-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/52746/#review153211
-----------------------------------------------------------


Fix it, then Ship it!




Wonder if we could use a helper that returns LOST or DROPPED state based on 
Frameworkinfo instead of duplicating that code everywhere. Not sure what to 
call it though.


src/slave/slave.cpp (line 3243)
<https://reviews.apache.org/r/52746/#comment222501>

    TODO at the end sounds weird.
    
    Lately we have been putting blank lines in front of TODOs for better 
formatting.
    
    ```
    // update for...
    //
    // TODO(vinod): Consider...
    //..
    //..
    //
    // TODO(vinod): Use foreachvalue...
    
    ```



src/slave/slave.cpp (line 3558)
<https://reviews.apache.org/r/52746/#comment222502>

    new line.



src/slave/slave.cpp (line 3561)
<https://reviews.apache.org/r/52746/#comment222503>

    new line.


- Vinod Kone


On Oct. 12, 2016, 7:35 p.m., Neil Conway wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/52746/
> -----------------------------------------------------------
> 
> (Updated Oct. 12, 2016, 7:35 p.m.)
> 
> 
> Review request for mesos and Vinod Kone.
> 
> 
> Bugs: MESOS-6332
>     https://issues.apache.org/jira/browse/MESOS-6332
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> If the agent cannot launch a task due to a variety of possible error
> conditions, we now send TASK_DROPPED to partition-aware frameworks
> rather than TASK_LOST.
> 
> 
> Diffs
> -----
> 
>   src/slave/slave.cpp 6bd9b49c3bbdb973a0d03552ae8fe55b33371083 
> 
> Diff: https://reviews.apache.org/r/52746/diff/
> 
> 
> Testing
> -------
> 
> `make check`.
> 
> Note that none of these code paths appear to have unit tests, so I couldn't 
> update the unit tests to ensure that `TASK_DROPPED` is sent appropriately. We 
> should add unit tests for at least some of these situations, if feasible -- 
> but I'll defer that for now.
> 
> 
> Thanks,
> 
> Neil Conway
> 
>

Reply via email to