slurmray 3.5.9__py3-none-any.whl → 3.6.0__py3-none-any.whl

slurmray/RayLauncher.py

@@ -168,10 +168,12 @@ class RayLauncher:
         """
         print("Serializing function and arguments...")
 
-        # Remove the old python script
-        for file in os.listdir(self.project_path):
-            if file.endswith(".pkl"):
-                os.remove(os.path.join(self.project_path, file))
+        # Check if there is already a func.pkl and args.pkl file
+        if os.path.exists(
+            os.path.join(self.project_path, "func.pkl")
+        ) and os.path.exists(os.path.join(self.project_path, "args.pkl")):
+            print("Function and arguments already serialized.")
+            return
 
         # Pickle the function
         with open(os.path.join(self.project_path, "func.pkl"), "wb") as f:
@@ -299,9 +301,7 @@ class RayLauncher:
 
         # Wait for log file to be created
         current_queue = None
-        queue_log_file = os.path.join(
-            self.project_path, "{}_queue.log".format(job_name)
-        )
+        queue_log_file = os.path.join(self.project_path, "queue.log")
         with open(queue_log_file, "w") as f:
             f.write("")
         print(
@@ -373,7 +373,11 @@ class RayLauncher:
                         text += "\n"
                         f.write(text)
 
+                        # Print the queue
+                        print(text)
+
         # Wait for the job to finish while printing the log
+        print("Job started! Waiting for the job to finish...")
         log_cursor_position = 0
         job_finished = False
         while not job_finished:
@@ -407,7 +411,7 @@ class RayLauncher:
                 if self.server_password is None:
                     # Add ssh key
                     self.server_password = getpass("Enter your cluster password: ")
-                
+
                 ssh_client.connect(
                     hostname=self.server_ssh,
                     username=self.server_username,
@@ -440,11 +444,12 @@ class RayLauncher:
             # lines = [re.sub(r'bitsandbytes\n', 'bitsandbytes --global-option="--cuda_ext"\n', line) for line in lines]
             lines = [re.sub(r"slurmray\n", "", line) for line in lines]
             # Add slurmray --pre
-            lines.append("slurmray --pre\n")
+            lines.append("slurmray --pre \n")
             # Solve torch buf (https://github.com/pytorch/pytorch/issues/111469)
             if "torchaudio\n" or "torchvision\n" in lines:
-                lines.append("torch==2.1.1\n")
-                lines.append("--index-url https://download.pytorch.org/whl/cu121\n")
+                lines.append(
+                    "torch==2.1.1 --index-url https://download.pytorch.org/whl/cu121\n"
+                )
 
         with open(f"{self.project_path}/requirements.txt", "w") as file:
             file.writelines(lines)
@@ -484,13 +489,24 @@ class RayLauncher:
                 break
             print(line, end="")
 
+        stdout.channel.recv_exit_status()
+
         # Downloading result
         print("Downloading result...")
-        sftp.get(
-            "slurmray-server/.slogs/server/result.pkl",
-            os.path.join(self.project_path, "result.pkl"),
-        )
-        print("Result downloaded!")
+        try:
+            sftp.get(
+                "slurmray-server/.slogs/server/result.pkl",
+                os.path.join(self.project_path, "result.pkl"),
+            )
+            print("Result downloaded!")
+        except FileNotFoundError:
+            # Check for errors
+            stderr_lines = stderr.readlines()
+            if stderr_lines:
+                print("\nErrors:\n")
+                for line in stderr_lines:
+                    print(line, end="")
+                print("An error occured, please check the logs.")
 
     def __write_server_script(self):
         """This funtion will write a script with the given specifications to run slurmray on the cluster"""
@@ -541,20 +557,24 @@ if __name__ == "__main__":
         return result
 
     launcher = RayLauncher(
-        project_name="example", # Name of the project (will create a directory with this name in the current directory)
-        func=example_func, # Function to execute
-        args={"x": 1}, # Arguments of the function
-        files=["slurmray/RayLauncher.py"], # List of files to push to the cluster (file path will be recreated on the cluster)
-        modules=[], # List of modules to load on the curnagl Cluster (CUDA & CUDNN are automatically added if use_gpu=True)
-        node_nbr=1, # Number of nodes to use
-        use_gpu=True, # If you need A100 GPU, you can set it to True
-        memory=8, # In MegaBytes
-        max_running_time=5, # In minutes
-        runtime_env={"env_vars": {"NCCL_SOCKET_IFNAME": "eno1"}}, # Example of environment variable
-        server_run=True, # To run the code on the cluster and not locally
-        server_ssh="curnagl.dcsr.unil.ch", # Address of the SLURM server
-        server_username="hjamet", # Username to connect to the server
-        server_password=None, # Will be asked in the terminal
+        project_name="example",  # Name of the project (will create a directory with this name in the current directory)
+        func=example_func,  # Function to execute
+        args={"x": 1},  # Arguments of the function
+        files=[
+            "slurmray/RayLauncher.py"
+        ],  # List of files to push to the cluster (file path will be recreated on the cluster)
+        modules=[],  # List of modules to load on the curnagl Cluster (CUDA & CUDNN are automatically added if use_gpu=True)
+        node_nbr=1,  # Number of nodes to use
+        use_gpu=False,  # If you need A100 GPU, you can set it to True
+        memory=8,  # In MegaBytes
+        max_running_time=5,  # In minutes
+        runtime_env={
+            "env_vars": {"NCCL_SOCKET_IFNAME": "eno1"}
+        },  # Example of environment variable
+        server_run=True,  # To run the code on the cluster and not locally
+        server_ssh="curnagl.dcsr.unil.ch",  # Address of the SLURM server
+        server_username="hjamet",  # Username to connect to the server
+        server_password=None,  # Will be asked in the terminal
     )
 
     result = launcher()

slurmray/assets/slurmray_server.sh

@@ -17,7 +17,34 @@ fi
 source .venv/bin/activate
 
 # Install requirements
-pip3 install -r requirements.txt
+## Load all installed packages into a variable
+installed_packages=$(pip3 list --format=freeze)
+## Function to check if a package is installed
+is_package_installed() {
+  package=$1
+  echo "$installed_packages" | grep -i "^$package==" &> /dev/null
+  return $?
+}
+## Read the requirements.txt file line by line
+while IFS= read -r package
+do
+  # Check if the line is not empty
+  if [ -n "$package" ]; then
+    echo "Checking package: $package"
+    # Extract the package name without options
+    package_name=$(echo "$package" | awk '{print $1}' | cut -d'=' -f1)
+    if is_package_installed "$package_name"; then
+      echo "The package $package_name is already installed."
+    else
+      echo "Installing package: $package"
+      command="pip3 install $package"
+      eval "$command"
+      if [ $? -ne 0 ]; then
+        echo "Error while installing $package"
+      fi
+    fi
+  fi
+done < "requirements.txt"
 
 # Fix torch bug (https://github.com/pytorch/pytorch/issues/111469)
 export LD_LIBRARY_PATH=$HOME/slurmray-server/.venv/lib/python3.9/site-packages/nvidia/nvjitlink/lib:$LD_LIBRARY_PATH

slurmray/assets/slurmray_server_template.py

@@ -15,8 +15,5 @@ if __name__ == "__main__":
         server_ssh=None,
         server_username=None,
     )
-    
-    # Remove serialization
-    launcher.__serialize_func_and_args = lambda *args, **kwargs : print("No serialization done.")
 
-    result = launcher()
+    result = launcher()

{slurmray-3.5.9.dist-info → slurmray-3.6.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: slurmray
-Version: 3.5.9
+Version: 3.6.0
 Summary: SlurmRay is a module for effortlessly distributing tasks on a Slurm cluster using the Ray library. 
 Home-page: https://henri-jamet.vercel.app/
 License: Apache License
@@ -23,7 +23,7 @@ Description-Content-Type: text/markdown
 
 # SLURM_RAY
 
-👉[Full documentation](https://henri-jamet.vercel.app/cards/documentation/slurm-ray/slurm-ray/)
+👉[Full documentation](https://www.henri-jamet.com/docs/slurmray/slurm-ray/)
 
 ## Description
 

slurmray-3.6.0.dist-info/RECORD

@@ -0,0 +1,10 @@
+slurmray/RayLauncher.py,sha256=3pOOMGDENspcfHiEaxWoyLx6Wep5XHItRBuSXJ3cUvI,23279
+slurmray/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+slurmray/assets/sbatch_template.sh,sha256=c-7J4ItzrctDrbF5Znu8p1d_xIgayC9puhjX3nLMzsk,2273
+slurmray/assets/slurmray_server.sh,sha256=-PpX3AitLVfAYjyNqE3BjtDu5uvk11KoiaCUVgmtcEQ,1506
+slurmray/assets/slurmray_server_template.py,sha256=PF4Rl3TrTS8hI0jbCMlOuRmICkL_OucO6R-uKq83kvg,446
+slurmray/assets/spython_template.py,sha256=kRUvNQs9iCcg0wJLmm9LV0TnbUdlenZMYPr_bZPkXLg,597
+slurmray-3.6.0.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+slurmray-3.6.0.dist-info/METADATA,sha256=VrDOXUE4gBXscmducVlQ2pqzEl2kV0kRO9pBKO-gWQM,3530
+slurmray-3.6.0.dist-info/WHEEL,sha256=d2fvjOD7sXsVzChCqf0Ty0JbHKBaLYwDbGQDwQTnJ50,88
+slurmray-3.6.0.dist-info/RECORD,,

slurmray-3.5.9.dist-info/RECORD

@@ -1,10 +0,0 @@
-slurmray/RayLauncher.py,sha256=6ZS8o4CT2ulwTlmW0ahImNEHZy0h-srgNyJoxGew8lg,22617
-slurmray/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-slurmray/assets/sbatch_template.sh,sha256=c-7J4ItzrctDrbF5Znu8p1d_xIgayC9puhjX3nLMzsk,2273
-slurmray/assets/slurmray_server.sh,sha256=BpmyczNtMlsRimbUYU2XSE59YHGSsozo3rqSQcXNubQ,638
-slurmray/assets/slurmray_server_template.py,sha256=xjuF3nwvQONRxQNzXkAsgFkInY80y6ynkHE9zJjw0xk,575
-slurmray/assets/spython_template.py,sha256=kRUvNQs9iCcg0wJLmm9LV0TnbUdlenZMYPr_bZPkXLg,597
-slurmray-3.5.9.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-slurmray-3.5.9.dist-info/METADATA,sha256=lbPv9a_Mn5nyXayOfrQka6rZG6QmtRzOVOsyjAxdJO4,3549
-slurmray-3.5.9.dist-info/WHEEL,sha256=d2fvjOD7sXsVzChCqf0Ty0JbHKBaLYwDbGQDwQTnJ50,88
-slurmray-3.5.9.dist-info/RECORD,,