-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/34720/#review87767
-----------------------------------------------------------



src/slave/slave.hpp
<https://reviews.apache.org/r/34720/#comment140194>

    The 'reason' is for the slave to encode the reason  behind a terminal 
status update for those pending/unterminated tasks when executor is terminated.



src/slave/slave.cpp
<https://reviews.apache.org/r/34720/#comment140186>

    Do you want to introduce a configurable minimal interval between two 
'corrections()' pull? (like we did for resource estimator)?
    
    You probably want to follow the implementation of `fowardOversubscribed` 
and `_forwardOversubscribed`.



src/slave/slave.cpp
<https://reviews.apache.org/r/34720/#comment140192>

    " because the framework cannot be found"



src/slave/slave.cpp
<https://reviews.apache.org/r/34720/#comment140189>

    indent



src/slave/slave.cpp
<https://reviews.apache.org/r/34720/#comment140191>

    "is not running" is confusing. How about:
    
    "because the executor cannot be found"



src/slave/slave.cpp
<https://reviews.apache.org/r/34720/#comment140193>

    Can you add a TODO here saying that we may want to  check if containerId 
matches or not due to race (i.e., qos controller wants to kill container 1 of 
executor 1, but the executor you are checking here is for container 2 of 
executor 1).
    
    That means we need to add ContainerID in ResourceUsage/QoSCorrection as well



src/slave/slave.cpp
<https://reviews.apache.org/r/34720/#comment140190>

    "because the executor is " << executor->state



src/slave/slave.cpp
<https://reviews.apache.org/r/34720/#comment140195>

    overrides `executor->reason`



src/slave/slave.cpp
<https://reviews.apache.org/r/34720/#comment140196>

    infer ... from 'killed' field in 'termination'.


- Jie Yu


On June 12, 2015, 9:48 p.m., Niklas Nielsen wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/34720/
> -----------------------------------------------------------
> 
> (Updated June 12, 2015, 9:48 p.m.)
> 
> 
> Review request for mesos, Bartek Plotka, Jie Yu, Joris Van Remoortere, and 
> Vinod Kone.
> 
> 
> Bugs: MESOS-2653
>     https://issues.apache.org/jira/browse/MESOS-2653
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> See summary
> 
> 
> Diffs
> -----
> 
>   src/slave/slave.hpp 0df1b55791963fb4159b7ea5318d09dde4f7d8c7 
>   src/slave/slave.cpp 9af69d8f0b28c9441c684886c52320378f9b2869 
> 
> Diff: https://reviews.apache.org/r/34720/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> 
> Thanks,
> 
> Niklas Nielsen
> 
>

Reply via email to