From b6a556e0079fde206aee0751119201fc70f6777c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Jan=C3=9Fen?= <janssen@mpie.de>
Date: Wed, 16 Jul 2025 19:41:50 +0200
Subject: [PATCH 1/8] Define error log file in resource_dict

---
 .github/workflows/pipeline.yml                |  2 +-
 executorlib/backend/cache_parallel.py         |  5 +++++
 executorlib/backend/interactive_parallel.py   |  5 +++++
 executorlib/backend/interactive_serial.py     |  5 +++++
 executorlib/standalone/error.py               | 22 +++++++++++++++++++
 executorlib/task_scheduler/file/backend.py    |  5 +++++
 executorlib/task_scheduler/file/hdf.py        |  4 ++++
 executorlib/task_scheduler/file/shared.py     |  2 ++
 .../task_scheduler/interactive/shared.py      |  3 +++
 pyproject.toml                                |  1 -
 tests/test_cache_fileexecutor_serial.py       | 15 +++++++++++++
 tests/test_singlenodeexecutor_cache.py        | 12 ++++++++++
 12 files changed, 79 insertions(+), 2 deletions(-)
 create mode 100644 executorlib/standalone/error.py

diff --git a/.github/workflows/pipeline.yml b/.github/workflows/pipeline.yml
index 5b1e02a7..c51853d0 100644
--- a/.github/workflows/pipeline.yml
+++ b/.github/workflows/pipeline.yml
@@ -402,7 +402,7 @@ jobs:
       run: echo -e "channels:\n  - conda-forge\n" > .condarc
     - uses: conda-incubator/setup-miniconda@v3
       with:
-        python-version: '3.9'
+        python-version: '3.10'
         miniforge-version: latest
         condarc-file: .condarc
         environment-file: .ci_support/environment-old.yml
diff --git a/executorlib/backend/cache_parallel.py b/executorlib/backend/cache_parallel.py
index 9b1d25d8..49ad6d6e 100644
--- a/executorlib/backend/cache_parallel.py
+++ b/executorlib/backend/cache_parallel.py
@@ -4,6 +4,7 @@
 
 import cloudpickle
 
+from executorlib.standalone.error import backend_write_error_file
 from executorlib.task_scheduler.file.backend import (
     backend_load_file,
     backend_write_file,
@@ -53,6 +54,10 @@ def main() -> None:
                 output={"error": error},
                 runtime=time.time() - time_start,
             )
+            backend_write_error_file(
+                error=error,
+                apply_dict=apply_dict,
+            )
     else:
         if mpi_rank_zero:
             backend_write_file(
diff --git a/executorlib/backend/interactive_parallel.py b/executorlib/backend/interactive_parallel.py
index 5ae2c320..3d5aadcc 100644
--- a/executorlib/backend/interactive_parallel.py
+++ b/executorlib/backend/interactive_parallel.py
@@ -6,6 +6,7 @@
 import cloudpickle
 import zmq
 
+from executorlib.standalone.error import backend_write_error_file
 from executorlib.standalone.interactive.backend import call_funct, parse_arguments
 from executorlib.standalone.interactive.communication import (
     interface_connect,
@@ -82,6 +83,10 @@ def main() -> None:
                         socket=socket,
                         result_dict={"error": error},
                     )
+                    backend_write_error_file(
+                        error=error,
+                        apply_dict=input_dict,
+                    )
             else:
                 # Send output
                 if mpi_rank_zero:
diff --git a/executorlib/backend/interactive_serial.py b/executorlib/backend/interactive_serial.py
index 34fb2288..c72d95a0 100644
--- a/executorlib/backend/interactive_serial.py
+++ b/executorlib/backend/interactive_serial.py
@@ -2,6 +2,7 @@
 from os.path import abspath
 from typing import Optional
 
+from executorlib.standalone.error import backend_write_error_file
 from executorlib.standalone.interactive.backend import call_funct, parse_arguments
 from executorlib.standalone.interactive.communication import (
     interface_connect,
@@ -58,6 +59,10 @@ def main(argument_lst: Optional[list[str]] = None):
                     socket=socket,
                     result_dict={"error": error},
                 )
+                backend_write_error_file(
+                    error=error,
+                    apply_dict=input_dict,
+                )
             else:
                 # Send output
                 interface_send(socket=socket, result_dict={"result": output})
diff --git a/executorlib/standalone/error.py b/executorlib/standalone/error.py
new file mode 100644
index 00000000..387e6374
--- /dev/null
+++ b/executorlib/standalone/error.py
@@ -0,0 +1,22 @@
+
+import traceback
+
+
+def backend_write_error_file(error: Exception, apply_dict: dict) -> None:
+    """
+    Write an error to a file if specified in the apply_dict.
+
+    Args:
+        error (Exception): The error to be written.
+        apply_dict (dict): Dictionary containing additional parameters.
+
+    Returns:
+        None
+    """
+    error_log_file = apply_dict.get("error_log_file", None)
+    if error_log_file is not None:
+        with open(error_log_file, "a") as f:
+            f.write("function: " + str(apply_dict["fn"]) + "\n")
+            f.write("args: " + str(apply_dict["args"]) + "\n")
+            f.write("kwargs: " + str(apply_dict["kwargs"]) + "\n")
+            traceback.print_exception(error, file=f)
\ No newline at end of file
diff --git a/executorlib/task_scheduler/file/backend.py b/executorlib/task_scheduler/file/backend.py
index 4ea27c17..cbe869cc 100644
--- a/executorlib/task_scheduler/file/backend.py
+++ b/executorlib/task_scheduler/file/backend.py
@@ -2,6 +2,7 @@
 import time
 from typing import Any
 
+from executorlib.standalone.error import backend_write_error_file
 from executorlib.task_scheduler.file.hdf import dump, load
 from executorlib.task_scheduler.file.shared import FutureItem
 
@@ -77,6 +78,10 @@ def backend_execute_task_in_file(file_name: str) -> None:
         }
     except Exception as error:
         result = {"error": error}
+        backend_write_error_file(
+            error=error,
+            apply_dict=apply_dict,
+        )
 
     backend_write_file(
         file_name=file_name,
diff --git a/executorlib/task_scheduler/file/hdf.py b/executorlib/task_scheduler/file/hdf.py
index d8cde507..812198c9 100644
--- a/executorlib/task_scheduler/file/hdf.py
+++ b/executorlib/task_scheduler/file/hdf.py
@@ -52,6 +52,10 @@ def load(file_name: str) -> dict:
             data_dict["kwargs"] = cloudpickle.loads(np.void(hdf["/input_kwargs"]))
         else:
             data_dict["kwargs"] = {}
+        if "error_log_file" in hdf:
+            data_dict["error_log_file"] = cloudpickle.loads(
+                np.void(hdf["/error_log_file"])
+            )
         return data_dict
 
 
diff --git a/executorlib/task_scheduler/file/shared.py b/executorlib/task_scheduler/file/shared.py
index 8aaa76bc..1944321f 100644
--- a/executorlib/task_scheduler/file/shared.py
+++ b/executorlib/task_scheduler/file/shared.py
@@ -126,6 +126,7 @@ def execute_tasks_h5(
             )
             cache_key = task_resource_dict.pop("cache_key", None)
             cache_directory = os.path.abspath(task_resource_dict.pop("cache_directory"))
+            error_log_file = task_resource_dict.pop("error_log_file", None)
             task_key, data_dict = serialize_funct_h5(
                 fn=task_dict["fn"],
                 fn_args=task_args,
@@ -133,6 +134,7 @@ def execute_tasks_h5(
                 resource_dict=task_resource_dict,
                 cache_key=cache_key,
             )
+            data_dict["error_log_file"] = error_log_file
             if task_key not in memory_dict:
                 if os.path.join(
                     cache_directory, task_key + "_o.h5"
diff --git a/executorlib/task_scheduler/interactive/shared.py b/executorlib/task_scheduler/interactive/shared.py
index 7c27e043..cc22ca2c 100644
--- a/executorlib/task_scheduler/interactive/shared.py
+++ b/executorlib/task_scheduler/interactive/shared.py
@@ -26,6 +26,7 @@ def execute_tasks(
     cache_key: Optional[str] = None,
     queue_join_on_shutdown: bool = True,
     log_obj_size: bool = False,
+    error_log_file: Optional[str] = None,
     **kwargs,
 ) -> None:
     """
@@ -70,6 +71,8 @@ def execute_tasks(
                 future_queue.join()
             break
         elif "fn" in task_dict and "future" in task_dict:
+            if error_log_file is not None:
+                task_dict["error_log_file"] = error_log_file
             if cache_directory is None:
                 _execute_task_without_cache(
                     interface=interface, task_dict=task_dict, future_queue=future_queue
diff --git a/pyproject.toml b/pyproject.toml
index d51a8076..ee8dd253 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -23,7 +23,6 @@ classifiers = [
     "License :: OSI Approved :: BSD License",
     "Intended Audience :: Science/Research",
     "Operating System :: OS Independent",
-    "Programming Language :: Python :: 3.9",
     "Programming Language :: Python :: 3.10",
     "Programming Language :: Python :: 3.11",
     "Programming Language :: Python :: 3.12",
diff --git a/tests/test_cache_fileexecutor_serial.py b/tests/test_cache_fileexecutor_serial.py
index a9eb97a5..9f8211a1 100644
--- a/tests/test_cache_fileexecutor_serial.py
+++ b/tests/test_cache_fileexecutor_serial.py
@@ -88,6 +88,21 @@ def test_executor_error(self):
             fs1 = exe.submit(get_error, a=1)
             with self.assertRaises(ValueError):
                 fs1.result()
+        self.assertEqual(len(os.listdir(cwd)), 1)
+
+    def test_executor_error(self):
+        cwd = os.path.join(os.path.dirname(__file__), "executables")
+        with FileTaskScheduler(
+            resource_dict={"cwd": cwd, "error_log_file": "error.out"}, 
+            execute_function=execute_in_subprocess
+        ) as exe:
+            fs1 = exe.submit(get_error, a=1)
+            with self.assertRaises(ValueError):
+                fs1.result()
+        working_directory_file_lst = os.listdir(cwd)
+        self.assertEqual(len(working_directory_file_lst), 2)
+        self.assertTrue("error.out" in working_directory_file_lst)
+        os.remove(os.path.join(cwd, "error.out"))
 
     def test_executor_function(self):
         fs1 = Future()
diff --git a/tests/test_singlenodeexecutor_cache.py b/tests/test_singlenodeexecutor_cache.py
index 531d369d..650c0044 100644
--- a/tests/test_singlenodeexecutor_cache.py
+++ b/tests/test_singlenodeexecutor_cache.py
@@ -58,6 +58,18 @@ def test_cache_error(self):
             with self.assertRaises(ValueError):
                 print(f.result())
 
+    def test_cache_error_file(self):
+        cache_directory = os.path.abspath("cache_error")
+        error_out =  "error.out"
+        with SingleNodeExecutor(cache_directory=cache_directory) as exe:
+            self.assertTrue(exe)
+            cloudpickle_register(ind=1)
+            f = exe.submit(get_error, a=1, resource_dict={"error_log_file": error_out})
+            with self.assertRaises(ValueError):
+                print(f.result())
+        self.assertTrue(os.path.exists(error_out))
+        os.remove(error_out)
+
     def tearDown(self):
         shutil.rmtree("executorlib_cache", ignore_errors=True)
         shutil.rmtree("cache_error", ignore_errors=True)

From 925a62905e985a717d95b7898e9c777df505ae7f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Jan=C3=9Fen?= <janssen@mpie.de>
Date: Wed, 16 Jul 2025 19:49:57 +0200
Subject: [PATCH 2/8] fix tests

---
 executorlib/standalone/cache.py         | 1 +
 tests/test_cache_fileexecutor_serial.py | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/executorlib/standalone/cache.py b/executorlib/standalone/cache.py
index c3d89483..92e1ae68 100644
--- a/executorlib/standalone/cache.py
+++ b/executorlib/standalone/cache.py
@@ -10,6 +10,7 @@
     "error": "error",
     "runtime": "runtime",
     "queue_id": "queue_id",
+    "error_log_file": "error_log_file",
 }
 
 
diff --git a/tests/test_cache_fileexecutor_serial.py b/tests/test_cache_fileexecutor_serial.py
index 9f8211a1..fcbebc1a 100644
--- a/tests/test_cache_fileexecutor_serial.py
+++ b/tests/test_cache_fileexecutor_serial.py
@@ -90,7 +90,7 @@ def test_executor_error(self):
                 fs1.result()
         self.assertEqual(len(os.listdir(cwd)), 1)
 
-    def test_executor_error(self):
+    def test_executor_error_file(self):
         cwd = os.path.join(os.path.dirname(__file__), "executables")
         with FileTaskScheduler(
             resource_dict={"cwd": cwd, "error_log_file": "error.out"}, 

From 297e0cae93cb7f61ba31afd618d19da1da43e498 Mon Sep 17 00:00:00 2001
From: pyiron-runner <pyiron@mpie.de>
Date: Wed, 16 Jul 2025 17:50:59 +0000
Subject: [PATCH 3/8] Format black

---
 executorlib/standalone/error.py | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/executorlib/standalone/error.py b/executorlib/standalone/error.py
index 387e6374..5b0cbabd 100644
--- a/executorlib/standalone/error.py
+++ b/executorlib/standalone/error.py
@@ -1,4 +1,3 @@
-
 import traceback
 
 
@@ -19,4 +18,4 @@ def backend_write_error_file(error: Exception, apply_dict: dict) -> None:
             f.write("function: " + str(apply_dict["fn"]) + "\n")
             f.write("args: " + str(apply_dict["args"]) + "\n")
             f.write("kwargs: " + str(apply_dict["kwargs"]) + "\n")
-            traceback.print_exception(error, file=f)
\ No newline at end of file
+            traceback.print_exception(error, file=f)

From 6d141367760c827ba7f853db2ecd10b3fde492ef Mon Sep 17 00:00:00 2001
From: "pre-commit-ci[bot]"
 <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Wed, 16 Jul 2025 17:52:21 +0000
Subject: [PATCH 4/8] [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci
---
 executorlib/standalone/error.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/executorlib/standalone/error.py b/executorlib/standalone/error.py
index 5b0cbabd..410a9c22 100644
--- a/executorlib/standalone/error.py
+++ b/executorlib/standalone/error.py
@@ -12,7 +12,7 @@ def backend_write_error_file(error: Exception, apply_dict: dict) -> None:
     Returns:
         None
     """
-    error_log_file = apply_dict.get("error_log_file", None)
+    error_log_file = apply_dict.get("error_log_file")
     if error_log_file is not None:
         with open(error_log_file, "a") as f:
             f.write("function: " + str(apply_dict["fn"]) + "\n")

From ceca9ca83bda3ce2d4078d4b2e99f556084182c1 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Jan=C3=9Fen?= <janssen@mpie.de>
Date: Wed, 16 Jul 2025 19:57:33 +0200
Subject: [PATCH 5/8] Add docstrings

---
 executorlib/executor/flux.py   | 10 ++++++++++
 executorlib/executor/single.py | 10 ++++++++++
 executorlib/executor/slurm.py  | 10 ++++++++++
 3 files changed, 30 insertions(+)

diff --git a/executorlib/executor/flux.py b/executorlib/executor/flux.py
index 145ccc73..33b45306 100644
--- a/executorlib/executor/flux.py
+++ b/executorlib/executor/flux.py
@@ -41,6 +41,8 @@ class FluxJobExecutor(BaseExecutor):
                                                            Defaults to None.
                               - exclusive (bool): Whether to exclusively reserve the compute nodes, or allow sharing
                                                   compute notes. Defaults to False.
+                              - error_log_file (str): Name of the error log file to use for storing exceptions raised
+                                                      by the Python functions submitted to the Executor.
         flux_executor (flux.job.FluxExecutor): Flux Python interface to submit the workers to flux
         flux_executor_pmi_mode (str): PMI interface to use (OpenMPI v5 requires pmix) default is None (Flux only)
         flux_executor_nesting (bool): Provide hierarchically nested Flux job scheduler inside the submitted function.
@@ -126,6 +128,8 @@ def __init__(
                                                                Defaults to None.
                                   - exclusive (bool): Whether to exclusively reserve the compute nodes, or allow sharing
                                                       compute notes. Defaults to False.
+                                  - error_log_file (str): Name of the error log file to use for storing exceptions
+                                                          raised by the Python functions submitted to the Executor.
             flux_executor (flux.job.FluxExecutor): Flux Python interface to submit the workers to flux
             flux_executor_pmi_mode (str): PMI interface to use (OpenMPI v5 requires pmix) default is None (Flux only)
             flux_executor_nesting (bool): Provide hierarchically nested Flux job scheduler inside the submitted function.
@@ -229,6 +233,8 @@ class FluxClusterExecutor(BaseExecutor):
                               - openmpi_oversubscribe (bool): adds the `--oversubscribe` command line flag (OpenMPI and
                                                               SLURM only) - default False
                               - slurm_cmd_args (list): Additional command line arguments for the srun call (SLURM only)
+                              - error_log_file (str): Name of the error log file to use for storing exceptions raised
+                                                      by the Python functions submitted to the Executor.
         pysqa_config_directory (str, optional): path to the pysqa config directory (only for pysqa based backend).
         hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                       context of an HPC cluster this essential to be able to communicate to an
@@ -308,6 +314,8 @@ def __init__(
                                                                   and SLURM only) - default False
                                   - slurm_cmd_args (list): Additional command line arguments for the srun call (SLURM
                                                            only)
+                                  - error_log_file (str): Name of the error log file to use for storing exceptions
+                                                          raised by the Python functions submitted to the Executor.
             pysqa_config_directory (str, optional): path to the pysqa config directory (only for pysqa based backend).
             hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                       context of an HPC cluster this essential to be able to communicate to an
@@ -424,6 +432,8 @@ def create_flux_executor(
                                                            Defaults to None.
                               - exclusive (bool): Whether to exclusively reserve the compute nodes, or allow sharing
                                                   compute notes. Defaults to False.
+                              - error_log_file (str): Name of the error log file to use for storing exceptions raised
+                                                      by the Python functions submitted to the Executor.
         flux_executor (flux.job.FluxExecutor): Flux Python interface to submit the workers to flux
         flux_executor_pmi_mode (str): PMI interface to use (OpenMPI v5 requires pmix) default is None (Flux only)
         flux_executor_nesting (bool): Provide hierarchically nested Flux job scheduler inside the submitted function.
diff --git a/executorlib/executor/single.py b/executorlib/executor/single.py
index 30037505..9ad40c13 100644
--- a/executorlib/executor/single.py
+++ b/executorlib/executor/single.py
@@ -39,6 +39,8 @@ class SingleNodeExecutor(BaseExecutor):
                               - openmpi_oversubscribe (bool): adds the `--oversubscribe` command line flag (OpenMPI and
                                                               SLURM only) - default False
                               - slurm_cmd_args (list): Additional command line arguments for the srun call (SLURM only)
+                              - error_log_file (str): Name of the error log file to use for storing exceptions raised
+                                                      by the Python functions submitted to the Executor.
         hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                       context of an HPC cluster this essential to be able to communicate to an
                                       Executor running on a different compute node within the same allocation. And
@@ -116,6 +118,8 @@ def __init__(
                                                                   and SLURM only) - default False
                                   - slurm_cmd_args (list): Additional command line arguments for the srun call (SLURM
                                                            only)
+                                  - error_log_file (str): Name of the error log file to use for storing exceptions
+                                                          raised by the Python functions submitted to the Executor.
             hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                       context of an HPC cluster this essential to be able to communicate to an
                                       Executor running on a different compute node within the same allocation. And
@@ -202,6 +206,8 @@ class TestClusterExecutor(BaseExecutor):
                               - threads_per_core (int): number of OpenMP threads to be used for each function call
                               - gpus_per_core (int): number of GPUs per worker - defaults to 0
                               - cwd (str/None): current working directory where the parallel python task is executed
+                              - error_log_file (str): Name of the error log file to use for storing exceptions raised
+                                                      by the Python functions submitted to the Executor.
         hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                       context of an HPC cluster this essential to be able to communicate to an
                                       Executor running on a different compute node within the same allocation. And
@@ -273,6 +279,8 @@ def __init__(
                                   - threads_per_core (int): number of OpenMP threads to be used for each function call
                                   - gpus_per_core (int): number of GPUs per worker - defaults to 0
                                   - cwd (str/None): current working directory where the parallel python task is executed
+                                  - error_log_file (str): Name of the error log file to use for storing exceptions
+                                                          raised by the Python functions submitted to the Executor.
             hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                       context of an HPC cluster this essential to be able to communicate to an
                                       Executor running on a different compute node within the same allocation. And
@@ -381,6 +389,8 @@ def create_single_node_executor(
                                                               and SLURM only) - default False
                               - slurm_cmd_args (list): Additional command line arguments for the srun call (SLURM
                                                        only)
+                              - error_log_file (str): Name of the error log file to use for storing exceptions raised
+                                                      by the Python functions submitted to the Executor.
         hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                   context of an HPC cluster this essential to be able to communicate to an
                                   Executor running on a different compute node within the same allocation. And
diff --git a/executorlib/executor/slurm.py b/executorlib/executor/slurm.py
index bf3e48c7..c0353a97 100644
--- a/executorlib/executor/slurm.py
+++ b/executorlib/executor/slurm.py
@@ -41,6 +41,8 @@ class SlurmClusterExecutor(BaseExecutor):
                               - openmpi_oversubscribe (bool): adds the `--oversubscribe` command line flag (OpenMPI and
                                                               SLURM only) - default False
                               - slurm_cmd_args (list): Additional command line arguments for the srun call (SLURM only)
+                              - error_log_file (str): Name of the error log file to use for storing exceptions raised
+                                                      by the Python functions submitted to the Executor.
         pysqa_config_directory (str, optional): path to the pysqa config directory (only for pysqa based backend).
         hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                       context of an HPC cluster this essential to be able to communicate to an
@@ -120,6 +122,8 @@ def __init__(
                                                                   and SLURM only) - default False
                                   - slurm_cmd_args (list): Additional command line arguments for the srun call (SLURM
                                                            only)
+                                  - error_log_file (str): Name of the error log file to use for storing exceptions
+                                                          raised by the Python functions submitted to the Executor.
             pysqa_config_directory (str, optional): path to the pysqa config directory (only for pysqa based backend).
             hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                       context of an HPC cluster this essential to be able to communicate to an
@@ -226,6 +230,8 @@ class SlurmJobExecutor(BaseExecutor):
                                                            Defaults to None.
                               - exclusive (bool): Whether to exclusively reserve the compute nodes, or allow sharing
                                                   compute notes. Defaults to False.
+                              - error_log_file (str): Name of the error log file to use for storing exceptions raised
+                                                      by the Python functions submitted to the Executor.
         hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                       context of an HPC cluster this essential to be able to communicate to an
                                       Executor running on a different compute node within the same allocation. And
@@ -307,6 +313,8 @@ def __init__(
                                                            Defaults to None.
                                   - exclusive (bool): Whether to exclusively reserve the compute nodes, or allow sharing
                                                       compute notes. Defaults to False.
+                                  - error_log_file (str): Name of the error log file to use for storing exceptions
+                                                          raised by the Python functions submitted to the Executor.
             hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                       context of an HPC cluster this essential to be able to communicate to an
                                       Executor running on a different compute node within the same allocation. And
@@ -408,6 +416,8 @@ def create_slurm_executor(
                                                            Defaults to None.
                               - exclusive (bool): Whether to exclusively reserve the compute nodes, or allow sharing
                                                   compute notes. Defaults to False.
+                              - error_log_file (str): Name of the error log file to use for storing exceptions raised
+                                                      by the Python functions submitted to the Executor.
         hostname_localhost (boolean): use localhost instead of the hostname to establish the zmq connection. In the
                                   context of an HPC cluster this essential to be able to communicate to an
                                   Executor running on a different compute node within the same allocation. And

From 80d001bc6ea16c9cc8f143e3dfe0d84a835ff236 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Jan=C3=9Fen?= <janssen@mpie.de>
Date: Wed, 16 Jul 2025 20:00:58 +0200
Subject: [PATCH 6/8] Add HDF test

---
 tests/test_standalone_hdf.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/tests/test_standalone_hdf.py b/tests/test_standalone_hdf.py
index addcce55..ed25c1a9 100644
--- a/tests/test_standalone_hdf.py
+++ b/tests/test_standalone_hdf.py
@@ -75,6 +75,7 @@ def test_hdf_kwargs(self):
                 "args": (),
                 "kwargs": {"a": a, "b": b},
                 "queue_id": 123,
+                "error_log_file": "error.out",
             },
         )
         data_dict = load(file_name=file_name)

From 4a1be415ac3171def197a0394f414f4a6ef7acb2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Jan=C3=9Fen?= <janssen@mpie.de>
Date: Wed, 16 Jul 2025 20:12:42 +0200
Subject: [PATCH 7/8] test error function

---
 tests/test_standalone_error.py | 29 +++++++++++++++++++++++++++++
 1 file changed, 29 insertions(+)
 create mode 100644 tests/test_standalone_error.py

diff --git a/tests/test_standalone_error.py b/tests/test_standalone_error.py
new file mode 100644
index 00000000..a0074cc2
--- /dev/null
+++ b/tests/test_standalone_error.py
@@ -0,0 +1,29 @@
+import os
+import unittest
+from executorlib.standalone.error import backend_write_error_file
+
+
+class TestErrorWriter(unittest.TestCase):
+    def test_backend_write_error_file(self):
+        backend_write_error_file(
+            error=ValueError(),
+            apply_dict={
+                "error_log_file": "error.out",
+                "fn": 1,
+                "args": (1, 2, 3),
+                "kwargs": {"a": 1, "b": 2, "c": 3},
+            }
+        )
+        error_file_content = [
+            'function: 1\n',
+            'args: (1, 2, 3)\n',
+            "kwargs: {'a': 1, 'b': 2, 'c': 3}\n",
+            'ValueError\n'
+        ]
+        with open("error.out", "r") as f:
+            content = f.readlines()
+        self.assertEqual(error_file_content, content)
+
+    def tearDown(self):
+        if os.path.exists("error.out"):
+            os.remove("error.out")
\ No newline at end of file

From dc149191ef81e28ca4c92680d819f1786d9463f1 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Jan=C3=9Fen?= <janssen@mpie.de>
Date: Wed, 16 Jul 2025 20:20:45 +0200
Subject: [PATCH 8/8] new test

---
 tests/test_cache_backend_execute.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/tests/test_cache_backend_execute.py b/tests/test_cache_backend_execute.py
index bb634c33..bc0131bd 100644
--- a/tests/test_cache_backend_execute.py
+++ b/tests/test_cache_backend_execute.py
@@ -121,6 +121,7 @@ def test_execute_function_error(self):
         )
         file_name = os.path.join(cache_directory, task_key + "_i.h5")
         os.makedirs(cache_directory, exist_ok=True)
+        data_dict["error_log_file"] = os.path.join(cache_directory, "error.out")
         dump(file_name=file_name, data_dict=data_dict)
         backend_execute_task_in_file(file_name=file_name)
         future_obj = Future()
@@ -130,6 +131,11 @@ def test_execute_function_error(self):
         self.assertTrue(future_obj.done())
         with self.assertRaises(ValueError):
             future_obj.result()
+        with open(os.path.join(cache_directory, "error.out"), "r") as f:
+            content = f.readlines()
+        self.assertEqual(content[1], 'args: []\n')
+        self.assertEqual(content[2], "kwargs: {'a': 1}\n")
+        self.assertEqual(content[-1], 'ValueError: 1\n')
         self.assertTrue(
             get_runtime(file_name=os.path.join(cache_directory, task_key + "_o.h5"))
             > 0.0