes to complete
each. Because of the extra time, many of the tests failed as timeouts caused
messages to be resent and duplicate messages were detected.
I've attached a patch which modifies the logging configuration to log much less
output to the console when run with the ibm jdk. (This was the platform on
which the excessive logging seemed to cause the most slow down - to the point
were there were test failures). The pom can very easily be modified though to
log less to the console on all other platforms if so desired.
was:
The servicemix cxf binding component tests print a LOT of output to the
console. This slows down the test build/run significantly. Besides from the
excessive amount of output to the console being not very helpful (because of
the quantity), the build slow-down can actually cause test failures. Trying to
debug a CxfBcRMSequenceTest on aix which passed when launched via mvn on the
command line, but failed when launched from a TeamCity build, it seems that the
excessive logging which was then also pushed from the TeamCity buildagent to
the TeamCity server as the build progressed, caused the individual test cases
in the CxfBcRmSequenceTest to take any where from 2 to 12 minutes to complete
each. Because of the extra time, many of the tests failed as timeouts caused
messages to be resent and duplicate messages were detected.
I've attached a patch which modifies the logging configuration to log much less
output to the console when run with the ibm jdk. (This was the platform on
which the excessive logging seemed to cause the most slow down - to the point
were there were test failures). The pom can very easily be modified though to
log less to the console on all other platforms if so desired.
apply patch on behalf of Peter Jones with thanks
svn commit . -m "[SM-1601]Excessive console output from cxfbc tests"
Sending servicemix-cxf-bc/pom.xml
Sending
servicemix-cxf-bc/src/test/java/org/apache/servicemix/cxfbc/ws/rm/CxfBcRMSequenceTest.java
Adding servicemix-cxf-bc/src/test/resources/logging.properties
Transmitting file data ...
Committed revision 702483.
> Excessive console output from cxfbc tests
> -----------------------------------------
>
> Key: SM-1601
> URL: https://issues.apache.org/activemq/browse/SM-1601
> Project: ServiceMix
> Issue Type: Improvement
> Components: servicemix-cxf-bc
> Reporter: Peter Jones
> Assignee: Freeman Fang
> Fix For: 3.2.3
>
> Attachments: cxfbc-test-logging.patch.txt
>
>
> The servicemix cxf binding component tests print a LOT of output to the
> console. This slows down the test build/run significantly. Besides from the
> excessive amount of output to the console being not very helpful (because of
> the quantity), the build slow-down can actually cause test failures. Trying
> to debug a CxfBcRMSequenceTest on aix which passed when launched via mvn on
> the command line, but failed when launched from a TeamCity build, it seems
> that the excessive logging which was then also pushed from the TeamCity
> buildagent to the TeamCity server as the build progressed, caused the
> individual test cases in the CxfBcRmSequenceTest to take any where from 2 to
> 12 minutes to complete each. Because of the extra time, many of the tests
> failed as timeouts caused messages to be resent and duplicate messages were
> detected.
> I've attached a patch which modifies the logging configuration to log much
> less output to the console when run with the ibm jdk. (This was the platform
> on which the excessive logging seemed to cause the most slow down - to the
> point were there were test failures). The pom can very easily be modified
> though to log less to the console on all other platforms if so desired.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
