Skip to content

🐛 Async Engine: retrieval of files before the job is completed? #7023

@mbercx

Description

@mbercx

Describe the bug

I'm getting weird issues while running with the core.ssh_async transport. I'm getting calculations that seem to have their stdout retrieved before the job is complete.

Steps to reproduce

The issue is transient, so difficult to reproduce with a number of steps. That said, it was happening enough for me to quickly find 2 cases in ~50 runs. Probably more.

@khsrali this is happening on Thanos in the mc3d-relax project, check the dev profile, node PK = 13300. If you run

verdi calcjob outputcat 13300 | tail

You'll the calculation is interrupted, i.e. not the typical JOB DONE at the end. But if you go to the calculation directory

verdi calcjob gotocomputer 13300

and run tail aiida.out, you will see JOB DONE.

Your environment

  • Operating system [e.g. Linux]: Ubuntu 22.04.5 LTS (Thanos)
  • Python version [e.g. 3.7.1]: 3.13.7
  • aiida-core version [e.g. 1.2.1]: main branch: 7255f01

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions