torc 0.22.2

Workflow management system
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48

# Test: Multi-Node Direct Execution
#
# 2-node allocation with direct execution mode (no srun wrapping for jobs).
# The head worker spawns one torc-slurm-job-runner per node via
# srun --ntasks-per-node=1, and each per-node worker executes jobs directly.
# Tests that single-node jobs are distributed across nodes in direct mode.

name: multi_node_direct
description: 2-node allocation — 20 jobs x 5 CPUs via direct execution
project: slurm-tests

execution_config:
  mode: direct

resource_monitor:
  enabled: true
  granularity: time_series
  sample_interval_seconds: 2

resource_requirements:
  - name: work_resources
    num_cpus: 5
    num_nodes: 1
    memory: 2g
    runtime: PT3M

jobs:
  - name: work_{i}
    command: bash -c 'echo "Running on $(hostname)"; stress-ng --cpu 5 --timeout 30 --metrics-brief'
    resource_requirements: work_resources
    scheduler: two_node_scheduler
    parameters:
      i: 1:20

slurm_schedulers:
  - name: two_node_scheduler
    account: PLACEHOLDER_ACCOUNT
    partition: PLACEHOLDER_PARTITION
    nodes: 2
    walltime: "00:10:00"

actions:
  - trigger_type: "on_workflow_start"
    action_type: "schedule_nodes"
    scheduler: "two_node_scheduler"
    scheduler_type: "slurm"
    num_allocations: 1
    start_one_worker_per_node: true