lst_auto_rta.Auto_RTA

Automatically start/stop RTA reconstruction pipeline for new runs during a observation night

New runs are found querying the TCU pymongo database regularly. This script: - copies the conda environment and reconstruction model to the RAM of the slurm nodes - queries the TCU pymongo database for new runs regularly and for each new run:

  • stops the r0->dl1 daemons for the previous runs

  • starts the r0->dl1 daemons for the new run, using a static configuration (CDB configuration) from disk, and writing the dynamic configuration to disk as well.

  • starts the “engineering gui” scripts allowing to monitor data processing for the run.

  • stops at a fixed hour according to its configuration file

  • cleans the RAM of slurm worker nodes

Functions

assign_worker_to_data_connection(...)

Assign a worker node from slurm_nodes to each data server connection in tel_ids_to_data_servers

main()

Entrypoint of Auto_RTA:

nuke_rta(slurm_account)

Hard stop of All RTA jobs: scancels all jobs of slurm_account

scancel_jobs(job_ids, signal[, delay_s, ...])

Run scancel on `job_ids, sending signal after delay_s seconds.

srun_cmd_worker_nodes(connection_jobs_info, cmd)

Submit a slurm command with srun on all worker nodes in connection_jobs_info

stop_rta(slurm_reservations, slurm_account, ...)

Stop the r0dl1 daemons

write_reco_manager_observation_config(...)

Writes the observation configuration for the reco-manager

Classes

ConnectionJobInfo(tel_id, hostname, port, ...)