Parsl HTEX hangs due to incorrect `run_dir` references when running multiple executors in sequence within same process

[d33bs]

Describe the bug

Hello - thanks again for the great work on Parsl! I noticed recently when attempting to use multiple Parsl executors within the same Python process that the 2nd executor configured after clearing the first appears to reference the 1st executor's run_dir information. This causes an exception preventing multiple executors from being leveraged. This appears to occur when using customized run_dir settings and HTEX. The result is that Parsl hangs without exception (an error message appears in the run_dir logs about a missing file).

To Reproduce

See this notebook (or gist backup) as a reference.

import shutil
from parsl.app.app import python_app
import parsl
from parsl.config import Config
from parsl.executors import HighThroughputExecutor


@python_app
def add(x, y):
    return x + y

# HTEX run 1

config = Config(executors=[HighThroughputExecutor()])
config.run_dir = "./runinfo_1"

parsl.load(config)

futs = [add(i, i) for i in range(5)]
results = [f.result() for f in futs]
print(results)

parsl.dfk().cleanup()
parsl.clear()

shutil.rmtree("./runinfo_1")

# HTEX run 2

config = Config(executors=[HighThroughputExecutor()])
config.run_dir = "./runinfo_2"

parsl.load(config)

futs = [add(i, i) for i in range(5)]
results = [f.result() for f in futs]
print(results)

parsl.dfk().cleanup()
parsl.clear()

shutil.rmtree("./runinfo_2")

Expected behavior

I'd expect that after performing a dkf().cleanup() and parsl.clear() that the executors and their workers would be fully cleaned up and not reference previous configurations.

Environment

• OS: MacOS, Linux
• Python version: >= 3.9
• Parsl version: 2025.6.2

Distributed Environment

• Where are you running the Parsl script from ? Laptop, Google Colab
• Where do you need the workers to run ? Same as Parsl script