Skip to content

[SPARK-2687] [yarn]amClient should remove ContainerRequest#3245

Closed
lianhuiwang wants to merge 1 commit intoapache:masterfrom
lianhuiwang:amclient-remove-request-2
Closed

[SPARK-2687] [yarn]amClient should remove ContainerRequest#3245
lianhuiwang wants to merge 1 commit intoapache:masterfrom
lianhuiwang:amclient-remove-request-2

Conversation

@lianhuiwang
Copy link
Contributor

in https://issues.apache.org/jira/browse/YARN-1902, after receving allocated containers,if amClient donot remove ContainerRequest,RM will continually allocate container for spark AppMaster.@tgravescs

@SparkQA
Copy link

SparkQA commented Nov 13, 2014

Test build #23311 has started for PR 3245 at commit 4749908.

  • This patch merges cleanly.

@SparkQA
Copy link

SparkQA commented Nov 13, 2014

Test build #23311 has finished for PR 3245 at commit 4749908.

  • This patch passes all tests.
  • This patch merges cleanly.
  • This patch adds no public classes.

@AmplabJenkins
Copy link

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/23311/
Test PASSed.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

so even though the preferred node locality stuff isn't working right now I think I would rather see us attempt to do the right thing here. If there weren't any preferred nodes specified then taking any of the requests works just fine, but if there were host or rack specific requests we should remove the ones that were actually allocated to us.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

i donot think so. because it is FIFO, we can remove head of request queue because head of queue is allocated by Yarn when we receive allocated Containers.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

When you request containers from yarn and specify hosts or racks then it is not guaranteed that you get them back in the order you requests them. Certain hosts may be busy, not available, etc. You might get a rack request fullfilled instead of a host request, etc. Since the preferred locality stuff is broken right now anyway (SPARK-2089), we can fix this up separately if you would prefer?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yes,i know what you mean. preferred locality donot work in our application. i will take a look at it.thanks.

@andrewor14
Copy link
Contributor

Hey @lianhuiwang by the way the YARN alpha support is now dropped. It would be good if you can rebase the changes here to master to resolve the merge conflicts manually.

@sryza
Copy link
Contributor

sryza commented Dec 22, 2014

Just noticed this. My PR for SPARK-1714 #3765 also fixes this issue.

@lianhuiwang
Copy link
Contributor Author

ok, i will close this PR.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

6 participants