Skip to content

Brainstorming tripyview integration #1211

@JanStreffing

Description

@JanStreffing

How do we want to integrate:

  • As a component
  • As a further reading

Tripyview needs to make use of python try: for more robustness.

We need to be able to launch failed jobs manually with -t post. Might have to fix dates and run number for this to work.

We can use inspect feature to visualize workflow, and shall take advantage of workflow manager.

The post jobs need to run in separate slurm jobs, filling one compute node.

We should automatically install environment from conda env yaml file if it does not exist yet. Multiple options where to install:

  • In the experiment folder (but will lead to many files, problem e.g. on juwels)
  • In a container in the experiment folder (but not all machines have, e.g. sigularity)
  • In home folder (but not all machines have enough space there)
  • In a custom folder under /work/

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions