Comments

client/driver/executor/executor.go

@@ -120,7 +120,7 @@ type ExecCommand struct {
 	ResourceLimits bool
 
 	// Cgroup marks whether we put the process in a cgroup. Setting this field
-	// doesn't enforce resource limits. To enforce limits, set ResoruceLimits.
+	// doesn't enforce resource limits. To enforce limits, set ResourceLimits.
 	// Using the cgroup does allow more precise cleanup of processes.
 	BasicProcessCgroup bool
 }

client/driver/raw_exec_test.go

@@ -265,6 +265,9 @@ func TestRawExecDriver_Start_Kill_Wait(t *testing.T) {
 	}
 }
 
+// This test creates a process tree such that without cgroups tracking the
+// processes cleanup of the children would not be possible. Thus the test
+// asserts that the processes get killed properly when using cgroups.
 func TestRawExecDriver_Start_Kill_Wait_Cgroup(t *testing.T) {
 	tu.ExecCompatible(t)
 	t.Parallel()

website/source/docs/drivers/raw_exec.html.md

@@ -109,3 +109,10 @@ The `raw_exec` driver will set the following client attributes:
 ## Resource Isolation
 
 The `raw_exec` driver provides no isolation.
+
+If the launched process creates a new process group, it is possible that Nomad
+will leak processes on shutdown unless the application forwards signals
+properly. Nomad will not leak any processes if cgroups are being used to manage
+the process tree. Cgroups are used on Linux when Nomad is being run with
+appropriate priviledges, the cgroup system is mounted and the operator hasn't
+disabled cgroups for the driver.