mesos-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Bernd Mathiske" <be...@mesosphere.io>
Subject Re: Review Request 39276: Fixed a bug in which under certains circumstances HTTP 1.1 Pipelining is not respected.
Date Wed, 21 Oct 2015 15:42:05 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/39276/#review103410
-----------------------------------------------------------



3rdparty/libprocess/include/process/event.hpp (line 123)
<https://reviews.apache.org/r/39276/#comment161467>

    Suggestion: name this parameter "responseFuture"?
    
    (Below, "promise" and "future" is short, but it carries too little meaning to follow the
code easily without looking up and memorizing declarations.)



3rdparty/libprocess/src/process.cpp (line 216)
<https://reviews.apache.org/r/39276/#comment161470>

    Aren't we making a copy here, too, now? Let's add that comment here!



3rdparty/libprocess/src/process.cpp (line 1034)
<https://reviews.apache.org/r/39276/#comment161468>

    Why wrap request? This is already of the right type. And the incoming parameter is of
the exact same const ref type as the expected argument.
    
    Copying happens all the way down in this call chain.


- Bernd Mathiske


On Oct. 21, 2015, 6:18 a.m., Alexander Rojas wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/39276/
> -----------------------------------------------------------
> 
> (Updated Oct. 21, 2015, 6:18 a.m.)
> 
> 
> Review request for mesos, Anand Mazumdar, Bernd Mathiske, and Till Toenshoff.
> 
> 
> Bugs: MESOS-3705
>     https://issues.apache.org/jira/browse/MESOS-3705
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> When using the same socket to send multiple HTTP requests to different actors. If the
actor responsible for handling the first request is stuck handling another event while a subsequent
request can reply immediatly, the order of the responses is altered violating HTTP Pipelining.
> 
> This patch fixes that problem enforcing that every response is sent in the order the
corresponding request arrived. It also adds a test to reproduce the issue and verify the fix
works.
> 
> 
> Diffs
> -----
> 
>   3rdparty/libprocess/include/process/event.hpp 16ddbd77afa6efdf6bad201aa497ee102aa863ae

>   3rdparty/libprocess/src/process.cpp 954d31424bc8f8ecfa78b80513c480f2514ec271 
>   3rdparty/libprocess/src/tests/http_tests.cpp d13d3888abbf3db552df4a9f83e54667e598ded9

> 
> Diff: https://reviews.apache.org/r/39276/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Alexander Rojas
> 
>


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