[
https://issues.apache.org/jira/browse/DERBY-5601?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Kristian Waagan updated DERBY-5601:
-----------------------------------
Attachment: spawnedprocess_extended.diff
Attaching an extended version of SpawnedProcess, which isn't suspectible to
interrupts when waiting for the process to complete.
I've let the uses of SpawnedProcess.complete behave as earlier, so if the
interrupt seen earlier stopped us from waiting/destroying a process the test
may now hang for 45 minutes before killing the process. For testing purposes
you can set this to a lower value (use milliseconds) with
derby.tests.process.killThreshold. This value must be higher than the time it
takes to run the longest test using an external process, otherwise the process
will be terminated prematurely. On a hang you should be able to use jstack to
see where the code is waiting/stuck.
I've run suites.All with the patch on Windows Vista without errors, but there
were no process-related issues in that run anyway.
> Windows hang in NativeAuthenticationServiceTest testAll with prior interrupt
> error in SecureServerTest and other cascade errors and failures
> --------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: DERBY-5601
> URL: https://issues.apache.org/jira/browse/DERBY-5601
> Project: Derby
> Issue Type: Bug
> Components: Services, Test
> Affects Versions: 10.9.0.0
> Reporter: Kathey Marsden
> Priority: Blocker
> Attachments: runsteall.out, spawnedprocess_extended.diff
>
>
> The IBM Windows runs on trunk have been hanging in
> NativeAuthenticationServiceTest. This is preventing tests from continuing or
> being posted. The actual test run hangs at :
> ....
> emb)lang.AutoIncrementTest.test_4437_03_deferredInsert used 218 ms .
> (emb)lang.HalfCreatedDatabaseTest.test_4589 used 6859 ms .
> (emb)lang.HalfCreatedDatabaseTest.test_5526 used 16 ms .
> (emb)lang.NativeAuthenticationServiceTest.testAll used 24061 ms .
> (emb)lang.NativeAuthenticationServiceTest.testAll used 17781 ms .
> (emb)lang.NativeAuthenticationServiceTest.testAll used 11187 ms .
> (net)lang.NativeAuthenticationServiceTest.testAll used 18187 ms .
> (net)lang.NativeAuthenticationServiceTest.testAll
> and so no test continue and no results get posted to:
> http://people.apache.org/~myrnavl/derby_test_results/main/windows/testSummary-1239981.html
> The last run where this did not happen was:
> http://people.apache.org/~myrnavl/derby_test_results/main/windows/testSummary-1236968.html
> The actual command being executed is:
> 1 4748 2676 0 20:23:24 CONIN$ 46:30 C:\WINDOWS\system32\java.exe
> -Dder
> byTesting.ldapServer=socket.usca.ibm.com -DderbyTesting.ldapPort=389
> -DderbyTest
> ing.dnString=oakland.ibm.com -DderbyTesting.ldapUser=mamta
> -DderbyTesting.ldapPa
> ssword=xxxxx -Xmx512M -Dderby.tests.networkServerStartTimeout=480 -Dderby.te
> sts.basePort=3100 -Dderby.tests.trace=true
> -DderbyTesting.oldReleasePath=d:/jart
> est/releases_for_upgrade_tests junit.textui.TestRunner
> org.apache.derbyTesting.f
> unctionTests.suites.All
> I couldn't reproduce this running the single test on my machine. Marking
> this blocker as the Oracle Nightlies seem to be having serious issues too
> that date back further.
> http://dbtg.foundry.sun.com/derby/test/Daily/jvm1.6/testing/Limited/
> Right now we have no reliable regression testing going on.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
