[
https://issues.apache.org/jira/browse/TEZ-4149?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17109465#comment-17109465
]
László Bodor commented on TEZ-4149:
-----------------------------------
[~jeagles]: thanks for taking a look
_"I think there is still value in being able to start hdfs and yarn separately,
but I like that they are wrapped in a class like this. We can delay that
decision until a test case needs one and not the other."_
Yeah, I was thinking about that. There would be an optional "useHdfs(boolean)"
kind of method before calling "start()" to tell the wrapper whether to hdfs or
not (default: true). I'm including that into the next patch. I'm about to
create a separate jira for adopting this wrapper to all unit tests, and then we
can see if we need any additional APIs for the wrapper.
_What is the cost/benefit of using the runner factory as well at the Waiter
class. It's a little messy, but may be worth it is limited situations like
this._
I agree, it might look messy, both of them are used because this way I was able
to share a common multithreaded runner logic (runWithShutDownConditions)
between the 3 unit test methods in this class. Waiters cannot be eliminated as
we assert inside the testing methods, and they run in subthreads from now on,
we must not miss any assertion failures from those threads (if a subthread
called the original Assert.assertXYZ, and failed, it would never propagate to
the main thread, only through waiter.assertXYZ). Let me know if we can do this
less mess, otherwise I would go on with this approach.
_Let's fix the ASF license header on MiniClusterGroup_
I'll take care of it in the next patch.
> Speed up TezRecovery tests
> --------------------------
>
> Key: TEZ-4149
> URL: https://issues.apache.org/jira/browse/TEZ-4149
> Project: Apache Tez
> Issue Type: Sub-task
> Reporter: Jonathan Turner Eagles
> Assignee: László Bodor
> Priority: Major
> Attachments: TEZ-4149.01.patch, TEZ-4149.WIP.patch,
> org.apache.tez.test.TestRecovery-output.txt
>
>
> Currently, approximately 50% of the tests cases are chosen to run as there
> are many failure points chosen to test recovery on.
> This can lead to the introduction of bugs into the code as not all test cases
> are run for every Tez QA run.
> In addition, this can be a real development bottleneck as tests take around
> 20 minutes per cycle if all tests are run (10 minutes if 50% of the tests are
> run as usual)
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
