mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ben Mahler" <benjamin.mah...@gmail.com>
Subject Re: Review Request 42154: Moved push_back to end of removeFramework.
Date Tue, 12 Jan 2016 01:20:52 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/42154/#review113893
-----------------------------------------------------------

Ship it!


Thank you!

- Ben Mahler


On Jan. 11, 2016, 9:41 p.m., Kevin Klues wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/42154/
> -----------------------------------------------------------
> 
> (Updated Jan. 11, 2016, 9:41 p.m.)
> 
> 
> Review request for mesos and Ben Mahler.
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Previously, the framework being removed was pushed back to the circular
> buffer of completed frameworks as a shared_ptr halfway through the
> computation done to remove a framework from the master. The pointer to
> the framework was then accessed again after this point.
> 
> Pushing a shared_ptr this way is fine so long as the push_pack succeeds
> and the pointer is actually inserted into the buffer. However, with the
> introduction of the new flags to set the size of this buffer via
> max_completed_frameworks, it's possible for the size of this buffer to
> be 0. As such, the circular buffer was taking control of the pointer,
> noticing there was nowhere to push it, and then freeing it.  This causes
> problems since the pointer is then accessed later on.
> 
> This patch moves the insertion of the shared pointer into the circular
> buffer to the end of the removeFramework() call. This change should not
> have any adverse effects.
> 
> Review: https://reviews.apache.org/r/42154
> 
> 
> Diffs
> -----
> 
>   src/master/master.cpp 2d9b7f9540574aa3ef9e5af3b2b8922dffeebac8 
> 
> Diff: https://reviews.apache.org/r/42154/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Kevin Klues
> 
>


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