JUnit test for channel performance and ordering - long running tests
--------------------------------------------------------------------
Key: HEDERA-37
URL: https://forge.continuent.org/jira/browse/HEDERA-37
Project: Hedera
Type: Sub-task
Environment: JUnit4
Reporter: Linas Virbalas
Assigned to: Robert Hodges
1. We need long running tests with so many messages that it would take, for
example, 10 minutes to commute. If the GC layer isn't garbage collecting
properly it would show up in this type of test. It may be worthwhile to make
sure performance does not degrade over a period of time.
1.1. Implementation specification:
Test case: 2 members send each other n messages at first. Time is taken.
Than another test runs with also 2 members sending each other N>>n
messages. Time is taken and compared to the first test's time. Test case is
said to be successful if the elapsed time of the second test is
(approximately) proportional to the first's time or faster. This is useful
to assure that messages in the "big" test are being sent approximately at
the same speed as in the "small" test. It would automatically fail, if the
speed
would degrade over time.
2. Also this issue completely redesigns
ChannelPerformanceMultithreadedTestRunner class in order to completely wrap the
ChannelPerformanceClass by using a:
2.1. custom ThreadPoolExecutor named CatchyThreadPoolExecutor to be able to
completely timeout and shutdown foreign (ChannelPerformanceTest created)
threads;
2.2. custom UncaughtExceptionHandler named ThreadExceptionHandler to be aware
of exceptions that are happening in foreign threads;
2.3. and a custom SecurityManager named ExitHandlingSecurityManager to disallow
exits from JVM called from foreign threads.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
https://forge.continuent.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
http://www.atlassian.com/software/jira
_______________________________________________
Hedera mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/hedera