GData's TestGdataIndexer.testDestroy() intermittently hits spin loop & causes
build timeout
-------------------------------------------------------------------------------------------
Key: LUCENE-952
URL: https://issues.apache.org/jira/browse/LUCENE-952
Project: Lucene - Java
Issue Type: Bug
Components: Other
Affects Versions: 2.3
Reporter: Michael McCandless
Assignee: Michael McCandless
Several nightly builds (at least #136, #143 and #144) have failed due
to timeout at 45 minutes while running the TestGdataIndexer.testDestroy()
test case.
I tracked it down to this line:
// wait active for the commit
while(this.indexer.writer != null){}
Intermittently, that while loop will spin forever. I can only get the
failure to happen on Linux: it doesn't happen on Mac OS X (haven't
tried windows). The nightly build runs on Solaris 10, so it also
happens there.
It turns out, this is due to the fact that "writer" is not declared as
"volatile". This is because one thread is closing the indexer, which
sets writer to null, but another thread is running the while loop.
If this.indexer.writer was set to null before that while loop starts,
the test will run through fine; else, it won't.
I plan to fix this by adding this method to GDataIndexer class:
// Used only for testing
protected synchronized IndexWriter getWriter() {
return this.writer;
}
and changing unit test to call that method.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
