[
https://issues.apache.org/jira/browse/HADOOP-6194?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Steve Loughran updated HADOOP-6194:
-----------------------------------
Attachment: HADOOP-6194.patch
This is an updated patch which
* uses the name LifecycleService for the class, LifecycleServiceWithWorkers for
the subclass intended for services with workers.
* added much more javadocs and even some more test cases to the test
I'll stick up the HDFS services tomorrow; I will synchronise and run all the
tests. The big changes there are primarily in adding more shutdown logic to
remove all threads; this could be teased out to make the lifecycle patch look
smaller.
I have a patch for the JT and TT too; the only troublespot there is in startup,
where I believe it can block forever waiting for the filesystem to go live in
startup. There's also MAPREDUCE-437. These are separate issues from how you
start and stop the things, just needed to make sure that when you start or stop
them they clean up and don't ever block waiting for some external dependency
like a functional filesystem.
> Add service base class and tests to hadoop-common/util
> ------------------------------------------------------
>
> Key: HADOOP-6194
> URL: https://issues.apache.org/jira/browse/HADOOP-6194
> Project: Hadoop Common
> Issue Type: New Feature
> Components: util
> Affects Versions: 0.21.0
> Reporter: Steve Loughran
> Assignee: Steve Loughran
> Fix For: 0.22.0
>
> Attachments: HADOOP-6194-1.patch, HADOOP-6194.patch, HADOOP-6194.patch
>
>
> For a service base class, we need to add to common
> # The Service class
> # A mock Service class (it's not in test, its in src/, because it helps in
> functional tests downstream)
> # Unit tests
> This is the issue to cover this work
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
