fractal-server 2.14.0a12__py3-none-any.whl → 2.14.0a14__py3-none-any.whl

fractal_server/app/runner/task_files.py

@@ -34,6 +34,7 @@ class TaskFiles(BaseModel):
 
     # Per-single-component
     component: Optional[str] = None
+    prefix: Optional[str] = None
 
     def _check_component(self):
         if self.component is None:
@@ -55,46 +56,55 @@ class TaskFiles(BaseModel):
     def wftask_subfolder_local(self) -> Path:
         return self.root_dir_local / self.subfolder_name
 
+    @property
+    def prefix_component(self):
+        if self.prefix is None:
+            return self.component
+        else:
+            return f"{self.prefix}-{self.component}"
+
     @property
     def log_file_local(self) -> str:
         self._check_component()
         return (
-            self.wftask_subfolder_local / f"{self.component}-log.txt"
+            self.wftask_subfolder_local / f"{self.prefix_component}-log.txt"
         ).as_posix()
 
     @property
     def log_file_remote(self) -> str:
         self._check_component()
         return (
-            self.wftask_subfolder_remote / f"{self.component}-log.txt"
+            self.wftask_subfolder_remote / f"{self.prefix_component}-log.txt"
         ).as_posix()
 
     @property
     def args_file_local(self) -> str:
         self._check_component()
         return (
-            self.wftask_subfolder_local / f"{self.component}-args.json"
+            self.wftask_subfolder_local / f"{self.prefix_component}-args.json"
         ).as_posix()
 
     @property
     def args_file_remote(self) -> str:
         self._check_component()
         return (
-            self.wftask_subfolder_remote / f"{self.component}-args.json"
+            self.wftask_subfolder_remote / f"{self.prefix_component}-args.json"
         ).as_posix()
 
     @property
     def metadiff_file_local(self) -> str:
         self._check_component()
         return (
-            self.wftask_subfolder_local / f"{self.component}-metadiff.json"
+            self.wftask_subfolder_local
+            / f"{self.prefix_component}-metadiff.json"
         ).as_posix()
 
     @property
     def metadiff_file_remote(self) -> str:
         self._check_component()
         return (
-            self.wftask_subfolder_remote / f"{self.component}-metadiff.json"
+            self.wftask_subfolder_remote
+            / f"{self.prefix_component}-metadiff.json"
         ).as_posix()
 
     @property
@@ -104,3 +114,7 @@ class TaskFiles(BaseModel):
             metadiff_file_remote=self.metadiff_file_remote,
             log_file_remote=self.log_file_remote,
         )
+
+
+SUBMIT_PREFIX = "non_par"
+MULTISUBMIT_PREFIX = "par"

fractal_server/app/runner/v2/_slurm_ssh.py

@@ -24,7 +24,7 @@ from ...models.v2 import DatasetV2
 from ...models.v2 import WorkflowV2
 from ..exceptions import JobExecutionError
 from ..executors.slurm_common.get_slurm_config import get_slurm_config
-from ..executors.slurm_ssh.executor import FractalSlurmSSHExecutor
+from ..executors.slurm_ssh.runner import SlurmSSHRunner
 from ..set_start_and_last_task_index import set_start_and_last_task_index
 from .runner import execute_tasks_v2
 from fractal_server.images.models import AttributeFiltersType
@@ -76,18 +76,18 @@ def process_workflow(
         logger.error(error_msg)
         raise JobExecutionError(info=error_msg)
 
-    with FractalSlurmSSHExecutor(
+    with SlurmSSHRunner(
         fractal_ssh=fractal_ssh,
-        workflow_dir_local=workflow_dir_local,
-        workflow_dir_remote=workflow_dir_remote,
+        root_dir_local=workflow_dir_local,
+        root_dir_remote=workflow_dir_remote,
         common_script_lines=worker_init,
-    ) as executor:
+    ) as runner:
         execute_tasks_v2(
             wf_task_list=workflow.task_list[
                 first_task_index : (last_task_index + 1)
             ],
             dataset=dataset,
-            runner=executor,
+            runner=runner,
             workflow_dir_local=workflow_dir_local,
             workflow_dir_remote=workflow_dir_remote,
             logger_name=logger_name,

fractal_server/app/runner/v2/_slurm_sudo.py

@@ -22,7 +22,7 @@ from typing import Optional
 from ...models.v2 import DatasetV2
 from ...models.v2 import WorkflowV2
 from ..executors.slurm_common.get_slurm_config import get_slurm_config
-from ..executors.slurm_sudo.runner import RunnerSlurmSudo
+from ..executors.slurm_sudo.runner import SudoSlurmRunner
 from ..set_start_and_last_task_index import set_start_and_last_task_index
 from .runner import execute_tasks_v2
 from fractal_server.images.models import AttributeFiltersType
@@ -66,20 +66,20 @@ def process_workflow(
     if isinstance(worker_init, str):
         worker_init = worker_init.split("\n")
 
-    with RunnerSlurmSudo(
+    with SudoSlurmRunner(
         slurm_user=slurm_user,
         user_cache_dir=user_cache_dir,
         root_dir_local=workflow_dir_local,
         root_dir_remote=workflow_dir_remote,
         common_script_lines=worker_init,
         slurm_account=slurm_account,
-    ) as executor:
+    ) as runner:
         execute_tasks_v2(
             wf_task_list=workflow.task_list[
                 first_task_index : (last_task_index + 1)
             ],
             dataset=dataset,
-            runner=executor,
+            runner=runner,
             workflow_dir_local=workflow_dir_local,
             workflow_dir_remote=workflow_dir_remote,
             logger_name=logger_name,

fractal_server/app/runner/v2/db_tools.py

@@ -85,3 +85,4 @@ def bulk_upsert_image_cache_fast(
             ),
         )
         db.execute(stmt)
+        db.commit()

fractal_server/app/runner/v2/runner.py

@@ -126,6 +126,10 @@ def execute_tasks_v2(
             db.commit()
             db.refresh(history_run)
             history_run_id = history_run.id
+            logger.debug(
+                f"Created {history_run_id=}, for "
+                f"{wftask.id=} and {dataset.id=}"
+            )
 
         # TASK EXECUTION (V2)
         if task.type in ["non_parallel", "converter_non_parallel"]:

fractal_server/app/runner/v2/runner_functions.py

@@ -372,7 +372,7 @@ def run_v2_task_compound(
         root_dir_remote=workflow_dir_remote,
         task_order=wftask.order,
         task_name=wftask.task.name,
-        component=f"init_{_index_to_component(0)}",
+        component=_index_to_component(0),
     )
 
     runner_config_init = get_runner_config(
@@ -483,7 +483,7 @@ def run_v2_task_compound(
             root_dir_remote=workflow_dir_remote,
             task_order=wftask.order,
             task_name=wftask.task.name,
-            component=f"compute_{_index_to_component(ind)}",
+            component=_index_to_component(ind),
         )
         for ind in range(len(parallelization_list))
     ]

fractal_server/app/runner/v2/submit_workflow.py

@@ -27,7 +27,6 @@ from ...models.v2 import WorkflowV2
 from ...schemas.v2 import JobStatusTypeV2
 from ..exceptions import JobExecutionError
 from ..exceptions import TaskExecutionError
-from ..executors.slurm_sudo._subprocess_run_as_user import _mkdir_as_user
 from ..filenames import WORKFLOW_LOG_FILENAME
 from ._local import process_workflow as local_process_workflow
 from ._slurm_ssh import process_workflow as slurm_ssh_process_workflow
@@ -170,35 +169,27 @@ def submit_workflow(
             return
 
         try:
-            # Create WORKFLOW_DIR_LOCAL
-            if FRACTAL_RUNNER_BACKEND == "slurm":
-                original_umask = os.umask(0)
-                WORKFLOW_DIR_LOCAL.mkdir(parents=True, mode=0o755)
-                os.umask(original_umask)
-            else:
-                WORKFLOW_DIR_LOCAL.mkdir(parents=True)
-
-            # Define and create WORKFLOW_DIR_REMOTE
+            # Create WORKFLOW_DIR_LOCAL and define WORKFLOW_DIR_REMOTE
             if FRACTAL_RUNNER_BACKEND == "local":
+                WORKFLOW_DIR_LOCAL.mkdir(parents=True)
                 WORKFLOW_DIR_REMOTE = WORKFLOW_DIR_LOCAL
             elif FRACTAL_RUNNER_BACKEND == "slurm":
+                original_umask = os.umask(0)
+                WORKFLOW_DIR_LOCAL.mkdir(parents=True, mode=0o755)
+                os.umask(original_umask)
                 WORKFLOW_DIR_REMOTE = (
                     Path(user_cache_dir) / WORKFLOW_DIR_LOCAL.name
                 )
-                _mkdir_as_user(
-                    folder=str(WORKFLOW_DIR_REMOTE), user=slurm_user
-                )
             elif FRACTAL_RUNNER_BACKEND == "slurm_ssh":
-                # Folder creation is deferred to _process_workflow
+                WORKFLOW_DIR_LOCAL.mkdir(parents=True)
                 WORKFLOW_DIR_REMOTE = (
                     Path(user_settings.ssh_jobs_dir) / WORKFLOW_DIR_LOCAL.name
                 )
             else:
-                logger.error(
+                raise ValueError(
                     "Invalid FRACTAL_RUNNER_BACKEND="
                     f"{settings.FRACTAL_RUNNER_BACKEND}."
                 )
-
         except Exception as e:
             error_type = type(e).__name__
             fail_job(

fractal_server/app/schemas/v2/__init__.py

@@ -10,11 +10,13 @@ from .dumps import TaskDumpV2  # noqa F401
 from .dumps import TaskGroupDumpV2  # noqa F401
 from .dumps import WorkflowDumpV2  # noqa F401
 from .dumps import WorkflowTaskDumpV2  # noqa F401
+from .history import HistoryRunRead  # noqa F401
 from .history import HistoryRunReadAggregated  # noqa F401
 from .history import HistoryUnitRead  # noqa F401
 from .history import HistoryUnitStatus  # noqa F401
+from .history import HistoryUnitStatusQuery  # noqa F401
 from .history import ImageLogsRequest  # noqa F401
-from .history import ZarrUrlAndStatus  # noqa F401
+from .history import SingleImageWithStatus  # noqa F401
 from .job import JobCreateV2  # noqa F401
 from .job import JobReadV2  # noqa F401
 from .job import JobStatusTypeV2  # noqa F401

fractal_server/app/schemas/v2/history.py

@@ -7,6 +7,8 @@ from pydantic import AwareDatetime
 from pydantic import BaseModel
 from pydantic import field_serializer
 
+from ....images import SingleImage
+
 
 class HistoryUnitStatus(str, Enum):
     """
@@ -23,6 +25,15 @@ class HistoryUnitStatus(str, Enum):
     FAILED = "failed"
 
 
+class HistoryUnitStatusQuery(str, Enum):
+
+    SUBMITTED = "submitted"
+    DONE = "done"
+    FAILED = "failed"
+
+    UNSET = "unset"
+
+
 class HistoryUnitRead(BaseModel):
     id: int
     logfile: Optional[str] = None
@@ -30,6 +41,21 @@ class HistoryUnitRead(BaseModel):
     zarr_urls: list[str]
 
 
+class HistoryRunRead(BaseModel):
+    id: int
+    dataset_id: int
+    workflowtask_id: Optional[int] = None
+    workflowtask_dump: dict[str, Any]
+    task_group_dump: dict[str, Any]
+    timestamp_started: AwareDatetime
+    status: HistoryUnitStatus
+    num_available_images: int
+
+    @field_serializer("timestamp_started")
+    def serialize_datetime(v: datetime) -> str:
+        return v.isoformat()
+
+
 class HistoryRunReadAggregated(BaseModel):
     id: int
     timestamp_started: AwareDatetime
@@ -49,6 +75,5 @@ class ImageLogsRequest(BaseModel):
     zarr_url: str
 
 
-class ZarrUrlAndStatus(BaseModel):
-    zarr_url: str
+class SingleImageWithStatus(SingleImage):
     status: Optional[HistoryUnitStatus] = None

fractal_server/config.py

@@ -489,8 +489,12 @@ class Settings(BaseSettings):
     FRACTAL_SLURM_POLL_INTERVAL: int = 5
     """
     Interval to wait (in seconds) before checking whether unfinished job are
-    still running on SLURM (see `SlurmWaitThread` in
-    [`clusterfutures`](https://github.com/sampsyo/clusterfutures/blob/master/cfut/__init__.py)).
+    still running on SLURM.
+    """
+
+    FRACTAL_SLURM_INTERVAL_BEFORE_RETRIEVAL: int = 2
+    """
+    FIXME: this is a workaround, we are still investigating.
     """
 
     FRACTAL_SLURM_SBATCH_SLEEP: float = 0

fractal_server/images/tools.py

@@ -121,3 +121,26 @@ def merge_type_filters(
     merged_dict = task_input_types
     merged_dict.update(wftask_type_filters)
     return merged_dict
+
+
+def aggregate_attributes(images: list[dict[str, Any]]) -> dict[str, list[Any]]:
+    """
+    Given a list of images, this function returns a dictionary of all image
+    attributes, each mapped to a list of present values.
+    """
+    attributes = {}
+    for image in images:
+        for k, v in image["attributes"].items():
+            attributes.setdefault(k, []).append(v)
+        for k, v in attributes.items():
+            attributes[k] = list(set(v))
+    return attributes
+
+
+def aggregate_types(images: list[dict[str, Any]]) -> list[str]:
+    """
+    Given a list of images, this function returns a list of all image types.
+    """
+    return list(
+        set(type for image in images for type in image["types"].keys())
+    )

fractal_server/migrations/versions/5b6007027595_on_cascade.py

@@ -0,0 +1,250 @@
+"""on cascade
+
+Revision ID: 5b6007027595
+Revises: af1ef1c83c9b
+Create Date: 2025-04-02 17:03:59.542921
+
+"""
+from alembic import op
+
+
+# revision identifiers, used by Alembic.
+revision = "5b6007027595"
+down_revision = "af1ef1c83c9b"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table("datasetv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            "fk_datasetv2_project_id_projectv2", type_="foreignkey"
+        )
+        batch_op.create_foreign_key(
+            batch_op.f("fk_datasetv2_project_id_projectv2"),
+            "projectv2",
+            ["project_id"],
+            ["id"],
+            ondelete="CASCADE",
+        )
+
+    with op.batch_alter_table("jobv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            "fk_jobv2_dataset_id_datasetv2", type_="foreignkey"
+        )
+        batch_op.drop_constraint(
+            "fk_jobv2_workflow_id_workflowv2", type_="foreignkey"
+        )
+        batch_op.drop_constraint(
+            "fk_jobv2_project_id_projectv2", type_="foreignkey"
+        )
+        batch_op.create_foreign_key(
+            batch_op.f("fk_jobv2_project_id_projectv2"),
+            "projectv2",
+            ["project_id"],
+            ["id"],
+            ondelete="SET NULL",
+        )
+        batch_op.create_foreign_key(
+            batch_op.f("fk_jobv2_dataset_id_datasetv2"),
+            "datasetv2",
+            ["dataset_id"],
+            ["id"],
+            ondelete="SET NULL",
+        )
+        batch_op.create_foreign_key(
+            batch_op.f("fk_jobv2_workflow_id_workflowv2"),
+            "workflowv2",
+            ["workflow_id"],
+            ["id"],
+            ondelete="SET NULL",
+        )
+
+    with op.batch_alter_table("linkusergroup", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            "fk_linkusergroup_group_id_usergroup", type_="foreignkey"
+        )
+        batch_op.drop_constraint(
+            "fk_linkusergroup_user_id_user_oauth", type_="foreignkey"
+        )
+        batch_op.create_foreign_key(
+            batch_op.f("fk_linkusergroup_group_id_usergroup"),
+            "usergroup",
+            ["group_id"],
+            ["id"],
+            ondelete="CASCADE",
+        )
+        batch_op.create_foreign_key(
+            batch_op.f("fk_linkusergroup_user_id_user_oauth"),
+            "user_oauth",
+            ["user_id"],
+            ["id"],
+            ondelete="CASCADE",
+        )
+
+    with op.batch_alter_table("taskgroupactivityv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            "fk_taskgroupactivityv2_taskgroupv2_id_taskgroupv2",
+            type_="foreignkey",
+        )
+        batch_op.create_foreign_key(
+            batch_op.f("fk_taskgroupactivityv2_taskgroupv2_id_taskgroupv2"),
+            "taskgroupv2",
+            ["taskgroupv2_id"],
+            ["id"],
+            ondelete="SET NULL",
+        )
+
+    with op.batch_alter_table("taskgroupv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            "fk_taskgroupv2_user_group_id_usergroup", type_="foreignkey"
+        )
+        batch_op.create_foreign_key(
+            batch_op.f("fk_taskgroupv2_user_group_id_usergroup"),
+            "usergroup",
+            ["user_group_id"],
+            ["id"],
+            ondelete="SET NULL",
+        )
+
+    with op.batch_alter_table("workflowtaskv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            "fk_workflowtaskv2_workflow_id_workflowv2", type_="foreignkey"
+        )
+        batch_op.create_foreign_key(
+            batch_op.f("fk_workflowtaskv2_workflow_id_workflowv2"),
+            "workflowv2",
+            ["workflow_id"],
+            ["id"],
+            ondelete="CASCADE",
+        )
+
+    with op.batch_alter_table("workflowv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            "fk_workflowv2_project_id_projectv2", type_="foreignkey"
+        )
+        batch_op.create_foreign_key(
+            batch_op.f("fk_workflowv2_project_id_projectv2"),
+            "projectv2",
+            ["project_id"],
+            ["id"],
+            ondelete="CASCADE",
+        )
+
+    # ### end Alembic commands ###
+
+
+def downgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table("workflowv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            batch_op.f("fk_workflowv2_project_id_projectv2"),
+            type_="foreignkey",
+        )
+        batch_op.create_foreign_key(
+            "fk_workflowv2_project_id_projectv2",
+            "projectv2",
+            ["project_id"],
+            ["id"],
+        )
+
+    with op.batch_alter_table("workflowtaskv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            batch_op.f("fk_workflowtaskv2_workflow_id_workflowv2"),
+            type_="foreignkey",
+        )
+        batch_op.create_foreign_key(
+            "fk_workflowtaskv2_workflow_id_workflowv2",
+            "workflowv2",
+            ["workflow_id"],
+            ["id"],
+        )
+
+    with op.batch_alter_table("taskgroupv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            batch_op.f("fk_taskgroupv2_user_group_id_usergroup"),
+            type_="foreignkey",
+        )
+        batch_op.create_foreign_key(
+            "fk_taskgroupv2_user_group_id_usergroup",
+            "usergroup",
+            ["user_group_id"],
+            ["id"],
+        )
+
+    with op.batch_alter_table("taskgroupactivityv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            batch_op.f("fk_taskgroupactivityv2_taskgroupv2_id_taskgroupv2"),
+            type_="foreignkey",
+        )
+        batch_op.create_foreign_key(
+            "fk_taskgroupactivityv2_taskgroupv2_id_taskgroupv2",
+            "taskgroupv2",
+            ["taskgroupv2_id"],
+            ["id"],
+        )
+
+    with op.batch_alter_table("linkusergroup", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            batch_op.f("fk_linkusergroup_user_id_user_oauth"),
+            type_="foreignkey",
+        )
+        batch_op.drop_constraint(
+            batch_op.f("fk_linkusergroup_group_id_usergroup"),
+            type_="foreignkey",
+        )
+        batch_op.create_foreign_key(
+            "fk_linkusergroup_user_id_user_oauth",
+            "user_oauth",
+            ["user_id"],
+            ["id"],
+        )
+        batch_op.create_foreign_key(
+            "fk_linkusergroup_group_id_usergroup",
+            "usergroup",
+            ["group_id"],
+            ["id"],
+        )
+
+    with op.batch_alter_table("jobv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            batch_op.f("fk_jobv2_workflow_id_workflowv2"), type_="foreignkey"
+        )
+        batch_op.drop_constraint(
+            batch_op.f("fk_jobv2_dataset_id_datasetv2"), type_="foreignkey"
+        )
+        batch_op.drop_constraint(
+            batch_op.f("fk_jobv2_project_id_projectv2"), type_="foreignkey"
+        )
+        batch_op.create_foreign_key(
+            "fk_jobv2_project_id_projectv2",
+            "projectv2",
+            ["project_id"],
+            ["id"],
+        )
+        batch_op.create_foreign_key(
+            "fk_jobv2_workflow_id_workflowv2",
+            "workflowv2",
+            ["workflow_id"],
+            ["id"],
+        )
+        batch_op.create_foreign_key(
+            "fk_jobv2_dataset_id_datasetv2",
+            "datasetv2",
+            ["dataset_id"],
+            ["id"],
+        )
+
+    with op.batch_alter_table("datasetv2", schema=None) as batch_op:
+        batch_op.drop_constraint(
+            batch_op.f("fk_datasetv2_project_id_projectv2"), type_="foreignkey"
+        )
+        batch_op.create_foreign_key(
+            "fk_datasetv2_project_id_projectv2",
+            "projectv2",
+            ["project_id"],
+            ["id"],
+        )
+
+    # ### end Alembic commands ###

fractal_server/migrations/versions/fbce16ff4e47_new_history_items.py

@@ -1,7 +1,7 @@
 """new history items
 
 Revision ID: fbce16ff4e47
-Revises: af1ef1c83c9b
+Revises: 5b6007027595
 Create Date: 2025-03-14 15:25:01.083619
 
 """
@@ -12,7 +12,7 @@ from sqlalchemy.dialects import postgresql
 
 # revision identifiers, used by Alembic.
 revision = "fbce16ff4e47"
-down_revision = "af1ef1c83c9b"
+down_revision = "5b6007027595"
 branch_labels = None
 depends_on = None
 

fractal_server/tasks/v2/utils_background.py

@@ -3,7 +3,6 @@ from typing import Optional
 from typing import TypeVar
 
 from sqlalchemy.orm import Session as DBSyncSession
-from sqlmodel import select
 
 from fractal_server.app.models.v2 import TaskGroupActivityV2
 from fractal_server.app.models.v2 import TaskGroupV2
@@ -44,24 +43,6 @@ def fail_and_cleanup(
     task_group_activity.log = get_current_log(log_file_path)
     task_group_activity = add_commit_refresh(obj=task_group_activity, db=db)
     if task_group_activity.action == TaskGroupActivityActionV2.COLLECT:
-        logger.info(f"Now delete TaskGroupV2 with {task_group.id=}")
-
-        logger.info("Start of TaskGroupActivityV2 cascade operations.")
-        stm = select(TaskGroupActivityV2).where(
-            TaskGroupActivityV2.taskgroupv2_id == task_group.id
-        )
-        res = db.execute(stm)
-        task_group_activity_list = res.scalars().all()
-        for task_group_activity in task_group_activity_list:
-            logger.info(
-                f"Setting TaskGroupActivityV2[{task_group_activity.id}]"
-                ".taskgroupv2_id to None."
-            )
-            task_group_activity.taskgroupv2_id = None
-            db.add(task_group_activity)
-        logger.info("End of TaskGroupActivityV2 cascade operations.")
-        logger.info(f"TaskGroupV2 with {task_group.id=} deleted")
-
         db.delete(task_group)
     db.commit()
     reset_logger_handlers(logger)

{fractal_server-2.14.0a12.dist-info → fractal_server-2.14.0a14.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: fractal-server
-Version: 2.14.0a12
+Version: 2.14.0a14
 Summary: Backend component of the Fractal analytics platform
 License: BSD-3-Clause
 Author: Tommaso Comparin