Skip to content

Latest commit

 

History

History
127 lines (96 loc) · 2.76 KB

quickstart.rst

File metadata and controls

127 lines (96 loc) · 2.76 KB

Quick Start

This guide will help you get up and running with SO Campaign Manager quickly.

Basic Workflow

  1. Create a configuration file - Define your campaign in a TOML file
  2. Run the campaign - Execute using the command-line interface
  3. Monitor progress - Track execution and results

Example Configuration

Create a file called campaign.toml:

[campaign]
deadline = "2d"

[campaign.resources]
nodes = 4
cores-per-node = 112

[campaign.ml-mapmaking]
context = "file:///path/to/context.yaml"
area = "file:///path/to/area.fits"
output_dir = "/path/to/output"
bands = "f090"
wafer = "ws0"
comps = "TQU"
maxiter = 10
query = "obs_id='1575600533.1575611468.ar5_1'"
tiled = 1
site = "act"

[campaign.ml-mapmaking.resources]
ranks = 1
threads = 32
memory = "80000"  # in MBs
runtime = "80000" # in seconds

[campaign.ml-mapmaking.environment]
MOBY2_TOD_STAGING_PATH = "/tmp/"
DOT_MOBY2 = "/path/to/act_dot_moby2"
SOTODLIB_SITECONFIG = "/path/to/site.yaml"

Running Your First Campaign

Execute your campaign with:

socm -t campaign.toml

This will:

  1. Parse your configuration
  2. Create the required workflows
  3. Set up the computing resources
  4. Submit jobs to SLURM
  5. Monitor execution

Understanding the Output

The campaign manager will:

  • Create output directories as specified in your configuration
  • Log execution details
  • Generate maps and analysis products based on your workflows

Next Steps

Simple Python Example

You can also use SO Campaign Manager programmatically:

from socm.bookkeeper import Bookkeeper
from socm.core import Campaign, Resource
from socm.workflows import MLMapmakingWorkflow

# Create workflows
workflow = MLMapmakingWorkflow(
    name="test_workflow",
    executable="so-site-pipeline",
    subcommand="make-filterbin-map",
    context="path/to/context.yaml",
    # ... other parameters
)

# Create campaign
campaign = Campaign(
    id=1,
    workflows=[workflow],
    campaign_policy="time"
)

# Define resource
resource = Resource(
    name="tiger3",
    nodes=4,
    cores_per_node=112,
    memory_per_node=100000000,
    default_queue="tiger-test"
)

# Execute
bookkeeper = Bookkeeper(
    campaign=campaign,
    resources={"tiger3": resource},
    policy="time",
    target_resource="tiger3"
)

bookkeeper.run()