mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gilbert Song <songzihao1...@gmail.com>
Subject Review Request 65139: Reverted "Fixed `wait()` and `destroy()` in composing containerizer.".
Date Sat, 13 Jan 2018 00:15:17 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/65139/
-----------------------------------------------------------

Review request for mesos, Andrei Budnik, Alexander Rukletsov, Benjamin Mahler, Jie Yu, and
Vinod Kone.


Bugs: MESOS-8391
    https://issues.apache.org/jira/browse/MESOS-8391


Repository: mesos


Description
-------

This reverts commit 95decd404438abd422794524e01d72a889821566.

There are two reasons to revert this commit:
  1. After the agent recovers, if the nested containers that are
     launched beforehand are killed, they will no longer be updated
     with new status, because the `WAIT_NESTED_CONTAINER` call from
     the default executor will end up with a future forever. Please
     see MESOS-8391 for details.
  2. The original commit makes the composing containerizer wait()
     and destroy() rely on the same future of a ContainerTermination
     promise. This would get into the bug that composing containerizer
     destroy() may fail due to the wait() future got discarded.
     Need to protect it by using `undiscardable()`. Please see
     MESOS-7926 for details.


Diffs
-----

  src/slave/containerizer/composing.cpp 9ace70d9fbd78182715c5ef13fcaf7ad45f76f97 


Diff: https://reviews.apache.org/r/65139/diff/1/


Testing
-------

make check


Thanks,

Gilbert Song


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