You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mesos.apache.org by Brenden Matthews <br...@diddyinc.com> on 2013/08/02 03:46:25 UTC

Review Request 13211: Fixed bug where active TaskTrackers are killed.

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/13211/
-----------------------------------------------------------

Review request for mesos.


Repository: mesos-git


Description
-------

Fixed bug where active TaskTrackers are killed.

Sometimes, TaskTrackers which are active will fail to get marked as such
because the scheduler is either a) too busy or b) has not received any
offers in a while.  This resolves the problem of TaskTrackers getting
killed by the launch timeout even though they are alive and well.

Furthermore, some parts of the scheduler don't need to be completely
synchronized because they don't modify shared data.


Diffs
-----

  hadoop/mesos/src/java/org/apache/hadoop/mapred/MesosScheduler.java 1923d498121f1ff4e875d9685ed66de2b7fb22ec 

Diff: https://reviews.apache.org/r/13211/diff/


Testing
-------

Used in production at Airbnb.

make check && cd hadoop && make hadoop-2.0.0-mr1-cdh4.2.1


Thanks,

Brenden Matthews


Re: Review Request 13211: Fixed bug where active TaskTrackers are killed.

Posted by Brenden Matthews <br...@diddyinc.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/13211/
-----------------------------------------------------------

(Updated Aug. 2, 2013, 1:52 a.m.)


Review request for mesos.


Changes
-------

Added dependency.


Repository: mesos-git


Description
-------

Fixed bug where active TaskTrackers are killed.

Sometimes, TaskTrackers which are active will fail to get marked as such
because the scheduler is either a) too busy or b) has not received any
offers in a while.  This resolves the problem of TaskTrackers getting
killed by the launch timeout even though they are alive and well.

Furthermore, some parts of the scheduler don't need to be completely
synchronized because they don't modify shared data.


Diffs
-----

  hadoop/mesos/src/java/org/apache/hadoop/mapred/MesosScheduler.java 1923d498121f1ff4e875d9685ed66de2b7fb22ec 

Diff: https://reviews.apache.org/r/13211/diff/


Testing
-------

Used in production at Airbnb.

make check && cd hadoop && make hadoop-2.0.0-mr1-cdh4.2.1


Thanks,

Brenden Matthews