move cgroup requirements to planned additions

also document that the executor will only try to remove all the children in a
best-effort way.

Docker-DCO-1.1-Signed-off-by: Daniel, Dao Quang Minh <dqminh89@gmail.com> (github: dqminh)
This commit is contained in:
Daniel, Dao Quang Minh
2014-12-11 13:59:10 -05:00
parent e9c1b0144a
commit eb2ae34c80
+4 -2
View File
@@ -333,6 +333,8 @@ the container is unpaused. The started process will only run when the container
primary process (PID 1) is running, and will not be restarted when the container
is restarted.
#### Planned additions
The started process will have its own cgroups nested inside the container's
cgroups. This is used for process tracking and optionally resource allocation
handling for the new process. Freezer cgroup is required, the rest of the cgroups
@@ -340,5 +342,5 @@ are optional. The process executor must place its pid inside the correct
cgroups before starting the process. This is done so that no child processes or
threads can escape the cgroups.
When the process stopped, all child processes spawned by the process will be
stopped and the process's cgroups will be removed.
When the process is stopped, the process executor will try (in a best-effort way)
to stop all its children and remove the sub-cgroups.