mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Timothy Chen" <tnac...@apache.org>
Subject Re: Review Request 41078: Fixed tests to call socket accept before sending response.
Date Tue, 08 Dec 2015 07:20:19 GMT


> On Dec. 8, 2015, 7:08 a.m., Timothy Chen wrote:
> > Can you elaborate more without this fix what happened and why?
> 
> Jojy Varghese wrote:
>     This fix moves "Accept" call before sending a response back to the client so that
when the client sends the next request, the server is ready to accept it. Before this fix,
there was a race condition between server sending the response and client sending the next
request and server missing that request because the "accept" was not called yet.

Sounds very fishy to me, we don't do this in Mesos itself as we take multiple requests in
flight. What's the difference here? And is there something we should do so others don't have
to worry about this?


- Timothy


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/41078/#review109288
-----------------------------------------------------------


On Dec. 8, 2015, 7:03 a.m., Jojy Varghese wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/41078/
> -----------------------------------------------------------
> 
> (Updated Dec. 8, 2015, 7:03 a.m.)
> 
> 
> Review request for mesos and Timothy Chen.
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> By calling accept before sending a response, the server would be ready for
> client's next request.
> 
> 
> Diffs
> -----
> 
>   src/tests/containerizer/provisioner_docker_tests.cpp c63bf53fee40ef12536a16e11f4d5224c4e4278e

> 
> Diff: https://reviews.apache.org/r/41078/diff/
> 
> 
> Testing
> -------
> 
> make check.
> 
> 
> Thanks,
> 
> Jojy Varghese
> 
>


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