-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/48370/#review137576
-----------------------------------------------------------


Fix it, then Ship it!





src/slave/containerizer/mesos/isolators/gpu/nvidia.cpp (line 155)
<https://reviews.apache.org/r/48370/#comment202749>

    Looks like we should include what we were trying to do:
    
    ```
        return Error("Failed to obtain device ID for '/dev/nvidiactl': " +
                     device.error());
    ```
    
    Ditto below.


- Benjamin Mahler


On June 11, 2016, 3:06 a.m., Kevin Klues wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/48370/
> -----------------------------------------------------------
> 
> (Updated June 11, 2016, 3:06 a.m.)
> 
> 
> Review request for mesos and Benjamin Mahler.
> 
> 
> Bugs: MESOS-5556
>     https://issues.apache.org/jira/browse/MESOS-5556
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Previously, the major/minor numbers of `/dev/nvidiactl` and
> `/dev/nvidia-uvm` were hard-coded. This actually caused problems for
> `/dev/nvidia-uvm` because its major number is part of the
> "Experimental" device range on Linux.
> 
> Because this range is experimental, there is no guarantee which device
> number will be assigned to it on a given machine. We actually
> encountered this problem in the wild, prompting this change.
> 
> We now use `os:stat::rdev()` to extract the major/minor numbers
> programatically.
> 
> 
> Diffs
> -----
> 
>   src/slave/containerizer/mesos/isolators/gpu/nvidia.hpp 
> 181a2aad97da9ee0f6ffa42cdba9c93dc0077ff7 
>   src/slave/containerizer/mesos/isolators/gpu/nvidia.cpp 
> d7557a0c338e8c0e51461b2326600c03f89c2e8b 
> 
> Diff: https://reviews.apache.org/r/48370/diff/
> 
> 
> Testing
> -------
> 
> GTEST_FILTER="" make -j check && sudo GTEST_FILTER="*NVIDIA*" src/mesos-tests
> 
> 
> Thanks,
> 
> Kevin Klues
> 
>

Reply via email to