[Fix] reduce concurrent client/server processes to avoid port already in use (#3697)

tests/distributed/test_dist_graph_store.py

@@ -584,9 +584,9 @@ def test_server_client():
     check_server_client(True, 1, 1)
     check_server_client(False, 1, 1)
     check_server_client(True, 2, 2)
-    check_server_client(True, 1, 1, 5)
-    check_server_client(False, 1, 1, 5)
-    check_server_client(True, 2, 2, 5)
+    check_server_client(True, 1, 1, 2)
+    check_server_client(False, 1, 1, 2)
+    check_server_client(True, 2, 2, 2)
 
 @unittest.skipIf(os.name == 'nt', reason='Do not support windows yet')
 @unittest.skipIf(dgl.backend.backend_name == "tensorflow", reason="TF doesn't support distributed DistEmbedding")
@@ -597,9 +597,9 @@ def test_dist_emb_server_client():
     check_dist_emb_server_client(True, 1, 1)
     check_dist_emb_server_client(False, 1, 1)
     check_dist_emb_server_client(True, 2, 2)
-    check_dist_emb_server_client(True, 1, 1, 5)
-    check_dist_emb_server_client(False, 1, 1, 5)
-    check_dist_emb_server_client(True, 2, 2, 5)
+    check_dist_emb_server_client(True, 1, 1, 2)
+    check_dist_emb_server_client(False, 1, 1, 2)
+    check_dist_emb_server_client(True, 2, 2, 2)
 
 @unittest.skipIf(dgl.backend.backend_name == "tensorflow", reason="TF doesn't support some of operations in DistGraph")
 @unittest.skipIf(dgl.backend.backend_name == "mxnet", reason="Turn off Mxnet support")

tests/distributed/test_mp_dataloader.py

@@ -213,7 +213,7 @@ def check_neg_dataloader(g, tmpdir, num_server, num_workers):
 @pytest.mark.parametrize("num_workers", [0, 4])
 @pytest.mark.parametrize("drop_last", [True, False])
 @pytest.mark.parametrize("reshuffle", [True, False])
-@pytest.mark.parametrize("num_groups", [1, 5])
+@pytest.mark.parametrize("num_groups", [1, 2])
 def test_dist_dataloader(tmpdir, num_server, num_workers, drop_last, reshuffle, num_groups):
     reset_envs()
     generate_ip_config("mp_ip_config.txt", num_server, num_server)
@@ -451,7 +451,7 @@ if __name__ == "__main__":
         test_dataloader(Path(tmpdirname), 3, 4, 'node')
         test_dataloader(Path(tmpdirname), 3, 4, 'edge')
         test_neg_dataloader(Path(tmpdirname), 3, 4)
-        for num_groups in [1, 5]:
+        for num_groups in [1, 2]:
             test_dist_dataloader(Path(tmpdirname), 3, 0, True, True, num_groups)
             test_dist_dataloader(Path(tmpdirname), 3, 4, True, True, num_groups)
             test_dist_dataloader(Path(tmpdirname), 3, 0, True, False, num_groups)

tests/distributed/test_rpc.py

@@ -251,8 +251,8 @@ def test_multi_client_groups():
     num_servers = 1
     generate_ip_config(ip_config, num_machines, num_servers)
     # presssue test
-    num_clients = 15
-    num_groups = 15
+    num_clients = 2
+    num_groups = 2
     ctx = mp.get_context('spawn')
     pserver_list = []
     for i in range(num_servers*num_machines):