Finding exact command that was executed

Walser52 · September 22, 2025, 6:55am

I have a simple code (given below) for the Espresso plugin based on the tutorial. It’s not really working. It’s possible that the command I use on the terminal is slightly different. I tried setting up code with different settings from prepend-text etc.

The command I usually run is on the lines of:

mpirun -np 2 --mca btl self,vader pw.x < pwscf.in > pwscf.out

So:

Is there a way I can find the exact command that aiida executed so that I can compare?
Is there a way to update an existing code without creating a new one?

code_pk = 323
code = aiida.orm.load_code(code_pk)
structure = aiida.orm.load_node(105)
pseudo_family = load_group('SSSP/1.3/PBE/efficiency')
pseudos = pseudo_family.get_pseudos(structure=structure)
parameters = {
'CONTROL': {'calculation': 'scf', },
'SYSTEM': {'ecutwfc': 30.,  # wave function cutoff in Ry
'ecutrho': 240.,  # density cutoff in Ry
  },
}
KpointsData = DataFactory('core.array.kpoints')
kpoints = KpointsData()
kpoints.set_kpoints_mesh([4,4,4])
builder = code.get_builder()
builder.structure = structure
builder.pseudos = pseudos
builder.parameters = Dict(parameters)
builder.kpoints = kpoints
builder.metadata.options.resources = {'num_machines': 1}
from aiida.engine import submit
calcjob_node = submit(builder)

The process report is given below:

(aiidapy313) [misbah@hpc _prelim]$ verdi process report 356*** 356: None*** (empty scheduler output file)*** Scheduler errors:
As of version 3.0.0, the “sm” BTL is no longer available in Open MPI.

Efficient, high-speed same-node shared memory communication support inOpen MPI is available in the “vader” BTL.  To use the vader BTL, youcan re-run your job with:

mpirun --mca btl vader,self,... your_mpi_application

A requested component was not found, or was unable to be opened.  Thismeans that this component is either not installed or is unable to beused on your system (e.g., sometimes this means that shared librariesthat the component requires are unable to be found/loaded).  Note thatOpen MPI stopped checking at the first component that it did not find.

Host:      hpc.giki.edu.pkFramework: btlComponent: sm

It looks like MPI_INIT failed for some reason; your parallel process islikely to abort.  There are many reasons that a parallel process canfail during MPI_INIT; some of which are due to configuration or environmentproblems.  This failure appears to be an internal failure; here’s someadditional information (which may only be relevant to an Open MPIdeveloper):

mca_bml_base_open() failed → Returned “Not found” (-13) instead of “Success” (0)

[hpc:44175] *** An error occurred in MPI_Init[hpc:44175] *** reported by process [465764353,0][hpc:44175] *** on a NULL communicator[hpc:44175] *** Unknown error[hpc:44175] *** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,[hpc:44175] ***    and potentially your MPI job)_aiidasubmit.sh: line 8: --mca: command not found

*** 5 LOG MESSAGES:
±> WARNING at 2025-09-22 06:53:03.494035+00:00| There was nonempty stderr in the whoami command: open terminal failed: not a terminal
±> WARNING at 2025-09-22 06:53:07.442177+00:00| key ‘symmetries’ is not present in raw output dictionary
±> ERROR at 2025-09-22 06:53:07.521282+00:00| ERROR_OUTPUT_STDOUT_INCOMPLETE
±> ERROR at 2025-09-22 06:53:07.527939+00:00| Both the stdout and XML output files could not be read or parsed.
±> WARNING at 2025-09-22 06:53:07.530391+00:00| output parser returned exit code<305>: Both the stdout and XML output files could not be read or parsed.

Thank you for your help.

t-reents · September 22, 2025, 8:20am

Hi @Walser52

To inspect the submission file that was used, you can run verdi calcjob inputcat <your-pk> _aiidasubmit.sh. This will print/cat the content of the file in your terminal.

Moreover, since you are using a single PwCalculation, you can also set builder.metadata.dry_run = True before submitting. This will not submit any calculation but rather writes the input and submission file into submit_test (and creates a sub-directory based on the date and time for each dry-run) in your current working directory. In this way, you can also inspect how the submission will look like without actually submitting it to a remote machine.

Hope that helps.

mbercx · September 22, 2025, 8:32am

Lightning-fast response @t-reents. ^^ Some additional comments from my end:

Another hidden gem is verdi calcjob gotocomputer. This will have you ssh into the remote machine, where you can see all the files of the corresponding calculation job. I often try to figure out what I need in the _aiidasubmit.sh file to get the calculation to run for that machine, then adapt my metadata.options accordingly.
Regarding updating a code, AiiDA doesn’t let you at the moment. It’s been a pet peeve of mine for some time. The reason is of course that in case you can update a code, you would change the history or provenance of any calculations that used that code. Having a “copy-on-edit” feature has also been proposed in this context, but we never settled on a solution that got implemented.

Note that the message in the report is not generated by AiiDA, it will simply pipe the stderr into _scheduler-stderr.txt and report that. So the actual message is from the scheduler on your remote machine.

EDIT: Let me also link a road map item on this topic:

github.com/aiidateam/team-compass

Usability: Make computers and codes entities more flexible

opened 04:33PM - 01 Mar 23 UTC

ramirezfranciscof

roadmap/proposed

## Motivation When computers and codes were originally designed, a certain le…vel of rigidity on their properties was established in order to try to impose a stricter level of control over the preservation of the provenance. However, the frequent changes an updates that computer clusters perform on their systems forces users to be also updating the ways in which they access those resources. Part of this required versatility was taken into account by allowing some parameters of computer access to be set in the `configuration` of the computer, rather than in its attributes during setup. However, many users find that this is not enough and express frustration at the limitations that these entities still have. Some examples include: - After updating software in clusters, some software change location and thus the respective code needs to be updated or changed to a new one. Creating a new code is the natural solution, but it tends to then over-complicate querying later. - One may want to set up a computer to be able to be used with different schedulers (specially considering the inclusion of meta-schedulers such as [aiida-hyperqueue](https://github.com/aiidateam/aiida-hyperqueue)), but by doing so the remote folders created by one are technically in a different computer and thus copying of files between those is not allowed. ## Desired Outcome Rethink how immutability works in the computer and code nodes, what restrictions can be relaxed, which ones we want to keep to ensure a cleaner provenance, and what other options can we offer users to facilitate dealing with these immutability problems in the cases that need to remain. ## Impact This will potentially improve usability for all users of AiiDA. ## Complexity Since this is not a bug or missing feature, but a direct consequence of how we enforce the "respect" for the provenance and this clashing or resulting inconvenience in practical usage, the biggest issue here is to rethink a more general principle of AiiDA. Since this is coming from unforeseen practical inconvenience, it is worth it to first try to do a thorough compilation of user frustrations and also some thinking into other potential problems. ## Progress The first task should be to do a comprehensive analysis of the situation and gather any remaining use case that may be relevant to consider.

@geiger_j this is something I might work on in the next coding days.

Walser52 · September 22, 2025, 10:52am

Thanks a lot. I now understand that I should have used the option mpirun_extra_params to modify my mpirun. The option I actually used was append-text and that generated the following output:

#!/bin/bashexec > _scheduler-stdout.txt
exec 2> _scheduler-stderr.txt
‘mpirun’ ‘-np’ ‘1’ ‘/usr/local/bin/pw.x’ ‘-in’ ‘aiida.in’  > ‘aiida.out’

–mca btl vader,self

Unfortunately I’ve tried this on the terminal and it says no such option. How do I pass the –mca btl vader,self to my mpirun?

verdi code create core.code.installed --label pw --computer localhost --filepath-executable /usr/local/bin/pw.x --default-calc-job-plugin quantumespresso.pw --non-interactive --mpirun_extra_params “–mca btl vader,self”

Edit:
I have been able to setup the right command from the builder by using:

builder.metadata.options.mpirun_extra_params = ['--mca', 'btl','vader,self']

Still not sure how to setup the code from the terminal so that it does it automatically.

t-reents · September 22, 2025, 11:41am

Happy to hear that it helped.

builder.metadata.options.mpirun_extra_params = [‘–mca’, ‘btl’,‘vader,self’]

This is indeed the right way to do it.

As you also pointed out, you can’t directly specify this as a separate option in the cli when creating a Code. The only way I see to do it permanently is to change/setup the mpirun-command of a Computer. However, this might not be the best approach, since you might have multiple Codes associated to that Computer that don’t require this option.
Hence, the best way is indeed to just set the mpirun_extra_params in your submission scripts.

Topic		Replies	Views
aiida.engine.processes.calcjobs.tasks.PreSubmitException: exception occurred in presubmit call New to AiiDA	2	107	February 2, 2024
Srun does not appear in the job submission file New to AiiDA	2	51	October 28, 2024
Configure Aiida to use remote HPC General Usage question	40	736	February 14, 2024
Stderr=sbatch: error: Batch job submission failed: Memory required by task is not available New to AiiDA	8	192	February 18, 2024
Error 305 on tutorial pw.x simulation New to AiiDA question	0	36	June 9, 2025

Finding exact command that was executed

Related topics