fractal-server 2.14.6__tar.gz → 2.14.8__tar.gz

{fractal_server-2.14.6 → fractal_server-2.14.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: fractal-server
-Version: 2.14.6
+Version: 2.14.8
 Summary: Backend component of the Fractal analytics platform
 License: BSD-3-Clause
 Author: Tommaso Comparin

fractal_server-2.14.8/fractal_server/__init__.py

@@ -0,0 +1 @@
+__VERSION__ = "2.14.8"

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/models/v2/history.py

@@ -27,6 +27,7 @@ class HistoryRun(SQLModel, table=True):
         ondelete="SET NULL",
     )
     job_id: int = Field(foreign_key="jobv2.id")
+    task_id: int | None = Field(foreign_key="taskv2.id", ondelete="SET NULL")
 
     workflowtask_dump: dict[str, Any] = Field(
         sa_column=Column(JSONB, nullable=False),

fractal_server-2.14.8/fractal_server/app/routes/api/v2/_aux_functions_task_version_update.py

@@ -0,0 +1,42 @@
+from typing import Any
+
+
+def get_new_workflow_task_meta(
+    *,
+    old_workflow_task_meta: dict | None,
+    old_task_meta: dict | None,
+    new_task_meta: dict | None,
+) -> dict[str, Any]:
+    """
+    Prepare new meta field based on old/new tasks and old workflow task.
+    """
+
+    # When the whole `old_workflow_task_meta` is user-provided, use it
+    # as the outcome
+    if old_task_meta is None:
+        return old_workflow_task_meta
+
+    # When `old_workflow_task_meta` is unset, use the new-task meta as default.
+    if old_workflow_task_meta is None:
+        return new_task_meta
+
+    if new_task_meta is None:
+        new_task_meta = {}
+
+    # Find properties that were added to the old defaults
+    additions = {
+        k: v
+        for k, v in old_workflow_task_meta.items()
+        if v != old_task_meta.get(k)
+    }
+    # Find properties that were removed from the old defaults
+    removals = old_task_meta.keys() - old_workflow_task_meta.keys()
+
+    # Add `additions` and remove `removals`.
+    new_workflowtask_meta = {
+        k: v
+        for k, v in (new_task_meta | additions).items()
+        if k not in removals
+    }
+
+    return new_workflowtask_meta

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/routes/api/v2/history.py

@@ -23,6 +23,7 @@ from fractal_server.app.models import UserOAuth
 from fractal_server.app.models.v2 import HistoryImageCache
 from fractal_server.app.models.v2 import HistoryRun
 from fractal_server.app.models.v2 import HistoryUnit
+from fractal_server.app.models.v2 import TaskV2
 from fractal_server.app.routes.auth import current_active_user
 from fractal_server.app.routes.pagination import get_pagination_params
 from fractal_server.app.routes.pagination import PaginationRequest
@@ -31,9 +32,11 @@ from fractal_server.app.schemas.v2 import HistoryRunRead
 from fractal_server.app.schemas.v2 import HistoryRunReadAggregated
 from fractal_server.app.schemas.v2 import HistoryUnitRead
 from fractal_server.app.schemas.v2 import HistoryUnitStatus
-from fractal_server.app.schemas.v2 import HistoryUnitStatusQuery
+from fractal_server.app.schemas.v2 import HistoryUnitStatusWithUnset
 from fractal_server.app.schemas.v2 import ImageLogsRequest
 from fractal_server.app.schemas.v2 import SingleImageWithStatus
+from fractal_server.images.image_status import enrich_image_list
+from fractal_server.images.image_status import IMAGE_STATUS_KEY
 from fractal_server.images.tools import aggregate_attributes
 from fractal_server.images.tools import aggregate_types
 from fractal_server.images.tools import filter_image_list
@@ -60,7 +63,6 @@ def check_historyrun_related_to_dataset_and_wftask(
 
 
 class ImageWithStatusPage(PaginationResponse[SingleImageWithStatus]):
-
     attributes: dict[str, list[Any]]
     types: list[str]
 
@@ -77,7 +79,6 @@ async def get_workflow_tasks_statuses(
     user: UserOAuth = Depends(current_active_user),
     db: AsyncSession = Depends(get_async_db),
 ) -> JSONResponse:
-
     # Access control
     workflow = await _get_workflow_check_owner(
         project_id=project_id,
@@ -150,7 +151,6 @@ async def get_history_run_list(
     user: UserOAuth = Depends(current_active_user),
     db: AsyncSession = Depends(get_async_db),
 ) -> list[HistoryRunReadAggregated]:
-
     # Access control
     await get_wftask_check_owner(
         project_id=project_id,
@@ -199,7 +199,36 @@ async def get_history_run_list(
     for run_id, unit_status, count in unit_counts:
         count_map[run_id][f"num_{unit_status}_units"] = count
 
-    runs = [dict(**run.model_dump(), **count_map[run.id]) for run in runs]
+    res = await db.execute(
+        select(
+            TaskV2.id,
+            TaskV2.version,
+            TaskV2.args_schema_parallel,
+            TaskV2.args_schema_non_parallel,
+        ).where(
+            TaskV2.id.in_(
+                [run.task_id for run in runs if run.task_id is not None]
+            )
+        )
+    )
+
+    task_args = {
+        _id: {
+            "version": version,
+            "args_schema_parallel": parallel,
+            "args_schema_non_parallel": non_parallel,
+        }
+        for _id, version, parallel, non_parallel in res.all()
+    }
+
+    runs = [
+        dict(
+            **run.model_dump(),
+            **count_map[run.id],
+            **task_args.get(run.task_id, {}),
+        )
+        for run in runs
+    ]
 
     return runs
 
@@ -215,7 +244,6 @@ async def get_history_run_units(
     db: AsyncSession = Depends(get_async_db),
     pagination: PaginationRequest = Depends(get_pagination_params),
 ) -> PaginationResponse[HistoryUnitRead]:
-
     # Access control
     await get_wftask_check_owner(
         project_id=project_id,
@@ -271,12 +299,11 @@ async def get_history_images(
     dataset_id: int,
     workflowtask_id: int,
     request_body: ImageQuery,
-    unit_status: HistoryUnitStatusQuery | None = None,
+    unit_status: HistoryUnitStatusWithUnset | None = None,
     user: UserOAuth = Depends(current_active_user),
     db: AsyncSession = Depends(get_async_db),
     pagination: PaginationRequest = Depends(get_pagination_params),
 ) -> ImageWithStatusPage:
-
     # Access control and object retrieval
     wftask = await get_wftask_check_owner(
         project_id=project_id,
@@ -318,99 +345,62 @@ async def get_history_images(
     actual_filters.update(type_filters_patch)
     logger.debug(f"{prefix} {actual_filters=}")
     # (1D) Get all matching images from the dataset
+
     pre_filtered_dataset_images = filter_image_list(
         images=dataset.images,
         type_filters=inferred_dataset_type_filters,
     )
+
+    full_images_list = await enrich_image_list(
+        dataset_id=dataset_id,
+        workflowtask_id=workflowtask_id,
+        images=pre_filtered_dataset_images,
+        db=db,
+    )
+
+    if unit_status is not None:
+        request_body.attribute_filters[IMAGE_STATUS_KEY] = unit_status
+
     filtered_dataset_images = filter_image_list(
-        pre_filtered_dataset_images,
+        full_images_list,
         type_filters=request_body.type_filters,
         attribute_filters=request_body.attribute_filters,
     )
     logger.debug(f"{prefix} {len(dataset.images)=}")
     logger.debug(f"{prefix} {len(filtered_dataset_images)=}")
-    # (1E) Extract the list of URLs for filtered images
-    filtered_dataset_images_url = list(
-        img["zarr_url"] for img in filtered_dataset_images
-    )
-
-    # (2) Get `(zarr_url, status)` pairs for all images that have already
-    # been processed, and
-    # (3) When relevant, find images that have not been processed
-    base_stmt = (
-        select(HistoryImageCache.zarr_url, HistoryUnit.status)
-        .join(HistoryUnit)
-        .where(HistoryImageCache.dataset_id == dataset_id)
-        .where(HistoryImageCache.workflowtask_id == workflowtask_id)
-        .where(HistoryImageCache.latest_history_unit_id == HistoryUnit.id)
-        .where(HistoryImageCache.zarr_url.in_(filtered_dataset_images_url))
-    )
-
-    if unit_status in [HistoryUnitStatusQuery.UNSET, None]:
-        stmt = base_stmt.order_by(HistoryImageCache.zarr_url)
-        res = await db.execute(stmt)
-        list_processed_url_status = res.all()
-        list_processed_url = list(
-            item[0] for item in list_processed_url_status
-        )
-        list_non_processed_url_status = list(
-            (url, None)
-            for url in filtered_dataset_images_url
-            if url not in list_processed_url
-        )
-        if unit_status == HistoryUnitStatusQuery.UNSET:
-            list_processed_url_status = []
-    else:
-        stmt = base_stmt.where(HistoryUnit.status == unit_status).order_by(
-            HistoryImageCache.zarr_url
-        )
-        res = await db.execute(stmt)
-        list_processed_url_status = res.all()
-        list_non_processed_url_status = []
-
-    logger.debug(f"{prefix} {len(list_processed_url_status)=}")
-    logger.debug(f"{prefix} {len(list_non_processed_url_status)=}")
-
-    # (3) Combine outputs from 1 and 2
-    full_list_url_status = (
-        list_processed_url_status + list_non_processed_url_status
-    )
-    logger.debug(f"{prefix} {len(full_list_url_status)=}")
 
     attributes = aggregate_attributes(pre_filtered_dataset_images)
     types = aggregate_types(pre_filtered_dataset_images)
 
-    sorted_list_url_status = sorted(
-        full_list_url_status,
-        key=lambda url_status: url_status[0],
-    )
-    logger.debug(f"{prefix} {len(sorted_list_url_status)=}")
-
     # Final list of objects
 
-    total_count = len(sorted_list_url_status)
+    total_count = len(filtered_dataset_images)
     page_size = pagination.page_size or total_count
-
-    paginated_list_url_status = sorted_list_url_status[
+    sorted_images_list = sorted(
+        filtered_dataset_images,
+        key=lambda image: image["zarr_url"],
+    )
+    paginated_images_list = sorted_images_list[
         (pagination.page - 1) * page_size : pagination.page * page_size
     ]
 
-    # Aggregate information to create 'SingleImageWithStatus'
-    items = [
+    # FIXME: This is only for backwards-compatibility. To remove when we
+    # update the webclient
+    paginated_images_list = [
         {
-            **filtered_dataset_images[
-                filtered_dataset_images_url.index(url_status[0])
-            ],
-            "status": url_status[1],
+            **img,
+            "status": (
+                lambda x: None if x == HistoryUnitStatusWithUnset.UNSET else x
+            )(img["attributes"].pop(IMAGE_STATUS_KEY)),
         }
-        for url_status in paginated_list_url_status
+        for img in paginated_images_list
     ]
 
     return dict(
         current_page=pagination.page,
         page_size=page_size,
         total_count=total_count,
-        items=items,
+        items=paginated_images_list,
         attributes=attributes,
         types=types,
     )

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/routes/api/v2/task_version_update.py

@@ -17,12 +17,12 @@ from ....models import LinkUserGroup
 from ....models.v2 import TaskV2
 from ._aux_functions import _get_workflow_check_owner
 from ._aux_functions import _get_workflow_task_check_owner
+from ._aux_functions_task_version_update import get_new_workflow_task_meta
 from ._aux_functions_tasks import _check_type_filters_compatibility
 from ._aux_functions_tasks import _get_task_group_or_404
 from ._aux_functions_tasks import _get_task_read_access
 from fractal_server.app.models import UserOAuth
 from fractal_server.app.models.v2 import TaskGroupV2
-from fractal_server.app.models.v2 import WorkflowTaskV2
 from fractal_server.app.routes.auth import current_active_user
 from fractal_server.app.schemas.v2 import WorkflowTaskReadV2
 from fractal_server.app.schemas.v2 import WorkflowTaskReplaceV2
@@ -181,7 +181,7 @@ async def replace_workflowtask(
 ) -> WorkflowTaskReadV2:
 
     # Get objects from database
-    old_wftask, workflow = await _get_workflow_task_check_owner(
+    workflow_task, workflow = await _get_workflow_task_check_owner(
         project_id=project_id,
         workflow_id=workflow_id,
         workflow_task_id=workflow_task_id,
@@ -197,14 +197,14 @@ async def replace_workflowtask(
 
     # Preliminary checks
     if not _is_type_update_valid(
-        old_type=old_wftask.task_type,
+        old_type=workflow_task.task_type,
         new_type=new_task.type,
     ):
         raise HTTPException(
             status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
             detail=(
                 "Cannot change task type from "
-                f"{old_wftask.task_type} to {new_task.type}."
+                f"{workflow_task.task_type} to {new_task.type}."
             ),
         )
 
@@ -220,51 +220,27 @@ async def replace_workflowtask(
         )
     _check_type_filters_compatibility(
         task_input_types=new_task.input_types,
-        wftask_type_filters=old_wftask.type_filters,
+        wftask_type_filters=workflow_task.type_filters,
     )
 
-    # Task arguments
-    if replace.args_non_parallel is None:
-        _args_non_parallel = old_wftask.args_non_parallel
-    else:
-        _args_non_parallel = replace.args_non_parallel
-    if replace.args_parallel is None:
-        _args_parallel = old_wftask.args_parallel
-    else:
-        _args_parallel = replace.args_parallel
-
-    # If user's changes to `meta_non_parallel` are compatible with new task,
-    # keep them; else, get `meta_non_parallel` from new task
-    if (
-        old_wftask.meta_non_parallel != old_wftask.task.meta_non_parallel
-    ) and (old_wftask.task.meta_non_parallel == new_task.meta_non_parallel):
-        _meta_non_parallel = old_wftask.meta_non_parallel
-    else:
-        _meta_non_parallel = new_task.meta_non_parallel
-    # Same for `meta_parallel`
-    if (old_wftask.meta_parallel != old_wftask.task.meta_parallel) and (
-        old_wftask.task.meta_parallel == new_task.meta_parallel
-    ):
-        _meta_parallel = old_wftask.meta_parallel
-    else:
-        _meta_parallel = new_task.meta_parallel
-
-    new_workflow_task = WorkflowTaskV2(
-        task_id=new_task.id,
-        task_type=new_task.type,
-        task=new_task,
-        # old-task values
-        type_filters=old_wftask.type_filters,
-        # possibly new values
-        args_non_parallel=_args_non_parallel,
-        args_parallel=_args_parallel,
-        meta_non_parallel=_meta_non_parallel,
-        meta_parallel=_meta_parallel,
+    workflow_task.task_id = new_task.id
+    workflow_task.task_type = new_task.type
+    workflow_task.meta_non_parallel = get_new_workflow_task_meta(
+        old_task_meta=workflow_task.task.meta_non_parallel,
+        old_workflow_task_meta=workflow_task.meta_non_parallel,
+        new_task_meta=new_task.meta_non_parallel,
+    )
+    workflow_task.meta_parallel = get_new_workflow_task_meta(
+        old_task_meta=workflow_task.task.meta_parallel,
+        old_workflow_task_meta=workflow_task.meta_parallel,
+        new_task_meta=new_task.meta_parallel,
     )
+    if replace.args_non_parallel is not None:
+        workflow_task.args_non_parallel = replace.args_non_parallel
+    if replace.args_parallel is not None:
+        workflow_task.args_parallel = replace.args_parallel
 
-    workflow_task_order = old_wftask.order
-    workflow.task_list.remove(old_wftask)
-    workflow.task_list.insert(workflow_task_order, new_workflow_task)
+    db.add(workflow_task)
     await db.commit()
-    await db.refresh(new_workflow_task)
-    return new_workflow_task
+    await db.refresh(workflow_task)
+    return workflow_task

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/runner/executors/slurm_common/base_slurm_runner.py

@@ -73,6 +73,7 @@ class BaseSlurmRunner(BaseRunner):
     jobs: dict[str, SlurmJob]
     python_worker_interpreter: str
     slurm_runner_type: Literal["ssh", "sudo"]
+    slurm_account: str | None = None
 
     def __init__(
         self,
@@ -83,6 +84,7 @@ class BaseSlurmRunner(BaseRunner):
         common_script_lines: list[str] | None = None,
         user_cache_dir: str | None = None,
         poll_interval: int | None = None,
+        slurm_account: str | None = None,
     ):
         self.slurm_runner_type = slurm_runner_type
         self.root_dir_local = root_dir_local
@@ -91,6 +93,7 @@ class BaseSlurmRunner(BaseRunner):
         self._check_slurm_account()
         self.user_cache_dir = user_cache_dir
         self.python_worker_interpreter = python_worker_interpreter
+        self.slurm_account = slurm_account
 
         settings = Inject(get_settings)
 
@@ -185,6 +188,24 @@ class BaseSlurmRunner(BaseRunner):
     ) -> str:
         logger.debug("[_submit_single_sbatch] START")
 
+        # Include SLURM account in `slurm_config`. Note: we make this change
+        # here, rather than exposing a new argument of `get_slurm_config`,
+        # because it's a backend-specific argument while `get_slurm_config` has
+        # a generic interface.
+        if self.slurm_account is not None:
+            slurm_config.account = self.slurm_account
+
+        # Include common_script_lines in extra_lines
+        if len(self.common_script_lines) > 0:
+            logger.debug(
+                f"Add {self.common_script_lines} to "
+                f"{slurm_config.extra_lines=}."
+            )
+            current_extra_lines = slurm_config.extra_lines or []
+            slurm_config.extra_lines = (
+                current_extra_lines + self.common_script_lines
+            )
+
         for task in slurm_job.tasks:
             # Write input file
             if self.slurm_runner_type == "ssh":

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/runner/executors/slurm_ssh/runner.py

@@ -30,6 +30,7 @@ class SlurmSSHRunner(BaseSlurmRunner):
         user_cache_dir: str | None = None,
         poll_interval: int | None = None,
         # Specific
+        slurm_account: str | None = None,
         fractal_ssh: FractalSSH,
     ) -> None:
         """
@@ -49,6 +50,7 @@ class SlurmSSHRunner(BaseSlurmRunner):
             user_cache_dir=user_cache_dir,
             poll_interval=poll_interval,
             python_worker_interpreter=settings.FRACTAL_SLURM_WORKER_PYTHON,
+            slurm_account=slurm_account,
         )
 
     def _mkdir_local_folder(self, folder: str) -> None:

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/runner/executors/slurm_sudo/runner.py

@@ -63,7 +63,6 @@ class SudoSlurmRunner(BaseSlurmRunner):
         """
 
         self.slurm_user = slurm_user
-        self.slurm_account = slurm_account
         settings = Inject(get_settings)
 
         super().__init__(
@@ -76,6 +75,7 @@ class SudoSlurmRunner(BaseSlurmRunner):
             python_worker_interpreter=(
                 settings.FRACTAL_SLURM_WORKER_PYTHON or sys.executable
             ),
+            slurm_account=slurm_account,
         )
 
     def _mkdir_local_folder(self, folder: str) -> None:

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/runner/v2/_slurm_ssh.py

@@ -43,10 +43,11 @@ def process_workflow(
     logger_name: str,
     job_attribute_filters: AttributeFilters,
     job_type_filters: dict[str, bool],
+    user_id: int,
+    # SLURM-ssh-specific
     fractal_ssh: FractalSSH,
+    slurm_account: str | None = None,
     worker_init: str | None = None,
-    user_id: int,
-    **kwargs,  # not used
 ) -> None:
     """
     Process workflow (SLURM backend public interface)
@@ -79,6 +80,7 @@ def process_workflow(
         fractal_ssh=fractal_ssh,
         root_dir_local=workflow_dir_local,
         root_dir_remote=workflow_dir_remote,
+        slurm_account=slurm_account,
         common_script_lines=worker_init,
     ) as runner:
         execute_tasks_v2(

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/runner/v2/_slurm_sudo.py

@@ -39,7 +39,7 @@ def process_workflow(
     job_attribute_filters: AttributeFilters,
     job_type_filters: dict[str, bool],
     user_id: int,
-    # Slurm-specific
+    # SLURM-sudo-specific
     user_cache_dir: str | None = None,
     slurm_user: str | None = None,
     slurm_account: str | None = None,

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/runner/v2/runner.py

@@ -153,6 +153,7 @@ def execute_tasks_v2(
                 dataset_id=dataset.id,
                 workflowtask_id=wftask.id,
                 job_id=job_id,
+                task_id=wftask.task.id,
                 workflowtask_dump=workflowtask_dump,
                 task_group_dump=task_group_dump,
                 num_available_images=num_available_images,

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/runner/v2/submit_workflow.py

@@ -236,6 +236,7 @@ def submit_workflow(
         elif FRACTAL_RUNNER_BACKEND == "slurm_ssh":
             logger.debug(f"ssh_user: {user_settings.ssh_username}")
             logger.debug(f"base dir: {user_settings.ssh_tasks_dir}")
+            logger.debug(f"slurm_account: {job.slurm_account}")
             logger.debug(f"worker_init: {worker_init}")
         logger.debug(f"job.id: {job.id}")
         logger.debug(f"job.working_dir: {job.working_dir}")
@@ -257,7 +258,10 @@ def submit_workflow(
             )
         elif FRACTAL_RUNNER_BACKEND == "slurm_ssh":
             process_workflow = slurm_ssh_process_workflow
-            backend_specific_kwargs = dict(fractal_ssh=fractal_ssh)
+            backend_specific_kwargs = dict(
+                fractal_ssh=fractal_ssh,
+                slurm_account=job.slurm_account,
+            )
         else:
             raise RuntimeError(
                 f"Invalid runner backend {FRACTAL_RUNNER_BACKEND=}"

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/schemas/v2/__init__.py

@@ -14,7 +14,7 @@ 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 HistoryUnitStatusWithUnset  # noqa F401
 from .history import ImageLogsRequest  # noqa F401
 from .history import SingleImageWithStatus  # noqa F401
 from .job import JobCreateV2  # noqa F401

{fractal_server-2.14.6 → fractal_server-2.14.8}/fractal_server/app/schemas/v2/history.py

@@ -24,12 +24,19 @@ class HistoryUnitStatus(StrEnum):
     FAILED = "failed"
 
 
-class HistoryUnitStatusQuery(StrEnum):
+class HistoryUnitStatusWithUnset(StrEnum):
+    """
+    Available status for history queries
+
+    Attributes:
+        SUBMITTED:
+        DONE:
+        FAILED:
+    """
 
     SUBMITTED = "submitted"
     DONE = "done"
     FAILED = "failed"
-
     UNSET = "unset"
 
 
@@ -63,6 +70,9 @@ class HistoryRunReadAggregated(BaseModel):
     num_submitted_units: int
     num_done_units: int
     num_failed_units: int
+    args_schema_parallel: dict[str, Any] | None = None
+    args_schema_non_parallel: dict[str, Any] | None = None
+    version: str | None = None
 
     @field_serializer("timestamp_started")
     def serialize_datetime(v: datetime) -> str:
@@ -75,5 +85,6 @@ class ImageLogsRequest(BaseModel):
     zarr_url: str
 
 
+# FIXME: remove this when we update the webclient
 class SingleImageWithStatus(SingleImage):
     status: HistoryUnitStatus | None = None

fractal_server-2.14.8/fractal_server/data_migrations/2_14_6.py

@@ -0,0 +1,48 @@
+import logging
+
+from sqlmodel import select
+
+from fractal_server.app.db import get_sync_db
+from fractal_server.app.models import HistoryRun
+from fractal_server.app.models import TaskV2
+from fractal_server.app.models import WorkflowTaskV2
+
+logger = logging.getLogger("fix_db")
+logger.setLevel(logging.INFO)
+
+
+def fix_db():
+    logger.info("START execution of fix_db function")
+
+    with next(get_sync_db()) as db:
+
+        stm = select(HistoryRun).order_by(HistoryRun.id)
+        history_runs = db.execute(stm).scalars().all()
+
+        for hr in history_runs:
+            logger.info(f"HistoryRun[{hr.id}] START")
+
+            wft = db.get(WorkflowTaskV2, hr.workflowtask_id)
+            if wft is None:
+                logger.warning(
+                    f"WorkflowTaskV2[{hr.workflowtask_id}] not found. "
+                    "Trying to use HistoryRun.workflowtask_dump"
+                )
+                task_id = hr.workflowtask_dump.get("task_id")
+                if task_id is not None and db.get(TaskV2, task_id) is not None:
+                    hr.task_id = task_id
+                else:
+                    logger.warning(f"TaskV2[{task_id}] not found")
+            else:
+                hr.task_id = wft.task_id
+                logger.info(
+                    f"HistoryRun[{hr.id}].task_id set to {wft.task_id}"
+                )
+
+            db.add(hr)
+            logger.info(f"HistoryRun[{hr.id}] END")
+
+        db.commit()
+        logger.info("Changes committed.")
+
+    logger.info("END execution of fix_db function")