Failed import of old calculation results from remote computer

[ValkScripter]

Greetings developers,

I have dived into Aiida just a few days ago.
I have learned and tested successfully the submission of CalcJobs on remote computers to run QuantumEspresso calculations.
So I wanted to make Aiida 'read' previous results that I had been running before. Here comes my problem.

If I understood correctly I should pass to the CalcJob (in my case through the quantumespresso.pw builder) the input named remote_folder which is of the type RemoteData so that the job will not actually run the job but instead rewrite the inputs - that I should give again to the job - and retrieve the results from this remote_folder.
So I created this remote_folder with the following command line:

remote = RemoteData(remote_path='/absolute/path/to/folder/on/remote/computer', computer=load_computer("my_remote_computer"))
builder.remote_folder = remote

with the builder created from a code, itself belonging to a remote computer (with core.ssh transport and core.pbspro scheduler):

builder = load_code("pw@my_remote_computer").get_builder()

I checked that the remote_folder was working correctly with functional SSH connection in order for the CalcJob to read from it, by doing a :

remote_folder.list_dir()

Results

['file1.txt',
'prefix.pwi',
'prefix.pwo',
...
]

After giving to the builder all the inputs and metadata that it needed (including input_filename and output_filename),

builder.metadata.options.input_filename = "prefix.pwi"
builder.metadata.options.output_filename = "prefix.pwo"

I launched the job (with either a run or a submit) and ultimately got an error with the following traceback:

+-> REPORT at ...
...
 |   File ".../aiida/lib/python3.12/site-packages/aiida/engine/processes/calcjobs/calcjob.py", line 636, in _perform_import
 |     retrieve_calculation(self.node, transport, retrieved_temporary_folder.abspath)
 |   File ".../aiida/lib/python3.12/site-packages/aiida/engine/daemon/execmanager.py", line 480, in retrieve_calculation
 |     transport.chdir(workdir)
 |   File ".../aiida/lib/python3.12/site-packages/aiida/transports/plugins/local.py", line 111, in chdir
 |     raise IOError(f"'{new_path}' is not a valid directory")
 | OSError: '/absolute/path/to/folder/on/remote/computer' is not a valid directory

I investigated and found that the _perform_import function in aiida/engine/processes/calcjobs/calcjob.py passes always - with the with statement - a core.local transport to the retrieve_calculation function in file aiida/engine/daemon/execmanager.py, instead of a core.ssh transport, which I assume would be needed in my case.
I believe this is why the command transport.chdir(workdir) fails and that the given remote directory path is said to be not valid.
Below is the corresponding code part:

aiida-core/src/aiida/engine/processes/calcjobs/calcjob.py

Lines 631 to 639 in cf091e8

	with LocalTransport() as transport:
	with SandboxFolder(filepath_sandbox) as folder:
	with SandboxFolder(filepath_sandbox) as retrieved_temporary_folder:
	self.presubmit(folder)
	self.node.set_remote_workdir(self.inputs.remote_folder.get_remote_path())
	retrieved = retrieve_calculation(self.node, transport, retrieved_temporary_folder.abspath)
	if retrieved is not None:
	self.out(self.node.link_label_retrieved, retrieved)
	self.update_outputs()

1/ Am I right or am I missing something?
2/ In the former case, what should I do to circumvent this problem?
3/ In the latter case, what would be the missing piece?

Thank you for your time and attention.

[unkcpz]

Hi @ValkScripter, thanks for the question. The communication platform is moved to https://aiida.discourse.group/ see https://github.com/orgs/aiidateam/discussions/6113.
Could you move the question there? Thanks!

[ValkScripter]

Hi @unkcpz, thank you for redirecting me to the communication platform.
There is a question there giving me something that looks like an answer to my question already:
https://aiida.discourse.group/t/immigration-of-existing-vasp-calculations/326/2
Thanks again!