Submit Flow¶
Status: current
The live submit path is experiment-YAML based.
Ray Launch¶
Flow:
- Expand any relative base YAML and validate
ExperimentConfig. - Build a
RunConfig. - Call
graphids.exp.run.launch_run. - Ray Train schedules the worker.
- The worker saves configuration and records startup, executes
fitortest, then publishes final metrics or failure in one manifest.
SLURM Submit¶
Flow:
- Expand any relative base YAML and validate
ExperimentConfig. - Build a
RunConfigas a submit-time validation check. - Render a Ray allocation sbatch script with resource directives from
resources. - Write the script to
{slurm_log_dir}/scripts/ray-{experiment_name}.sbatchand an expanded config snapshot named by experiment and content hash. - Submit it with
sbatch.
The compute node then runs:
Dry Run¶
This prints the sbatch script without writing config snapshots, scripts, or submitting jobs. The printed snapshot path is created by actual submission.
Files Of Interest¶
graphids/cli/exp.pygraphids/exp/config.pygraphids/exp/run.pygraphids/exp/slurm.pyscripts/slurm/_preamble.shscripts/slurm/_epilog.sh
The old gx run, gx exec --row, gx submit --row, and
gx plans submit path is historical and should not be used for new runs.
Submission supports one node/device. --source-root selects a frozen checkout,
and --dependency supplies a Slurm dependency expression. Python callers use
build_experiment_job, write_job and submit_script; recipes do not modify
rendered shell text.