-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/55901/#review162922
-----------------------------------------------------------
Patch looks great!
Reviews applied: [55899, 55900, 55901]
Passed command: export OS='ubuntu:14.04' BUILDTOOL='autotools' COMPILER='gcc' CONFIGURATION='--verbose'
ENVIRONMENT='GLOG_v=1 MESOS_VERBOSE=1'; ./support/docker-build.sh
- Mesos Reviewbot
On Jan. 25, 2017, 12:33 a.m., Gastón Kleiman wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/55901/
> -----------------------------------------------------------
>
> (Updated Jan. 25, 2017, 12:33 a.m.)
>
>
> Review request for mesos, Alexander Rukletsov, Anand Mazumdar, haosdent huang, and Vinod
Kone.
>
>
> Bugs: MESOS-6280
> https://issues.apache.org/jira/browse/MESOS-6280
>
>
> Repository: mesos
>
>
> Description
> -------
>
> [WIP] Added support for command health checks to the default executor.
>
> This is still a work in progress, but I am posting it in order to get some feedback on
the general approach.
>
> Open questions/issues:
>
> - The code that handles communication with the Agent is on the `DefaultExecutor`, so
the health checker calls `evolve` when serializing the API calls. Should it use the v1 protos
instead, and skip the `evolve` call?
> - CMD health checks for command tasks inherit the env from the executor, but allow
overriding env variables. I don't think that inheriting the executor env makes sense for cmd
health checks launched by the `DefaultExecutor` - each task can have its own env, so the health
check command should inherit the task env and merge it with the one in the check's `CommandInfo`.
> I added some code that takes the task env from `TaskInfo` and merges it check's env,
but I don't think that this is an ideal approach, since it will miss env variables set/modified
by containerizers/hooks/isolators. Do you think that it would make sense for all Debug Containers
to inherit the environment of the parent container?
> - There's a TODO for stopping/resuming health checking when the `DefaultExecutor` is
disconnected from the agent. After talking to AlexR, I believe that we should do this for
all types of checks, not just for `CMD`.
> - The test that I introduced passes on Linux, but not on macOS, I still have to do
some more debugging.
>
>
> Diffs
> -----
>
> src/checks/health_checker.hpp 6e558f2061a9e31157c47d31cb64b3a8568aace3
> src/checks/health_checker.cpp 50aa2858e807b27bbab58a3618f5200cfe4eca9e
> src/launcher/default_executor.cpp a03794934adb93868734f8cf00b337a1bff9b5ab
> src/tests/health_check_tests.cpp debbd3c09b7555145aaf3f62a24d795d1423a269
>
> Diff: https://reviews.apache.org/r/55901/diff/
>
>
> Testing
> -------
>
> Introduced a new test: `HealthCheckTest.DefaultExecutorCmdHealthCheck`. It passes on
Linux, but not on macOS.
>
>
> Thanks,
>
> Gastón Kleiman
>
>
|