[HELP] I have persistent issue with save task.

[lamb-dao]

Many thanks in advance for your help

Here is minimal code, with narative comments on the cause. In brief, when I import vectors and save or rasterize and save, there is a pyogrio-sqlite OpenFailedError. I have been unable to find the cause or even a way to change the error to get some clues on how to proceed. I am somewhat desperate to solve this.

the rest of this text can be dropped into a python buffer

# &&& simplified version for testing error

def bind_observations(polygons=DATA_ids, observations=DATA_table, ddir=DATA_DIR):
    "Append table of observation data to polygons, ensure common column, then row bind on samples. Returns: fiona readable object"
    gdf = gpd.read_file(polygons)
    df = pd.read_csv(observations)
    # ensure correct attributes
    target_shared = 'sample'
    change_from = 'SAMPLE'
    cols_gdf = gdf.columns
    # must have target_shared
    if not target_shared in cols_gdf:
        raise ValueError(f"Expected column not found in {cols_gdf} Expected: {target_shared}")
    cols_df = df.columns
    # must have change_from
    if not change_from in cols_df:
        raise ValueError(f"Expected column not found in {cols_df} Expected: {change_from}")
    # make rename for merge
    df.rename(columns={change_from: target_shared}, inplace=True)
    # merge
    merged_gdf = gdf.merge(df, on=target_shared, how='left')
    # merged_gdf.columns
    # write to disk
    path = os.path.join(ddir, 'bound_observations.gpkg')
    abs_path = os.path.abspath(path)
    merged_gdf.to_file(abs_path, driver='GPKG', layer='name')
    return abs_path

bind_observations()
# '/bulk-2/model_output/data/bound_observations.gpkg'
test = gpd.read_file('/bulk-2/model_output/data/bound_observations.gpkg')
test.crs
test.columns

def bind_observations_temp(polygons=DATA_ids, observations=DATA_table, ddir=DATA_DIR):
    "Minimal case, only geometries and single int attribute"
    gdf = gpd.read_file(polygons)
    # write to disk
    path = os.path.join(ddir, 'temp_observations.gpkg')
    abs_path = os.path.abspath(path)
    gdf.to_file(abs_path, driver='GPKG', layer='identities')
    return abs_path

bind_observations_temp()
# '/bulk-2/model_output/data/temp_observations.gpkg'
test = gpd.read_file('/bulk-2/model_output/data/temp_observations.gpkg')
test.crs
test.columns

vector_feature = (FeatureType.VECTOR_TIMELESS, "IDENTITIES")

# init tasks
load_task = LoadTask(EOPATCH_TRAIN_DIR)
vector_task = VectorImportTask(vector_feature, bind_observations_temp())
rasterize_height_task = VectorToRasterTask(
    vector_feature, # as used in vector import task
    (FeatureType.DATA_TIMELESS, "HEIGHT"), #name of rasterized new layer, DATA for float values
    values_column="HEIGHT-CM", # select col from merged_gdf.columns to rasterize
    raster_shape=(FeatureType.MASK_TIMELESS, "IN_POLYGON"),
    raster_dtype=np.float32, # float
)
# this target dir is made empty before each test
save_task = SaveTask('/home/user/temp', overwrite_permission=OverwritePermission.OVERWRITE_FEATURES)

# node list
# success
# workflow_nodes = linearly_connect_tasks(load_task, <other data loading tasks have been successfully saved>, save_task)
# success
# workflow_nodes = linearly_connect_tasks(load_task, vector_task)
# fail
# workflow_nodes = linearly_connect_tasks(load_task, vector_task, save_task)
'''
test 2
manually place a copy of observations, as if vector_task had succeded to test rasterize height task

 cd /bulk-2/model_output/eopatch_0/vector_timeless/
 cp /bulk-2/model_output/data/IDENTITIES.gpkg .
 cd ../../eopatch_1/vector_timeless/
 cp /bulk-2/model_output/data/IDENTITIES.gpkg .
 cd ../../eopatch_2/vector_timeless/
 cp /bulk-2/model_output/data/IDENTITIES.gpkg .
 cd ../../eopatch_3/vector_timeless/
 cp /bulk-2/model_output/data/IDENTITIES.gpkg .

'''
# success
# workflow_nodes = linearly_connect_tasks(load_task, rasterize_height_task)
# fail
workflow_nodes = linearly_connect_tasks(load_task, rasterize_height_task, save_task)

# workflow
workflow = EOWorkflow(workflow_nodes)

# additional arguments
execution_args = []
for idx, bbox in enumerate(area_grid(DATA_train)): # area_grid partitions the polygon and returns bbox list
    execution_args.append(
        {
            workflow_nodes[0]: {"eopatch_folder": f"eopatch_{idx}"}, # load task is first
            workflow_nodes[-1]: {"eopatch_folder": f"eopatch_{idx}"} # save task is last
        }
    )

# Execute the workflow
executor = EOExecutor(workflow, execution_args, save_logs=True)
executor.run(workers=1)
executor.make_report()

failed_ids = executor.get_failed_executions()
if failed_ids:
    raise RuntimeError(
        f"Execution failed EOPatches with IDs:\n{failed_ids}\n"
        f"For more info check report at {executor.get_report_path()}"
    )

'''

In both cases, the combination of a (vector import task or rasterize task) followed by the save task causes this pyogiro-sqlite error


>> error
2025-03-29 16:09:02,118 eolearn.core.eoworkflow ERROR    Task 'SaveTask' with id SaveTask-06f7db1f0ce211f09159-6bd05c84396c failed with stack trace:
Traceback (most recent call last):
  File "pyogrio/_io.pyx", line 1603, in pyogrio._io.ogr_create
  File "pyogrio/_err.pyx", line 179, in pyogrio._err.exc_wrap_pointer
pyogrio._err.CPLE_OpenFailedError: sqlite3_open(<gzip _io.BufferedWriter name=b'/home/user/temp/tmpcowhph8o__tempfs__/tmp_feature' 0x7213b77aaa10>) failed: unable to open database file
pyogrio OpenFailedError: sqlite3_open failed: unable to open database file

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eoworkflow.py", line 271, in _execute_task
    return task.execute(*task_args, **task_kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/core_tasks.py", line 162, in execute
    eopatch.save(
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata.py", line 644, in save
    save_eopatch(
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata_io.py", line 162, in save_eopatch
    new_files = list(executor.map(lambda saver: saver(), data_for_saving))
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/concurrent/futures/_base.py", line 619, in result_iterator
    yield _result_or_cancel(fs.pop())
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/concurrent/futures/_base.py", line 317, in _result_or_cancel
    return fut.result(timeout)
           ^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/concurrent/futures/_base.py", line 449, in result
    return self.__get_result()
           ^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/concurrent/futures/_base.py", line 401, in __get_result
    raise self._exception
  File "/home/user/.conda/envs/model/lib/python3.11/concurrent/futures/thread.py", line 58, in run
    result = self.fn(*self.args, **self.kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata_io.py", line 162, in <lambda>
    new_files = list(executor.map(lambda saver: saver(), data_for_saving))
                                                ^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata_io.py", line 607, in save
    cls._save(data, tempfs, "tmp_feature", compress_level)
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata_io.py", line 623, in _save
    cls._write_to_file(data, gzip_file, path)
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata_io.py", line 676, in _write_to_file
    return data.to_file(file, driver="GPKG", encoding="utf-8", layer=layer, index=False)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/geopandas/geodataframe.py", line 1536, in to_file
    _to_file(self, filename, driver, schema, index, **kwargs)
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/geopandas/io/file.py", line 686, in _to_file
    _to_file_pyogrio(df, filename, driver, schema, crs, mode, metadata, **kwargs)
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/geopandas/io/file.py", line 748, in _to_file_pyogrio
    pyogrio.write_dataframe(df, filename, driver=driver, metadata=metadata, **kwargs)
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/pyogrio/geopandas.py", line 548, in write_dataframe
    write(
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/pyogrio/raw.py", line 530, in write
    ogr_write(
  File "pyogrio/_io.pyx", line 1799, in pyogrio._io.ogr_write
  File "pyogrio/_io.pyx", line 1612, in pyogrio._io.ogr_create
pyogrio.errors.DataSourceError: sqlite3_open(<gzip _io.BufferedWriter name=b'/home/user/temp/tmpcowhph8o__tempfs__/tmp_feature' 0x7213b77aaa10>) failed: unable to open database file
<< end error

Just doing the task without save seems to succeed, the log file shows the expected columns of IDENTITIES after import, or the created HEIGHT raster

test 1
  vector_timeless={
    IDENTITIES: geopandas.GeoDataFrame(columns=['sample', 'NAME', 'HEIGHT-CM', 'ROW-TYPE', 'HULLESS-CONDITION', 'SBLOTCH-RATING', 'WEIGHT', 'DIAMETER', 'AREA', 'STEM-WEIGHT', 'DENSITY', 'ROWS', 'BARLEY-WHEAT', 'HULLED', 'SBLOTCH-LMH', 'geometry'], length=1212, crs=EPSG:32614)
  }

test 2
  data_timeless={
    HEIGHT: numpy.ndarray(shape=(500, 500, 1), dtype=float32)
  }

But on save of either the error is thrown.
checking the previously empty output dir I see that the eopatch_<n> has been created which includes a dir for the feature I am trying to create. this dir will be empty. Although it seems that permissions are not an issue

The env has been installed with conda, and to my reading satisfies dependencies of pyogrio
(model) conda list | grep pyogrio
pyogrio                   0.7.2           py311hbac4ec9_0    conda-forge
(model) conda list | grep gdal
gdal                      3.7.2           py311h815a124_1    conda-forge
libgdal                   3.7.2                h323ed7e_1    conda-forge
(model) conda list | grep sqlite
libsqlite                 3.49.1               hee588c1_2    conda-forge
sqlite                    3.49.1               h9eae976_2    conda-forge
(model) conda list | grep geopandas
geopandas                 1.0.1              pyhd8ed1ab_3    conda-forge
geopandas-base            1.0.1              pyha770c72_3    conda-forge
(model) conda list | grep shapely
shapely                   2.0.2           py311he06c224_0    conda-forge

I have searched your repo for calls to pyogrio and found nothing.
I have made a series of changes, in my system, moving to other disks, ensuring the program is being opened from the same disk etc.
Searches for this error string show very few relevant hits.
Is this something you have seen before? I am at my wits end. Your deeper experience in spatial data may suggest ways ahead to solve this.
'''

[lamb-dao]

Adding some details to this.
It is known that the geometries are valid, I am running a function that mimics your validation steps before saving.
Also the permissions for the dirs I have this test reading and writing to are set to 777 to preclude any permission issues.

There is a message printed in the repl (in addition to the run log above)

Warning 1: File /vsimem/f7c0b1474f57418ea9fab7a0f2f24d70 has GPKG application_id, but non conformant file extension
Warning 1: The filename extension should be 'gpkg' instead of '' to conform to the GPKG specification.
ERROR 4: sqlite3_open(<gzip _io.BufferedWriter name=b'/home/user/temp/tmpedtuiz70__tempfs__/tmp_feature' 0x7edc70300220>) failed: unable to open database file

[lamb-dao]

looks related, and quite recent
geopandas/pyogrio#508

[zigaLuksic]

Hi @lamb-dao
it seems like pyogrio does not agree with additionally zipped files. This is the reason why we added a constraint for geopandas < 1 (because older versions do not use pyogrio).

You can try downgrading geopandas. If you are proficient enough with python, you can try changing eolearn/core/eodata_io.py line 239 to compress_level=0 to see if this fixes your issue.

@mlubej we decided not to support gpd >= 1 because it would break backwards compatibility, but i think we'll need to go with the times and make a breaking release.

[lamb-dao]

Well thank you @zigaLuksic

The eo-learn project is certainly beyond my proficiency, but I was pretty close. in eodata_io.py at _save I had added logging on both the compressed and non-compressed calls, and was considering forcing it there.

With your suggestion at line 239 of _yield_savers modified to compress_level=0
Logging in _save shows in the repl that all calls go to the non-compressed call, and here in the run log we see the same error return from that.

I will downgrade gpd for now and report back.

with gratitude,
lamb

2025-03-31 13:14:46,704 eolearn.core.eoworkflow ERROR    Task 'SaveTask' with id SaveTask-fb285faf0e5b11f0aea5-79f427f9125a failed with stack trace:
Traceback (most recent call last):
  File "pyogrio/_io.pyx", line 1603, in pyogrio._io.ogr_create
  File "pyogrio/_err.pyx", line 179, in pyogrio._err.exc_wrap_pointer
pyogrio._err.CPLE_OpenFailedError: sqlite3_open(<_io.BufferedWriter name=b'/home/user/temp/tmpobt6_wxs__tempfs__/tmp_feature'>) failed: unable to open database file

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eoworkflow.py", line 271, in _execute_task
    return task.execute(*task_args, **task_kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/core_tasks.py", line 162, in execute
    eopatch.save(
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata.py", line 644, in save
    save_eopatch(
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata_io.py", line 162, in save_eopatch
    new_files = list(executor.map(lambda saver: saver(), data_for_saving))
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/concurrent/futures/_base.py", line 619, in result_iterator
    yield _result_or_cancel(fs.pop())
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/concurrent/futures/_base.py", line 317, in _result_or_cancel
    return fut.result(timeout)
           ^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/concurrent/futures/_base.py", line 449, in result
    return self.__get_result()
           ^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/concurrent/futures/_base.py", line 401, in __get_result
    raise self._exception
  File "/home/user/.conda/envs/model/lib/python3.11/concurrent/futures/thread.py", line 58, in run
    result = self.fn(*self.args, **self.kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata_io.py", line 162, in <lambda>
    new_files = list(executor.map(lambda saver: saver(), data_for_saving))
                                                ^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata_io.py", line 608, in save
    cls._save(data, tempfs, "tmp_feature", compress_level)
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata_io.py", line 623, in _save
    cls._write_to_file(data, file, path) #non compressed call
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/eolearn/core/eodata_io.py", line 688, in _write_to_file
    return data.to_file(file, driver="GPKG", encoding="utf-8", layer=layer, index=False)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/geopandas/geodataframe.py", line 1536, in to_file
    _to_file(self, filename, driver, schema, index, **kwargs)
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/geopandas/io/file.py", line 686, in _to_file
    _to_file_pyogrio(df, filename, driver, schema, crs, mode, metadata, **kwargs)
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/geopandas/io/file.py", line 748, in _to_file_pyogrio
    pyogrio.write_dataframe(df, filename, driver=driver, metadata=metadata, **kwargs)
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/pyogrio/geopandas.py", line 548, in write_dataframe
    write(
  File "/home/user/.conda/envs/model/lib/python3.11/site-packages/pyogrio/raw.py", line 530, in write
    ogr_write(
  File "pyogrio/_io.pyx", line 1799, in pyogrio._io.ogr_write
  File "pyogrio/_io.pyx", line 1612, in pyogrio._io.ogr_create
pyogrio.errors.DataSourceError: sqlite3_open(<_io.BufferedWriter name=b'/home/user/temp/tmpobt6_wxs__tempfs__/tmp_feature'>) failed: unable to open database file

[lamb-dao]

recreating the conda env with

mamba install -c conda-forge -c timeseriesai python shapely aenum scikit-learn lightgbm "geopandas<1" sentinelhub eo-learn eo-learn-visualization fastai numba timeseriesai::tsai

the vector__task completes!

[lamb-dao]

Thank you for your direction on this, and of course thank you for eo-learn.
I look forward to making contributions once this project wraps up.