Skip to content

eCLM on jupiter running with 2 or more nodes needs UCX-settings/DC module #121

@AGonzalezNicolas

Description

@AGonzalezNicolas

Environment used: jsc.2025.gnu.openmpi
To run eCLM with N=2 or more nodes the module UCX-settings/DC needs to be uploaded for the run (not needed for the build).

If UCX-settings/DC is not uploaded, the error is the following:
log-err file:

[jpbo-113-42.jupiter.internal:1748254] pml_ucx.c:431  Error: ucp_ep_create(proc=546) failed: Input/output error
[jpbo-113-42.jupiter.internal:1748049] pml_ucx.c:431  Error: ucp_ep_create(proc=464) failed: Input/output error
[jpbo-113-42.jupiter.internal:1748125] pml_ucx.c:431  Error: ucp_ep_create(proc=415) failed: Input/output error

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    Status

    Issues

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions