completed and tested timed containers with the real-world setup

Dockerfiles/service/app.py

@@ -18,6 +18,12 @@
 SUCCESS_MESSAGE = "Resources allocated successfully"
 app = Flask(__name__)
 
+
+DURATION = int(os.environ['DURATION'])
+logging.info(f'DURATION set to: {DURATION}')
+
+
+
 def stop_server():
     print('Stopping Flask app...')
     # Send a SIGTERM signal to the current process
@@ -35,7 +41,7 @@ def index():
 
     try:
 
-        logging.info("received new configuration from sdghafouri/utilization-server-smart-scheduler: {}".format(
+        logging.info("received new configuration from sdghafoLIFETIMEuri/utilization-server-smart-scheduler: {}".format(
             request.form
         ))
 
@@ -86,7 +92,7 @@ def index():
     # logging.info('running stress-ng: "{}"'.format(command))
     # os.system(command)
 
-    timer = threading.Timer(10, stop_server)
+    timer = threading.Timer(DURATION, stop_server)
     timer.start()
 
     # logging.info("serving 'app' on port {}".format(PORT))

docs/docker-server-setup.md

@@ -27,7 +27,7 @@
 open cv and dm_tree, solution -> wait for opencv and upgrade pip for dm_tree
 4. error 2
 ```
->>> import ray.rllib.agents.pg as pg
+>>> import ray.rllib.algorithms.pg as pg
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "/homes/sg324/python-envs/development/lib/python3.8/site-packages/ray/rllib/__init__.py", line 5, in <module>

docs/problems/eecs-server-setup.md

@@ -27,7 +27,7 @@
 open cv and dm_tree, solution -> wait for opencv and upgrade pip for dm_tree
 4. error 2
 ```
->>> import ray.rllib.agents.pg as pg
+>>> import ray.rllib.algorithms.pg as pg
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
      File "/homes/sg324/python-envs/development/lib/python3.8/site-packages/ray/rllib/__init__.py", line 5, in <module>

experiments/analysis/check_learned.py

@@ -19,11 +19,11 @@
 from ray.rllib.utils.framework import try_import_torch
 import pprint
 import gym
-import ray.rllib.agents.ppo as ppo
-import ray.rllib.agents.a3c as a3c
-import ray.rllib.agents.impala as impala
-import ray.rllib.agents.pg as pg
-import ray.rllib.agents.dqn as dqn
+import ray.rllib.algorithms.ppo as ppo
+import ray.rllib.algorithms.a3c as a3c
+import ray.rllib.algorithms.impala as impala
+import ray.rllib.algorithms.pg as pg
+import ray.rllib.algorithms.dqn as dqn
 from pprint import PrettyPrinter
 pp = PrettyPrinter(indent=4)
 from copy import deepcopy

experiments/analysis/test_rls-debug.py

@@ -16,11 +16,11 @@
 from ray.rllib.utils.framework import try_import_torch
 import pprint
 import gym
-import ray.rllib.agents.ppo as ppo
-import ray.rllib.agents.impala as impala
-import ray.rllib.agents.a3c as a3c
-import ray.rllib.agents.pg as pg
-import ray.rllib.agents.dqn as dqn
+import ray.rllib.algorithms.ppo as ppo
+import ray.rllib.algorithms.impala as impala
+import ray.rllib.algorithms.a3c as a3c
+import ray.rllib.algorithms.pg as pg
+import ray.rllib.algorithms.dqn as dqn
 import pandas as pd
 from pprint import PrettyPrinter
 pp = PrettyPrinter(indent=4)

experiments/analysis/test_rls.py

@@ -16,11 +16,11 @@
 from ray.rllib.utils.framework import try_import_torch
 import pprint
 import gym
-import ray.rllib.agents.ppo as ppo
-import ray.rllib.agents.impala as impala
-import ray.rllib.agents.a3c as a3c
-import ray.rllib.agents.pg as pg
-import ray.rllib.agents.dqn as dqn
+import ray.rllib.algorithms.ppo as ppo
+import ray.rllib.algorithms.impala as impala
+import ray.rllib.algorithms.a3c as a3c
+import ray.rllib.algorithms.pg as pg
+import ray.rllib.algorithms.dqn as dqn
 import pandas as pd
 from pprint import PrettyPrinter
 pp = PrettyPrinter(indent=4)
@@ -308,7 +308,7 @@ def fix_grid_searches(
 @click.option('--train-series', required=True, type=int, default=121)
 @click.option('--type-env', required=True,
               type=click.Choice(['sim-scheduler', 'kube-scheduler']),
-              default='sim-scheduler')
+              default='kube-scheduler')
 @click.option('--cluster-id', required=True, type=int, default=19)
 @click.option('--workload-id', required=True, type=int, default=0)
 @click.option('--experiment-id', required=True, type=int, default=0)

experiments/utils/callbacks.py

@@ -4,7 +4,7 @@
 from packaging import version
 from tabulate import tabulate
 if version.parse(ray.__version__) < version.parse('1.9.0'):
-    from ray.rllib.agents.callbacks import DefaultCallbacks
+    from ray.rllib.algorithms.callbacks import DefaultCallbacks
 else:
     from ray.rllib.algorithms.callbacks import DefaultCallbacks
 from ray.rllib.env import BaseEnv

experiments/utils/callbacks_update.py

@@ -6,7 +6,7 @@
 from packaging import version
 from tabulate import tabulate
 if version.parse(ray.__version__) < version.parse('1.9.0'):
-    from ray.rllib.agents.callbacks import DefaultCallbacks
+    from ray.rllib.algorithms.callbacks import DefaultCallbacks
 else:
     from ray.rllib.algorithms.callbacks import DefaultCallbacks
 from ray.rllib.env import BaseEnv

smart_scheduler/src/smart_scheduler/envs/kubernetes/kube_scheduler_env.py

@@ -411,7 +411,8 @@ def schedule(self, service_id: int, node_id: int) -> bool:
         if schedule_success:
             self.schedule_on_kube(
                 service=self.pending_services[service_index],
-                node=self.kube_nodes[node_id])
+                node=self.kube_nodes[node_id],
+                duration=self.pending_services[service_index].duration)
             self.pending_services.pop(service_index)
         # return true if successful
         return schedule_success
@@ -578,7 +579,7 @@ def done(self):
         return False
 
 
-    def schedule_on_kube(self, service: Service, node: KubeNode):
+    def schedule_on_kube(self, service: Service, node: KubeNode, duration: int):
         # TODO add scheduling success
         pod = construct_pod(
             name=service.service_name,
@@ -589,6 +590,7 @@ def schedule_on_kube(self, service: Service, node: KubeNode):
                 service.requests[0]),
             limit_cpu="{}".format(
                 service.requests[1]),
+            env={'DURATION': str(duration)}
         )
 
         svc = construct_svc(