sv2000 opened a new pull request #2940: GOBBLIN-1099: Handle orphaned Yarn containers in Gobblin-on-Yarn clus… URL: https://github.com/apache/incubator-gobblin/pull/2940 …ters Dear Gobblin maintainers, Please accept this PR. I understand that it will not be reviewed until I have checked off all the steps below! ### JIRA - [x] My PR addresses the following [Gobblin JIRA](https://issues.apache.org/jira/browse/GOBBLIN/) issues and references them in the PR title. For example, "[GOBBLIN-XXX] My Gobblin PR" - https://issues.apache.org/jira/browse/GOBBLIN-1099 ### Description - [x] Here are some details about my PR, including screenshots (if applicable): A Yarn application may leave behind orphaned containers, which can happen due to lost node managers. The orphaned containers however can continue to run (potentially forever) as participants in the Helix cluster. This can cause the following problems for a Gobblin-on-Yarn application: Double publish of data and commit of state Task failures and partition starvation during application restarts, as Helix may assign tasks to the orphaned containers which have a stale state and configuration Container failures on application restarts due to Helix instance name collisions with orphaned containers This PR incorporates the following changes to handle orphaned containers: 1. Disables live instances during Yarn application start up and shutdown in GobblinYarnAppLauncher. This ensures that any orphaned Helix instances are disabled from joining the cluster on restart and any tasks running on these instances are cancelled. 2. The GobblinApplicationMaster (inside YarnService) ensures that Helix instance name assigned to a newly allocated container is not a currently live instance to avoid instance name collisions. We also handle NM failures/restarts that result in Yarn RM "aborting" the container (i.e. container is deemed dead from Yarn's point of view, even though the container may physically be alive). In this case, we disable the instance to ensure it is fenced off from the Helix cluster. 3. Gobblin workers (i.e. GobblinYarnTaskRunner) retry in case of failure in joining a Helix cluster. The retry logic disconnects from the Helix cluster and drops the instance before reattempting to join the cluster. ### Tests - [x] My PR adds the following unit tests __OR__ does not need testing for this extremely good reason: Added unit test in GobblinTaskRunnerTest ### Commits - [x] My commits all reference JIRA issues in their subject lines, and I have squashed multiple commits if they address the same issue. In addition, my commits follow the guidelines from "[How to write a good git commit message](http://chris.beams.io/posts/git-commit/)": 1. Subject is separated from body by a blank line 2. Subject is limited to 50 characters 3. Subject does not end with a period 4. Subject uses the imperative mood ("add", not "adding") 5. Body wraps at 72 characters 6. Body explains "what" and "why", not "how"
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
