/opt/chainguard/nemo-rl/venvs/mcore/bin/nvrx-smonsvc

usage: nvrx-smonsvc [-h] [--attrsvc-endpoint ATTRSVC_URL] [--interval INTERVAL] [--user USER]
                    [--partitions PARTITIONS [PARTITIONS ...]] [--job-pattern JOB_PATTERN]
                    [--timeout TIMEOUT] [--port PORT] [-v]

Monitor SLURM jobs and integrate with attribution service

options:
  -h, --help            show this help message and exit
  --attrsvc-endpoint ATTRSVC_URL
                        Attribution service URL or UDS endpoint (env: NVRX_ATTRSVC_ENDPOINT,
                        default: http://localhost:8000)
  --interval INTERVAL   Poll interval in seconds (env: NVRX_SMONSVC_INTERVAL, default: 180)
  --user USER           Filter jobs by specific user (env: NVRX_SMONSVC_USER, default: all users)
  --partitions PARTITIONS [PARTITIONS ...]
                        SLURM partitions to monitor (env: NVRX_SMONSVC_PARTITIONS space-separated,
                        default: batch batch_long)
  --job-pattern JOB_PATTERN
                        Regex pattern to match job names (env: NVRX_SMONSVC_JOB_PATTERN)
  --timeout TIMEOUT     HTTP request timeout in seconds (env: NVRX_SMONSVC_TIMEOUT, default: 60)
  --port PORT           Port for HTTP server (env: NVRX_SMONSVC_PORT, disabled if not set)
  -v, --verbose         Enable debug logging (env: NVRX_SMONSVC_VERBOSE)

Examples:
  # Monitor jobs in default partitions (batch, batch_long)
  nvrx-smonsvc --attrsvc-endpoint http://localhost:8000 --port 8100

  # Monitor through attrsvc HTTP-over-UDS
  nvrx-smonsvc --attrsvc-endpoint unix:///tmp/nvrx-attrsvc.sock --port 8100

  # Monitor specific partitions
  nvrx-smonsvc --partitions gpu interactive --attrsvc-endpoint http://localhost:8000 --port 8100

  # Monitor specific user's jobs
  nvrx-smonsvc --user foo_bar --attrsvc-endpoint http://attrsvc.cluster.local:8000 --port 8100

  # Monitor jobs matching a pattern
  nvrx-smonsvc --job-pattern "train_.*" --attrsvc-endpoint http://localhost:8000 --port 8100

  # Run both services via SLURM
  sbatch scripts/nvrx_services.sbatch