You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mesos.apache.org by Ben Mahler <be...@gmail.com> on 2014/05/22 00:24:24 UTC

Review Request 21788: Reduced acknowledgment delay in the scheduler driver.

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/21788/
-----------------------------------------------------------

Review request for mesos, Benjamin Hindman, Jie Yu, and Vinod Kone.


Repository: mesos-git


Description
-------

We currently 'defer' in order to acknowledge processed status updates in the scheduler driver.

This is unfortunate because it adds unnecessary delay and unnecessary retries, especially when the scheduler's queue is backed up.

It is sufficient to use 'aborted' to see if we can send a status update. Ideally there would be proper locking on this variable rather than the use of 'volatile', but since we're mostly concerned about the driver aborting on the same thread as the status update was processed, I've left it as is.


Diffs
-----

  src/sched/sched.cpp 3684cfe9d153cab5f4ea86094fffd814ce74baa1 

Diff: https://reviews.apache.org/r/21788/diff/


Testing
-------

make check


Thanks,

Ben Mahler


Re: Review Request 21788: Reduced acknowledgment delay in the scheduler driver.

Posted by Jie Yu <yu...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/21788/#review43727
-----------------------------------------------------------

Ship it!


Ship It!

- Jie Yu


On May 21, 2014, 10:24 p.m., Ben Mahler wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/21788/
> -----------------------------------------------------------
> 
> (Updated May 21, 2014, 10:24 p.m.)
> 
> 
> Review request for mesos, Benjamin Hindman, Jie Yu, and Vinod Kone.
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> We currently 'defer' in order to acknowledge processed status updates in the scheduler driver.
> 
> This is unfortunate because it adds unnecessary delay and unnecessary retries, especially when the scheduler's queue is backed up.
> 
> It is sufficient to use 'aborted' to see if we can send a status update. Ideally there would be proper locking on this variable rather than the use of 'volatile', but since we're mostly concerned about the driver aborting on the same thread as the status update was processed, I've left it as is.
> 
> 
> Diffs
> -----
> 
>   src/sched/sched.cpp 3684cfe9d153cab5f4ea86094fffd814ce74baa1 
> 
> Diff: https://reviews.apache.org/r/21788/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Ben Mahler
> 
>


Re: Review Request 21788: Reduced acknowledgment delay in the scheduler driver.

Posted by Vinod Kone <vi...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/21788/#review43670
-----------------------------------------------------------

Ship it!


Ship It!

- Vinod Kone


On May 21, 2014, 10:24 p.m., Ben Mahler wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/21788/
> -----------------------------------------------------------
> 
> (Updated May 21, 2014, 10:24 p.m.)
> 
> 
> Review request for mesos, Benjamin Hindman, Jie Yu, and Vinod Kone.
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> We currently 'defer' in order to acknowledge processed status updates in the scheduler driver.
> 
> This is unfortunate because it adds unnecessary delay and unnecessary retries, especially when the scheduler's queue is backed up.
> 
> It is sufficient to use 'aborted' to see if we can send a status update. Ideally there would be proper locking on this variable rather than the use of 'volatile', but since we're mostly concerned about the driver aborting on the same thread as the status update was processed, I've left it as is.
> 
> 
> Diffs
> -----
> 
>   src/sched/sched.cpp 3684cfe9d153cab5f4ea86094fffd814ce74baa1 
> 
> Diff: https://reviews.apache.org/r/21788/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Ben Mahler
> 
>