Skip to content

IncidentSvc: fix ever-growing incident map

Frank Winklmeier requested to merge fwinkl/Gaudi:incidentsvc into master

The map used for the storage of the asynchronous incidents was never cleared during the job and each new EventContext added a new entry into the map that only got cleared during finalize() (see !758 (merged)). Change the storage to be slot-specific and clear the list of incidents when the event changes for a given slot. Thanks to @rbielski for reporting this as part of https://its.cern.ch/jira/browse/ATR-22070.

In addition a few unrelated cleanups (in a separate commit):

  • Make IncidentProcAlg a re-entrant algorithm.
  • Fix AsyncIncidents example job options.
  • Minor code cleanups in incident handling.
Edited by Frank Winklmeier

Merge request reports