Skip to content

Pilot Configuration

Paul Nilsson edited this page Apr 25, 2022 · 1 revision

The pilot configuration is stored in the default.cfg file (pilot3/pilot/util directory). The default values can be overridden locally but placing a config file in the pilot's launch directory. The various fields are described in the table below (in the order of appearance in the config file).

Section: Experiment

Field Default value Description
name ATLAS Pilot user name (currently unused).

Section: Pilot

Field Default value Description
pilotlog pilotlog.txt Main pilot log stored in the job log.
stageinlog stageinlog.txt Name of log file for stage-in.
stageoutlog stageoutlog.txt. Name of log file for stage-out.
pandajobdata pandaJobData.out Name of job definition file.
pandajob real Alternative: 'fake'. Used for pilot testing (see the following three fields).
testjobtype production Alternative: 'user'. Used for pilot testing.
testjobcommand normal Alternative: 'sleep' (sleep 1 will be executed as payload). Used for pilot testing.
testtransfertype NULL Alternative: 'direct'. Used for pilot testing.
pandaserver (see code) The URL for the PanDA server.
iddsserver (see code) The URL for the iDDS server.
heartbeat 1800 The heartbeat period in normal mode (30*60 = 1800 s).
debug_heartbeat 300 The heartbeat period in debug mode (5*60 = 300 s).
heartbeat_message heartbeat.json Heartbeat message file (only used when Pilot is not sending heartbeats to server).
jobid_file pandaIDs.out Job IDs can be stored to a file that is picked up by the wrapper.
free_space_limit 2 GB The minimum required disk space for the pilot to run a job.
free_space_limit_running 1 GB The minimum required disk space during running.
maximum_output_file_size 500 GB The maximum output file size. The default is currently 'infinity'.
maximum_input_file_sizes 14336 MB The maximum allowed sum of all input files (files accessed by direct access are not counted by the Pilot). Note: this is a fall-back value - the primary value is set in AGIS.
local_size_limit_stdout 2097152 kB Size limit of payload stdout size during running (default: 2 * 1024^2).
looping_verification_time 900 min Looping job time limits; if job does not write anything in N minutes, it is considered to be a looping. This field determines the frequency of the check.
looping_limit_default 7200 min Time limit for looping jobs. Any output must have been updated within this time
looping_limit_min_default 7200 min The minimum allowed looping limit, 2*3600 (reserved for future use).
kill_instruction_time 60 s Kill instruction monitoring time. Pilot looks for kill instruction file every N seconds. The file may be placed in the run directory by an external actor (e.g. Raythena actor). When intercepted, Pilot will abort the payload.
kill_instruction_filename pilot_kill_payload Kill instruction file name.
proxy_verification_time 600 s Proxy verification time (used by monitoring) in seconds.
payload_proxy_from_server True In case payload proxy should be downloaded from the server and be verified.
disk_space_verification_time 240 s Disk space monitoring time (check remaining disk space every N seconds).
memory_usage_verification_time 60 s Memory usage verification time (how often the memory monitor output will be checked)
process_verification_time 300 s Process verification time. This verification sets the environmental variable PILOT_MAXNPROC to the maximum number of found (child) processes corresponding to the main payload process id, PILOT_MAXNPROC may be used by the payload.
output_verification_time 300 s Output file size verification time. Currently not used.
thread_check 10 s The default thread check time in seconds, used by thread monitoring.
cpu_check 60 s The default CPU check time in seconds, used by CPU monitoring.
timing_file pilot_timing.json The timing file used to store various timing measurements.
error_log piloterrorlog.txt Optional error log. Currently not used.
redundant (cvmfs path) List of redundant files and directories to be removed prior to log file creation. Currently not used.

(more to come)