fractal-server 2.3.2__tar.gz → 2.3.4__tar.gz

{fractal_server-2.3.2 → fractal_server-2.3.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: fractal-server
-Version: 2.3.2
+Version: 2.3.4
 Summary: Server component of the Fractal analytics platform
 Home-page: https://github.com/fractal-analytics-platform/fractal-server
 License: BSD-3-Clause

fractal_server-2.3.4/fractal_server/__init__.py

@@ -0,0 +1 @@
+__VERSION__ = "2.3.4"

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/app/routes/api/__init__.py

@@ -2,9 +2,12 @@
 `api` module
 """
 from fastapi import APIRouter
+from fastapi import Depends
 
 from ....config import get_settings
 from ....syringe import Inject
+from ...models.security import UserOAuth
+from ...security import current_active_superuser
 
 
 router_api = APIRouter()
@@ -17,3 +20,9 @@ async def alive():
         alive=True,
         version=settings.PROJECT_VERSION,
     )
+
+
+@router_api.get("/settings/")
+async def view_settings(user: UserOAuth = Depends(current_active_superuser)):
+    settings = Inject(get_settings)
+    return settings.get_sanitized()

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/app/routes/api/v1/project.py

@@ -250,9 +250,18 @@ async def apply_workflow(
     db: AsyncSession = Depends(get_async_db),
 ) -> Optional[ApplyWorkflowReadV1]:
 
+    settings = Inject(get_settings)
+    if settings.FRACTAL_API_V1_MODE == "include_without_submission":
+        raise HTTPException(
+            status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
+            detail=(
+                "Legacy API is still accessible, "
+                "but the submission of legacy jobs is not available."
+            ),
+        )
+
     # Remove non-submitted V1 jobs from the app state when the list grows
     # beyond a threshold
-    settings = Inject(get_settings)
     if (
         len(request.app.state.jobsV1)
         > settings.FRACTAL_API_MAX_JOB_LIST_LENGTH

fractal_server-2.3.4/fractal_server/app/runner/compress_folder.py

@@ -0,0 +1,132 @@
+"""
+Wrap `tar` compression command.
+
+This module is used both locally (in the environment where `fractal-server`
+is running) and remotely (as a standalon Python module, executed over SSH).
+
+This is a twin-module of `extract_archive.py`.
+
+The reason for using the `tar` command via `subprocess` rather than Python
+built-in `tarfile` library has to do with performance issues we observed
+when handling files which were just created within a SLURM job, and in the
+context of a CephFS filesystem.
+"""
+import shutil
+import sys
+from pathlib import Path
+
+from fractal_server.app.runner.run_subprocess import run_subprocess
+from fractal_server.logger import get_logger
+from fractal_server.logger import set_logger
+
+
+def copy_subfolder(src: Path, dest: Path, logger_name: str):
+    cmd_cp = f"cp -r {src.as_posix()} {dest.as_posix()}"
+    logger = get_logger(logger_name=logger_name)
+    logger.debug(f"{cmd_cp=}")
+    res = run_subprocess(cmd=cmd_cp, logger_name=logger_name)
+    return res
+
+
+def create_tar_archive(
+    tarfile_path: Path,
+    subfolder_path_tmp_copy: Path,
+    logger_name: str,
+    remote_to_local: bool,
+):
+    logger = get_logger(logger_name)
+
+    if remote_to_local:
+        exclude_options = "--exclude *sbatch --exclude *_in_*.pickle "
+    else:
+        exclude_options = ""
+
+    cmd_tar = (
+        f"tar czf {tarfile_path} "
+        f"{exclude_options} "
+        f"--directory={subfolder_path_tmp_copy.as_posix()} "
+        "."
+    )
+    logger.debug(f"cmd tar:\n{cmd_tar}")
+    run_subprocess(cmd=cmd_tar, logger_name=logger_name)
+
+
+def remove_temp_subfolder(subfolder_path_tmp_copy: Path, logger_name: str):
+    logger = get_logger(logger_name)
+    try:
+        logger.debug(f"Now remove {subfolder_path_tmp_copy}")
+        shutil.rmtree(subfolder_path_tmp_copy)
+    except Exception as e:
+        logger.debug(f"ERROR during shutil.rmtree: {e}")
+
+
+def compress_folder(
+    subfolder_path: Path, remote_to_local: bool = False
+) -> str:
+    """
+    Compress e.g. `/path/archive` into `/path/archive.tar.gz`
+
+    Note that `/path/archive.tar.gz` may already exist. In this case, it will
+    be overwritten.
+
+    Args:
+        subfolder_path: Absolute path to the folder to compress.
+        remote_to_local: If `True`, exclude some files from the tar.gz archive.
+
+    Returns:
+        Absolute path to the tar.gz archive.
+    """
+
+    logger_name = "compress_folder"
+    logger = set_logger(logger_name)
+
+    logger.debug("START")
+    logger.debug(f"{subfolder_path=}")
+    parent_dir = subfolder_path.parent
+    subfolder_name = subfolder_path.name
+    tarfile_path = (parent_dir / f"{subfolder_name}.tar.gz").as_posix()
+    logger.debug(f"{tarfile_path=}")
+
+    subfolder_path_tmp_copy = (
+        subfolder_path.parent / f"{subfolder_path.name}_copy"
+    )
+    try:
+        copy_subfolder(
+            subfolder_path, subfolder_path_tmp_copy, logger_name=logger_name
+        )
+        create_tar_archive(
+            tarfile_path,
+            subfolder_path_tmp_copy,
+            logger_name=logger_name,
+            remote_to_local=remote_to_local,
+        )
+        return tarfile_path
+
+    except Exception as e:
+        logger.debug(f"ERROR: {e}")
+        sys.exit(1)
+
+    finally:
+        remove_temp_subfolder(subfolder_path_tmp_copy, logger_name=logger_name)
+
+
+def main(sys_argv: list[str]):
+
+    help_msg = (
+        "Expected use:\n"
+        "python -m fractal_server.app.runner.compress_folder "
+        "path/to/folder [--remote-to-local]\n"
+    )
+    num_args = len(sys_argv[1:])
+    if num_args == 0:
+        sys.exit(f"Invalid argument.\n{help_msg}\nProvided: {sys_argv[1:]=}")
+    elif num_args == 1:
+        compress_folder(subfolder_path=Path(sys_argv[1]))
+    elif num_args == 2 and sys_argv[2] == "--remote-to-local":
+        compress_folder(subfolder_path=Path(sys_argv[1]), remote_to_local=True)
+    else:
+        sys.exit(f"Invalid argument.\n{help_msg}\nProvided: {sys_argv[1:]=}")
+
+
+if __name__ == "__main__":
+    main(sys.argv)

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/app/runner/executors/slurm/ssh/executor.py

@@ -13,7 +13,6 @@
 import json
 import math
 import sys
-import tarfile
 import threading
 import time
 from concurrent.futures import Future
@@ -38,9 +37,11 @@ from ...slurm._slurm_config import SlurmConfig
 from .._batching import heuristics
 from ._executor_wait_thread import FractalSlurmWaitThread
 from fractal_server.app.runner.components import _COMPONENT_KEY_
+from fractal_server.app.runner.compress_folder import compress_folder
 from fractal_server.app.runner.exceptions import JobExecutionError
 from fractal_server.app.runner.exceptions import TaskExecutionError
 from fractal_server.app.runner.executors.slurm.ssh._slurm_job import SlurmJob
+from fractal_server.app.runner.extract_archive import extract_archive
 from fractal_server.config import get_settings
 from fractal_server.logger import set_logger
 from fractal_server.ssh._fabric import FractalSSH
@@ -822,17 +823,12 @@ class FractalSlurmSSHExecutor(SlurmExecutor):
 
         # Create compressed subfolder archive (locally)
         local_subfolder = self.workflow_dir_local / subfolder_name
-        tarfile_name = f"{subfolder_name}.tar.gz"
-        tarfile_path_local = (
-            self.workflow_dir_local / tarfile_name
-        ).as_posix()
+        tarfile_path_local = compress_folder(local_subfolder)
+        tarfile_name = Path(tarfile_path_local).name
+        logger.info(f"Subfolder archive created at {tarfile_path_local}")
         tarfile_path_remote = (
             self.workflow_dir_remote / tarfile_name
         ).as_posix()
-        with tarfile.open(tarfile_path_local, "w:gz") as tar:
-            for this_file in local_subfolder.glob("*"):
-                tar.add(this_file, arcname=this_file.name)
-        logger.info(f"Subfolder archive created at {tarfile_path_local}")
 
         # Transfer archive
         t_0_put = time.perf_counter()
@@ -1222,7 +1218,8 @@ class FractalSlurmSSHExecutor(SlurmExecutor):
         tar_command = (
             f"{self.python_remote} "
             "-m fractal_server.app.runner.compress_folder "
-            f"{(self.workflow_dir_remote / subfolder_name).as_posix()}"
+            f"{(self.workflow_dir_remote / subfolder_name).as_posix()} "
+            "--remote-to-local"
         )
         stdout = self.fractal_ssh.run_command(cmd=tar_command)
         print(stdout)
@@ -1240,8 +1237,7 @@ class FractalSlurmSSHExecutor(SlurmExecutor):
         )
 
         # Extract tarfile locally
-        with tarfile.open(tarfile_path_local) as tar:
-            tar.extractall(path=(self.workflow_dir_local / subfolder_name))
+        extract_archive(Path(tarfile_path_local))
 
         t_1 = time.perf_counter()
         logger.info("[_get_subfolder_sftp] End - " f"elapsed: {t_1-t_0:.3f} s")

fractal_server-2.3.4/fractal_server/app/runner/extract_archive.py

@@ -0,0 +1,85 @@
+"""
+Wrap `tar` extraction command.
+
+This module is used both locally (in the environment where `fractal-server`
+is running) and remotely (as a standalon Python module, executed over SSH).
+
+This is a twin-module of `compress_folder.py`.
+
+The reason for using the `tar` command via `subprocess` rather than Python
+built-in `tarfile` library has to do with performance issues we observed
+when handling files which were just created within a SLURM job, and in the
+context of a CephFS filesystem.
+"""
+import sys
+from pathlib import Path
+
+from .run_subprocess import run_subprocess
+from fractal_server.logger import set_logger
+
+
+def _remove_suffix(*, string: str, suffix: str) -> str:
+    if string.endswith(suffix):
+        return string[: -len(suffix)]
+    else:
+        raise ValueError(f"Cannot remove {suffix=} from {string=}.")
+
+
+def extract_archive(archive_path: Path):
+    """
+    Extract e.g. `/path/archive.tar.gz` archive into `/path/archive` folder
+
+    Note that `/path/archive` may already exist. In this case, files with
+    the same name are overwritten and new files are added.
+
+    Arguments:
+        archive_path: Absolute path to the archive file.
+    """
+
+    logger_name = "extract_archive"
+    logger = set_logger(logger_name)
+
+    logger.debug("START")
+    logger.debug(f"{archive_path.as_posix()=}")
+
+    # Check archive_path is valid
+    if not archive_path.exists():
+        sys.exit(f"Missing file {archive_path.as_posix()}.")
+
+    # Prepare subfolder path
+    parent_dir = archive_path.parent
+    subfolder_name = _remove_suffix(string=archive_path.name, suffix=".tar.gz")
+    subfolder_path = parent_dir / subfolder_name
+    logger.debug(f"{subfolder_path.as_posix()=}")
+
+    # Create subfolder
+    subfolder_path.mkdir(exist_ok=True)
+
+    # Run tar command
+    cmd_tar = (
+        f"tar -xzvf {archive_path} "
+        f"--directory={subfolder_path.as_posix()} "
+        "."
+    )
+    logger.debug(f"{cmd_tar=}")
+    run_subprocess(cmd=cmd_tar, logger_name=logger_name)
+
+    logger.debug("END")
+
+
+def main(sys_argv: list[str]):
+    help_msg = (
+        "Expected use:\n"
+        "python -m fractal_server.app.runner.extract_archive "
+        "path/to/archive.tar.gz"
+    )
+
+    if len(sys_argv[1:]) != 1 or not sys_argv[1].endswith(".tar.gz"):
+        sys.exit(f"Invalid argument.\n{help_msg}\nProvided: {sys_argv[1:]=}")
+    else:
+        tarfile_path = Path(sys_argv[1])
+        extract_archive(tarfile_path)
+
+
+if __name__ == "__main__":
+    main(sys.argv)

fractal_server-2.3.4/fractal_server/app/runner/run_subprocess.py

@@ -0,0 +1,27 @@
+import shlex
+import subprocess  # nosec
+from typing import Optional
+
+from fractal_server.logger import get_logger
+
+
+def run_subprocess(
+    cmd: str, logger_name: Optional[str] = None
+) -> subprocess.CompletedProcess:
+    logger = get_logger(logger_name)
+    try:
+        res = subprocess.run(  # nosec
+            shlex.split(cmd), check=True, capture_output=True, encoding="utf-8"
+        )
+        return res
+    except subprocess.CalledProcessError as e:
+        logger.debug(
+            f"Command '{e.cmd}' returned non-zero exit status {e.returncode}."
+        )
+        logger.debug(f"stdout: {e.stdout}")
+        logger.debug(f"stderr: {e.stderr}")
+        raise e
+    except Exception as e:
+        logger.debug(f"An error occurred while running command: {cmd}")
+        logger.debug(str(e))
+        raise e

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/config.py

@@ -497,7 +497,7 @@ class Settings(BaseSettings):
     [`clusterfutures`](https://github.com/sampsyo/clusterfutures/blob/master/cfut/__init__.py)).
     """
 
-    FRACTAL_SLURM_SBATCH_SLEEP: int = 0
+    FRACTAL_SLURM_SBATCH_SLEEP: float = 0
     """
     Interval to wait (in seconds) between two subsequent `sbatch` calls, when
     running a task that produces multiple SLURM jobs.
@@ -546,7 +546,9 @@ class Settings(BaseSettings):
     attribute in their input-arguments JSON file.
     """
 
-    FRACTAL_API_V1_MODE: Literal["include", "exclude"] = "include"
+    FRACTAL_API_V1_MODE: Literal[
+        "include", "include_without_submission", "exclude"
+    ] = "include"
     """
     Whether to include the v1 API.
     """
@@ -685,6 +687,25 @@ class Settings(BaseSettings):
         self.check_db()
         self.check_runner()
 
+    def get_sanitized(self) -> dict:
+        def _must_be_sanitized(string) -> bool:
+            if not string.upper().startswith("FRACTAL") or any(
+                s in string.upper()
+                for s in ["PASSWORD", "SECRET", "PWD", "TOKEN"]
+            ):
+                return True
+            else:
+                return False
+
+        sanitized_settings = {}
+        for k, v in self.dict().items():
+            if _must_be_sanitized(k):
+                sanitized_settings[k] = "***"
+            else:
+                sanitized_settings[k] = v
+
+        return sanitized_settings
+
 
 def get_settings(settings=Settings()) -> Settings:
     return settings

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/main.py

@@ -49,7 +49,7 @@ def collect_routers(app: FastAPI) -> None:
     settings = Inject(get_settings)
 
     app.include_router(router_api, prefix="/api")
-    if settings.FRACTAL_API_V1_MODE == "include":
+    if settings.FRACTAL_API_V1_MODE.startswith("include"):
         app.include_router(router_api_v1, prefix="/api/v1")
         app.include_router(
             router_admin_v1, prefix="/admin/v1", tags=["V1 Admin area"]

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/ssh/_fabric.py

@@ -286,7 +286,6 @@ class FractalSSH(object):
             folder: Absolute path to a folder that should be removed.
             safe_root: If `folder` is not a subfolder of the absolute
                 `safe_root` path, raise an error.
-            fractal_ssh:
         """
         invalid_characters = {" ", "\n", ";", "$", "`"}
 

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/string_tools.py

@@ -9,12 +9,12 @@ def sanitize_string(value: str) -> str:
 
     Make the string lower-case, and replace any special character with an
     underscore, where special characters are:
-    ```python repl
-    >>> string.punctuation
-    '!"#$%&\'()*+,-./:;<=>?@[\\]^_`{|}~'
-    >>> string.whitespace
-    ' \t\n\r\x0b\x0c'
-    ```
+
+
+        >>> string.punctuation
+        '!"#$%&\'()*+,-./:;<=>?@[\\\\]^_`{|}~'
+        >>> string.whitespace
+        ' \\t\\n\\r\\x0b\\x0c'
 
     Args:
         value: Input string
@@ -35,5 +35,11 @@ def slugify_task_name_for_source(task_name: str) -> str:
     from `fractal_server.string_tools.sanitize_string`, nor we can remove it.
     As 2.3.1, we are renaming it to `slugify_task_name_for_source`, to make
     it clear that it should not be used for other purposes.
+
+    Args:
+        task_name:
+
+    Return:
+        Slug-ified task name.
     """
     return task_name.replace(" ", "_").lower()

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/tasks/v2/background_operations_ssh.py

@@ -167,7 +167,7 @@ def background_collect_pip_ssh(
                     / ".fractal"
                     / f"{task_pkg.package_name}{package_version}"
                 ).as_posix()
-
+                logger.debug(f"{package_env_dir=}")
                 replacements = [
                     ("__PACKAGE_NAME__", task_pkg.package_name),
                     ("__PACKAGE_ENV_DIR__", package_env_dir),
@@ -197,10 +197,18 @@ def background_collect_pip_ssh(
                 # long operations that do not use the db
                 db.close()
 
+                # `remove_venv_folder_upon_failure` is set to True only if
+                # script 1 goes through, which means that the remote folder
+                # `package_env_dir` did not already exist. If this remote
+                # folder already existed, then script 1 fails and the boolean
+                # flag `remove_venv_folder_upon_failure` remains false.
+                remove_venv_folder_upon_failure = False
                 stdout = _customize_and_run_template(
                     script_filename="_1_create_venv.sh",
                     **common_args,
                 )
+                remove_venv_folder_upon_failure = True
+
                 stdout = _customize_and_run_template(
                     script_filename="_2_upgrade_pip.sh",
                     **common_args,
@@ -294,6 +302,7 @@ def background_collect_pip_ssh(
 
                 # Finalize (write metadata to DB)
                 logger.debug("finalising - START")
+
                 collection_state = db.get(CollectionStateV2, state_id)
                 collection_state.data["log"] = log_file_path.open("r").read()
                 collection_state.data["freeze"] = stdout_pip_freeze
@@ -312,16 +321,26 @@ def background_collect_pip_ssh(
                     exception=e,
                     db=db,
                 )
-                try:
-                    logger.info(f"Now delete remote folder {package_env_dir}")
-                    fractal_ssh.remove_folder(
-                        folder=package_env_dir,
-                        safe_root=settings.FRACTAL_SLURM_SSH_WORKING_BASE_DIR,
-                    )
-                    logger.info(f"Deleted remoted folder {package_env_dir}")
-                except Exception as e:
-                    logger.error(
-                        f"Deleting remote folder failed.\n"
-                        f"Original error:\n{str(e)}"
-                    )
+                if remove_venv_folder_upon_failure:
+                    try:
+                        logger.info(
+                            f"Now delete remote folder {package_env_dir}"
+                        )
+                        fractal_ssh.remove_folder(
+                            folder=package_env_dir,
+                            safe_root=settings.FRACTAL_SLURM_SSH_WORKING_BASE_DIR,  # noqa: E501
+                        )
+                        logger.info(
+                            f"Deleted remoted folder {package_env_dir}"
+                        )
+                    except Exception as e:
+                        logger.error(
+                            f"Removing remote folder failed.\n"
+                            f"Original error:\n{str(e)}"
+                        )
+                    else:
+                        logger.info(
+                            "Not trying to remove remote folder "
+                            f"{package_env_dir}."
+                        )
                 return

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/tasks/v2/templates/_1_create_venv.sh

@@ -12,12 +12,12 @@ PYTHON=__PYTHON__
 
 TIME_START=$(date +%s)
 
-
-# Create main folder
+# Check that package folder does not exist
 if [ -d "$PACKAGE_ENV_DIR" ]; then
     write_log "ERROR: Folder $PACKAGE_ENV_DIR already exists. Exit."
     exit 1
 fi
+
 write_log "START mkdir -p $PACKAGE_ENV_DIR"
 mkdir -p $PACKAGE_ENV_DIR
 write_log "END   mkdir -p $PACKAGE_ENV_DIR"

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/tasks/v2/templates/_2_upgrade_pip.sh

@@ -7,10 +7,6 @@ write_log(){
 
 # Variables to be filled within fractal-server
 PACKAGE_ENV_DIR=__PACKAGE_ENV_DIR__
-PACKAGE_NAME=__PACKAGE_NAME__
-PACKAGE=__PACKAGE__
-PYTHON=__PYTHON__
-INSTALL_STRING=__INSTALL_STRING__
 
 TIME_START=$(date +%s)
 

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/tasks/v2/templates/_3_pip_install.sh

@@ -8,12 +8,8 @@ write_log(){
 
 # Variables to be filled within fractal-server
 PACKAGE_ENV_DIR=__PACKAGE_ENV_DIR__
-PACKAGE_NAME=__PACKAGE_NAME__
-PACKAGE=__PACKAGE__
-PYTHON=__PYTHON__
 INSTALL_STRING=__INSTALL_STRING__
 
-
 TIME_START=$(date +%s)
 
 VENVPYTHON=${PACKAGE_ENV_DIR}/bin/python

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/tasks/v2/templates/_4_pip_freeze.sh

@@ -9,12 +9,6 @@ write_log(){
 
 # Variables to be filled within fractal-server
 PACKAGE_ENV_DIR=__PACKAGE_ENV_DIR__
-PACKAGE_NAME=__PACKAGE_NAME__
-PACKAGE=__PACKAGE__
-PYTHON=__PYTHON__
-INSTALL_STRING=__INSTALL_STRING__
-
-
 
 VENVPYTHON=${PACKAGE_ENV_DIR}/bin/python
 

{fractal_server-2.3.2 → fractal_server-2.3.4}/fractal_server/tasks/v2/templates/_5_pip_show.sh

@@ -9,9 +9,7 @@ write_log(){
 # Variables to be filled within fractal-server
 PACKAGE_ENV_DIR=__PACKAGE_ENV_DIR__
 PACKAGE_NAME=__PACKAGE_NAME__
-PACKAGE=__PACKAGE__
-PYTHON=__PYTHON__
-INSTALL_STRING=__INSTALL_STRING__
+
 
 
 TIME_START=$(date +%s)

{fractal_server-2.3.2 → fractal_server-2.3.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "fractal-server"
-version = "2.3.2"
+version = "2.3.4"
 description = "Server component of the Fractal analytics platform"
 authors = [
     "Tommaso Comparin <tommaso.comparin@exact-lab.it>",
@@ -91,7 +91,7 @@ filterwarnings = [
 ]
 
 [tool.bumpver]
-current_version = "2.3.2"
+current_version = "2.3.4"
 version_pattern = "MAJOR.MINOR.PATCH[PYTAGNUM]"
 commit_message = "bump version {old_version} -> {new_version}"
 commit = true

fractal_server-2.3.2/fractal_server/__init__.py

@@ -1 +0,0 @@
-__VERSION__ = "2.3.2"