[refactor] move container start/exit to up_pub

Move the container start/exit events to the upstream pub/sub event
stream. As these are more of a global event now that we support multiple
commands in the same container, it makes sense to move them to the more
general event stream.

This patch also remove the code in cmd waiting for container start or
exit, making (temporarily) the cmd unable to report power metrics. We
will fix that in a later commit.

This patch fixes complicated issues we had with how a job running
multiple commands in the container might not all wait for the end of the
container: now none of them do.

bin/cmd

@@ -44,13 +44,14 @@ class CommandLineInterface(object):
         # the command a container uuid as a way to make sure that we can make
         # the command idempotent.
         environ = os.environ
+        container_uuid = argv.ucontainername or str(uuid.uuid4())
         command = {'api': 'up_rpc_req',
                    'type': 'run',
                    'manifest': argv.manifest,
                    'path': argv.command,
                    'args': argv.args,
                    'environ': dict(environ),
-                   'container_uuid': argv.ucontainername or str(uuid.uuid4()),
+                   'container_uuid': container_uuid,
                    }
         msg = RPC_MSG['run'](**command)
         # command fsm
@@ -63,12 +64,7 @@ class CommandLineInterface(object):
         # the first message tells us if we started a container or not
         msg = self.client.recvmsg()
         assert msg.api == 'up_rpc_rep'
-        assert msg.type in ['start', 'process_start']
-        new_container = False
-        if msg.type == 'start':
-            new_container = True
-            msg = self.client.recvmsg()
-            assert msg.type == 'process_start'
+        assert msg.type == 'process_start'
         state = 'started'
         while(True):
             msg = self.client.recvmsg()
@@ -85,17 +81,13 @@ class CommandLineInterface(object):
                     erreof = True
             elif msg.type == 'process_exit':
                 logger.info("process ended: %r", msg)
-                if not new_container:
-                    state = 'exiting'
-            elif msg.type == 'exit':
-                if state == 'started':
-                    state = 'exiting'
-                    exitmsg = msg
-                else:
-                    logger.info("unexpected exit message: %r", msg)
+                state = 'exiting'
+                exitmsg = msg
+            else:
+                logger.error("unexpected message: %r", msg)
             if outeof and erreof and state == 'exiting':
                 state = 'exit'
-                logger.info("container ended: %r", exitmsg)
+                logger.info("command ended: %r", exitmsg)
                 break
 
     def do_list(self, argv):

nrm/daemon.py

@@ -100,14 +100,14 @@ class Daemon(object):
                     p = container.power['profile']
                     p['start'] = self.machine_info['energy']['energy']
                     p['start']['time'] = self.machine_info['time']
-                update = {'api': 'up_rpc_rep',
-                          'type': 'start',
+                update = {'api': 'up_pub',
+                          'type': 'container_start',
                           'container_uuid': container_uuid,
                           'errno': 0 if container else -1,
                           'power': container.power['policy'] or dict()
                           }
-                self.upstream_rpc_server.sendmsg(RPC_MSG['start'](**update),
-                                                 client)
+                self.upstream_pub_server.sendmsg(
+                        PUB_MSG['container_start'](**update))
                 self.container_owner[container.uuid] = client
 
             # now deal with the process itself
@@ -219,8 +219,8 @@ class Daemon(object):
                     # kill everything
                     if self.container_owner[container.uuid] == clientid:
                         # deal with container exit
-                        msg = {'api': 'up_rpc_rep',
-                               'type': 'exit',
+                        msg = {'api': 'up_pub',
+                               'type': 'container_exit',
                                'container_uuid': container.uuid,
                                'profile_data': dict(),
                                }
@@ -242,8 +242,8 @@ class Daemon(object):
                                         container.uuid, diff)
                             msg['profile_data'] = diff
                         self.container_manager.delete(container.uuid)
-                        self.upstream_rpc_server.sendmsg(
-                                RPC_MSG['exit'](**msg), clientid)
+                        self.upstream_pub_server.sendmsg(
+                                PUB_MSG['container_exit'](**msg))
                         del self.container_owner[container.uuid]
             else:
                 logger.debug("child update ignored")

nrm/messaging.py

@@ -29,23 +29,24 @@ MSGFORMATS['up_rpc_req'] = {'list': {},
                             'kill': {'container_uuid': basestring},
                             'setpower': {'limit': basestring},
                             }
-MSGFORMATS['up_rpc_rep'] = {'start': {'container_uuid': basestring,
-                                      'errno': int,
-                                      'power': dict},
-                            'list': {'payload': list},
+MSGFORMATS['up_rpc_rep'] = {'list': {'payload': list},
                             'stdout': {'container_uuid': basestring,
                                        'payload': basestring},
                             'stderr': {'container_uuid': basestring,
                                        'payload': basestring},
-                            'exit': {'container_uuid': basestring,
-                                     'profile_data': dict},
                             'process_start': {'container_uuid': basestring,
                                               'pid': int},
                             'process_exit': {'container_uuid': basestring,
                                              'status': basestring},
                             'getpower': {'limit': basestring},
                             }
-MSGFORMATS['up_pub'] = {'power': {'total': int, 'limit': float}}
+MSGFORMATS['up_pub'] = {'power': {'total': int, 'limit': float},
+                        'container_start': {'container_uuid': basestring,
+                                            'errno': int,
+                                            'power': dict},
+                        'container_exit': {'container_uuid': basestring,
+                                           'profile_data': dict},
+                        }
 
 # Mirror of the message formats, using namedtuples as the actual transport
 # for users of this messaging layer.