add repeatable seeder sets

benchmarks/deluge/config.py

@@ -49,12 +49,17 @@ DelugeDisseminationExperiment = IteratedExperiment[BoundExperiment[StaticDissemi
 
 
 class DelugeExperimentConfig(ExperimentBuilder[DelugeDisseminationExperiment]):
-    repetitions: int = Field(gt=0)
+    seeder_sets: int = Field(gt=0, default=1, description='Number of distinct seeder sets to experiment with')
-    file_size: int = Field(gt=0)
+    seeders: int = Field(gt=0, description='Number of seeders per seeder set')
-    seeders: int = Field(gt=0)
+
-    shared_volume_path: Path
+    repetitions: int = Field(gt=0, description='How many experiment repetitions to run for each seeder set')
-    tracker_announce_url: HttpUrl
+    file_size: int = Field(gt=0, description='File size, in bytes')
-    nodes: List[DelugeNodeConfig] | DelugeNodeSetConfig
+
+    shared_volume_path: Path = Field(description='Path to the volume shared between clients and experiment runner')
+    tracker_announce_url: HttpUrl = Field(description='URL to the tracker announce endpoint')
+
+    nodes: List[DelugeNodeConfig] | DelugeNodeSetConfig = Field(
+        description='Configuration for the nodes that make up the network')
 
     def build(self) -> DelugeDisseminationExperiment:
         nodes_specs = self.nodes.nodes if isinstance(self.nodes, DelugeNodeSetConfig) else self.nodes
@@ -76,15 +81,16 @@ class DelugeExperimentConfig(ExperimentBuilder[DelugeDisseminationExperiment]):
             polling_interval=0.5,
         )
 
-        repetitions = (
+        def repetitions():
-            env.bind(StaticDisseminationExperiment(
+            for seeder_set in range(self.seeder_sets):
+                seeders = list(islice(sample(len(network)), self.seeders))
+                for experiment_run in range(self.repetitions):
+                    yield env.bind(StaticDisseminationExperiment(
                         network=network,
-                seeders=list(islice(sample(len(network)), self.seeders)),
+                        seeders=seeders,
                         data=RandomTempData(size=self.file_size,
-                                    meta=DelugeMeta(f'dataset-{experiment_run}',
+                                            meta=DelugeMeta(f'dataset-{seeder_set}',
                                                             announce_url=tracker.announce_url))
                     ))
-            for experiment_run in range(self.repetitions)
-        )
 
-        return IteratedExperiment(repetitions)
+        return IteratedExperiment(repetitions())

benchmarks/deluge/tests/test_config.py

@@ -4,6 +4,7 @@ from unittest.mock import patch
 
 import yaml
 
+from benchmarks.core.experiments.static_experiment import StaticDisseminationExperiment
 from benchmarks.deluge.config import DelugeNodeSetConfig, DelugeNodeConfig, DelugeExperimentConfig
 from benchmarks.deluge.deluge_node import DelugeNode
 
@@ -44,6 +45,7 @@ def test_should_expand_node_sets_into_simple_nodes():
         ),
     ]
 
+
 def test_should_respect_first_node_index():
     nodeset = DelugeNodeSetConfig(
         name='deluge-{node_index}',
@@ -69,6 +71,7 @@ def test_should_respect_first_node_index():
         ),
     ]
 
+
 def test_should_build_experiment_from_config():
     config_file = StringIO("""
     deluge_experiment:
@@ -99,3 +102,40 @@ def test_should_build_experiment_from_config():
     assert cast(DelugeNode, repetitions[0].experiment.nodes[5]).daemon_args['port'] == 6890
 
 
+def test_should_create_n_repetitions_per_seeder_set():
+    config_file = StringIO("""
+    deluge_experiment:
+      seeder_sets: 2
+      repetitions: 3
+      seeders: 3
+      tracker_announce_url: http://localhost:2020/announce
+      file_size: 1024
+      shared_volume_path: /var/lib/deluge
+
+      nodes:
+        network_size: 100
+        name: 'deluge-{node_index}'
+        address: 'node-{node_index}.deluge.codexbenchmarks.svc.cluster.local'
+        daemon_port: 6890
+        listen_ports: [ 6891, 6892 ]
+    """)
+
+    config = DelugeExperimentConfig.model_validate(yaml.safe_load(config_file)['deluge_experiment'])
+
+    # Need to patch mkdir, or we'll try to actually create the folder when DelugeNode gets initialized.
+    with patch('pathlib.Path.mkdir'):
+        experiment = config.build()
+        repetitions = list(experiment.experiments)
+
+    assert len(repetitions) == 3 * 2
+
+    experiment_set1_1 = cast(StaticDisseminationExperiment, repetitions[0].experiment)
+    experiment_set1_2 = cast(StaticDisseminationExperiment, repetitions[2].experiment)
+    experiment_set2_1 = cast(StaticDisseminationExperiment, repetitions[3].experiment)
+
+    # FIXME Ehm... this test might actually fail with a very low probability if the seeder sets end
+    #  up being the same by chance, but the probability is very small (you're drawing 3 out of 100 twice
+    #  and the result needs to be the same). The fix would be having a deterministic sampler but I feel
+    #  lazy right now. :-)
+    assert experiment_set1_1.seeders == experiment_set1_2.seeders
+    assert experiment_set1_1.seeders != experiment_set2_1.seeders

experiments.k8s.yaml

@@ -1,4 +1,5 @@
 deluge_experiment:
+  seeder_sets: ${SEEDER_SETS}
   seeders: ${SEEDERS}
   tracker_announce_url: ${TRACKER_ANNOUNCE_URL}
   file_size: ${FILE_SIZE}

k8s/testrunner-job.yaml

@@ -5,11 +5,13 @@ metadata:
   namespace: codex-benchmarks
   labels:
     app: testrunner
+
 spec:
   template:
     metadata:
       labels:
         app: testrunner
+
     spec:
       containers:
         - name: testrunner
@@ -18,11 +20,13 @@ spec:
           args: [ "run", "deluge_experiment" ]
           env:
             - name: NETWORK_SIZE
-              value: "100"
+              value: "10"
             - name: SEEDERS
               value: "4"
             - name: REPETITIONS
               value: "10"
+            - name: SEEDER_SETS
+              value: "2"
             - name: FILE_SIZE
               value: "104857600"
             - name: TRACKER_ANNOUNCE_URL
@@ -47,9 +51,11 @@ spec:
             limits:
               cpu: "1"
               memory: "2Gi"
+
       volumes:
         - name: benchmark-volume
           persistentVolumeClaim:
             claimName: deluge-pvc
       restartPolicy: Never
+
   backoffLimit: 0