-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/50258/
-----------------------------------------------------------
Review request for mesos, Benjamin Mahler, Kevin Klues, Yubo Li, and Vikrama
Ditya.
Bugs: MESOS-5795
https://issues.apache.org/jira/browse/MESOS-5795
Repository: mesos
Description
-------
This change is to account for allocated GPU devices during docker recovery
invoked as part of slave recovery.
On docker container recovery, the list of devices is taken from the docker
inspect output. The nvidia
devices of the format /dev/nvidia# are further narrowed down to get the minor
number of the GPU.
The allocation module is invoked to account for the GPUs used by the recovered
containers.
Note: The next review request #50259 builds on top of this and uses the
abstraction provided by docker inspect
Diffs
-----
src/slave/containerizer/docker.hpp 43ca4317d608b3b43dd7bd0d1b55c721e7364885
src/slave/containerizer/docker.cpp f1ecf3b25d85597f6c3dcaa47968860ed119dbd5
src/tests/containerizer/nvidia_gpu_isolator_tests.cpp
7944c0da716b2438fa2b40f5504aefa346e31046
Diff: https://reviews.apache.org/r/50258/diff/
Testing
-------
GTEST_FILTER="NvidiaGpuDockerContainerizerTest.ROOT_DOCKER_LaunchWithGpuRecovery"
make -j check
Thanks,
Rajat Phull
