rhapsody_rm

Resource Manager for rhapsody.

rhapsody_rm discovers HPC node resources and carves them into partitions, then hands a partition to a rhapsody execution backend. It supports SLURM, PBSPro, LSF, Cobalt, Torque, and a FORK manager for single-node / local use.

The core is pure Python standard library — it has no runtime dependency on rhapsody. The two projects are coupled only by the documented partition dict contract (see CONTRACT.md).

Install

pip install .                 # core only
pip install '.[integration]'  # also installs rhapsody-py for the seam tests
pip install '.[dev]'          # test / lint tooling

Quick start

import rhapsody_rm

cfg = rhapsody_rm.RMConfig(requested_nodes=5, fake_resources=True)
rm  = rhapsody_rm.ResourceManager.get_instance(name="FORK", cfg=cfg)

print(f"found {len(rm.node_list)} nodes")

# allocate a partition and get a rhapsody-conforming spec
spec = rhapsody_rm.partition_spec(rm, part_id="p0", n_nodes=3)
# -> {"nodelist": [Node, ...], "env": {...}}  ->  resources["partition"]

Layout

src/rhapsody_rm/        # base.py + per-scheduler managers + integration.py
tests/                  # unit tests (run with: pytest)
CONTRACT.md             # the rhapsody <-> rhapsody_rm partition contract

Tests

pytest          # core unit tests (no scheduler or rhapsody needed)