aurora-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Kevin Sweeney" <kevi...@apache.org>
Subject Re: Review Request 26300: Don't kill GC Executor after period of inactivity
Date Wed, 08 Oct 2014 18:56:53 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/26300/
-----------------------------------------------------------

(Updated Oct. 8, 2014, 11:56 a.m.)


Review request for Aurora, Ben Mahler, Vinod Kone, and Brian Wickman.


Changes
-------

rebase


Bugs: AURORA-788
    https://issues.apache.org/jira/browse/AURORA-788


Repository: aurora


Description
-------

The GC executor is configured to exit after 15 minutes of inactivity. This leads to a race
where the mesos slave gets a launchTask message for a GC executor just as the executor has
exited, causing TASK_LOST noise. This also increases the risk that a slave will lose its GC
executor due to the scheduler not being able to find a slot for it (since GC executors will
have a higher churn rate).

Cluster operators will still be able to deploy new versions of the GC executor as the 24-hour
max lifetime limit is still in place. This patch only removes the inactivity limit.


Diffs (updated)
-----

  src/main/python/apache/aurora/executor/gc_executor.py 788671e32d2f995b954e906d8866019ed66c970d

  src/test/python/apache/aurora/executor/test_gc_executor.py 1905fe35de31e667f499f7945952f04dc13aac06


Diff: https://reviews.apache.org/r/26300/diff/


Testing
-------

./pants src/test/python/apache/aurora/executor:gc_executor


Thanks,

Kevin Sweeney


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message