The pilot sends detailed information about file transfers to Rucio. Here is a list of the different fields contained in the trace report.

Field name Explanation
appid PanDA job id
catStart time stamp when rucio has been queried for replica information
dataset dataset name (prodDBlock from job definition - or destinationDblock)
duid (currently not set by the pilot)
eventType type of trace; currently the following event types are sent by the pilot (ordered by frequency; in the case of user jobs, an _a is added to the type):
1. get_sm_a: stagein inputs for analysis jobs
2. get_sm: stagein inputs for non-analysis jobs
3. put_sm: stageout outputs for non-analysis jobs
4. get_es: stagein inputs for eventservice jobs
5. download: EXPLANATION
6. put_sm_logs: stageout logs for non-analysis jobs
7. put_sm_a: stageout outputs for analysis jobs
8. put_sm_logs_a: stageout logs for analysis jobs
9. upload: EXPLANATION
10. put_es: stageout outputs for eventservice jobs
11. put_sm_logs_os: stageout logs to objectstore (special transfer defined in agis per panda queue)
eventVersion pilot version
filename the local file name (LFN)
filesize file size
guid GUID with removed '-' signs,
guid.replace('-', '')
hostname host name,
ip IP number of local host:
localSite PanDA site name (if set by pilot's or DDM endpoint (if set by pilot's movers/ - should only be DDM endpoints?
protocol name of copy tool used by the pilot; e.g. xrdcp
relativeStart time stamp when stage-in or stage-out begins
remoteSite DDM endpoint
scope replica scope (note: missing in pilot's TraceReport class init function)
stateReason error message or explanation; e.g. BAD_COPYTOOL, OK, 'skip stagein file'
suspicious (currently not being set by the pilot - a corrupt file is currently reported to rucio via
in the rucio.client)
taskid PanDA task ID
timeEnd time stamp when transfer or replica lookup, etc has been failed by the pilot
timeStart start time of trace report;
transferStart currently the same time stamp as relativeStart, see above
url TURL in case of direct access for given replica. Note: in the case of direct access, the copy tool is not involved by rucio is queried for the replica info. The corresponding trace will contain the TURL, as well as clientState='FOUND_ROOT' and stateReason='direct_access'. In event service mode and when Prefetcher is used, the TURL, clientState='FOUND_ROOT' and stateReason='prefetch' are also added to the trace report
usr hash of the user DN (usrdn field):
usrdn user distinguished name (DN)
uuid hash of the pilot id (if it exists):
hashlib.md5('ppilot_%s' % job.jobDefinitionID).hexdigest()
commands.getoutput('uuidgen -t 2> /dev/null').replace('-','')
validateStart time stamp when pilot performs checksum verification during stage-in and stage-out
version (currently not used by the pilot)

-- PaulNilsson - 2018-02-16

This topic: PanDA > AtlasDistributedComputing > PanDA > Pilot2 > PilotRucioTraces
Topic revision: r4 - 2018-02-28 - IlijaVukotic
This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2021 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback