Project

General

Profile

Feature #14997

MetricManager should be a "black hole" for metrics

Added by Eric Flumerfelt almost 4 years ago. Updated almost 4 years ago.

Status:
Closed
Priority:
Normal
Category:
-
Start date:
01/03/2017
Due date:
% Done:

100%

Estimated time:
(Total: 0.00 h)
Experiment:
-
Co-Assignees:
Duration:

Description

Calls to MetricManager::sendMetric should return as quickly as possible, regardless of the speed of the underlying metric plugins.

This has been implemented by a shared queue of metric data and a metric-sending thread. There's still probably improvements that can be made (i.e. condition_variable signaling of the metric thread).


Subtasks

Bug #14996: double free or corruption in MetricManagerClosedEric Flumerfelt

Support #15083: Make MetricPlugin const-correctClosedEric Flumerfelt

History

#1 Updated by Eric Flumerfelt almost 4 years ago

  • Status changed from Assigned to Resolved
  • % Done changed from 70 to 100

Implemented signaling in MetricManager using std::condition_variable. The thread will wake up every 100 ms regardless to check if the run was ended.

#2 Updated by Eric Flumerfelt almost 4 years ago

  • Project changed from artdaq to artdaq Utilities
  • Status changed from Resolved to Closed
  • Target version changed from 577 to artdaq-utilities v1_01_00

Also available in: Atom PDF