-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/50298/#review143147
-----------------------------------------------------------


Fix it, then Ship it!




Thanks! Looks like our validation for this field has the potential to crash the 
master in light of the HTTP API (see below).

This patch can stand on its own, just need to pull out the 0.22.0 compatibility 
removal into a separate patch.


src/master/master.cpp 
<https://reviews.apache.org/r/50298/#comment208886>

    This looks unrelated to the summary / description, did it slip in? This 
looks ok, but how about a separate patch?



src/master/master.cpp (line 6139)
<https://reviews.apache.org/r/50298/#comment208889>

    Given that we also have the new HTTP API, have you audited whether we 
validate that executorinfo has the framework id set in that path?
    
    Looks to me like we need to update the validation function to return an 
Error rather than CHECK, since HTTP frameworks can send this to us:
    
    
https://github.com/apache/mesos/blob/f69a27c43eaf9ade386819dd707ce33041ee1f20/src/master/validation.cpp#L573-L577
    
    Feel free to send in a separate patch with a test (it should be easy to 
test since you can just unit test the validation function).


- Benjamin Mahler


On July 21, 2016, 10:34 p.m., Neil Conway wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/50298/
> -----------------------------------------------------------
> 
> (Updated July 21, 2016, 10:34 p.m.)
> 
> 
> Review request for mesos and Benjamin Mahler.
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Removed backward compat code for old (< 0.15) Mesos agents.
> 
> 
> Diffs
> -----
> 
>   src/master/master.cpp b87d18f457927114bebf6068d08ff0c4f210aece 
> 
> Diff: https://reviews.apache.org/r/50298/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Neil Conway
> 
>

Reply via email to