[
https://issues.apache.org/jira/browse/OOZIE-2584?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15354746#comment-15354746
]
Peter Bacsko commented on OOZIE-2584:
-------------------------------------
I was thinking - if the machine is very slow, 1000ms of timeout in
testTimeoutWaitingWriteLock can be a problem.
If it takes too much time to execute
{code}
l1.proceed();
l1.awaitTermination();
{code}
then L2 might time out. I will increase the value to 10000 - that really should
be enough. Same applies to testWriteLockSameThreadWait.
> Eliminate Thread.sleep() calls in TestMemoryLocks
> -------------------------------------------------
>
> Key: OOZIE-2584
> URL: https://issues.apache.org/jira/browse/OOZIE-2584
> Project: Oozie
> Issue Type: Bug
> Reporter: Peter Bacsko
> Assignee: Peter Bacsko
> Priority: Minor
> Attachments: OOZIE-2584-001.patch
>
>
> There are a lot of Thread.sleep() calls in TestMemoryLocks.
> For example:
> {code}
> new Thread(l1).start();
> Thread.sleep(500);
> new Thread(l2).start();
> Thread.sleep(500);
> {code}
> This solution is very fragile. On a slower build machine, Thread2 might
> actually start earlier than Thread1, causing test failure.
> Eliminate sleeps and use a more stable approach for start/stop coordination
> (possibly CountDownLatches).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
