Project

General

Profile

Bug #1524

Inputs hang each other up (especially ones that timeout)

Added by Luke Murphey over 7 years ago. Updated over 7 years ago.

Status:
Closed
Priority:
Normal
Assignee:
Target version:
Start date:
10/19/2016
Due date:
% Done:

100%


Description

To save memory, the input currently uses single instance mode (a single input running all of the inputs).

To fix this, I could:
  1. Change input away from single instance mode
    1. This is difficult to do because then I would have to use Splunk's interval which doesn't support time specifiers (like 5m).
    2. Would need to a process to convert these to Splunk's interval and the current interval; UI would likely need to accept and convert to/from Splunk's interval
  2. Switch to a multi-threading model

Subtasks

Task #1529: Change base class to use RLocksClosed

Task #1530: Create multiple threads for performing pingsClosedLuke Murphey

Task #1531: Clean up threads on shutdownClosedLuke Murphey

History

#1 Updated by Luke Murphey over 7 years ago

  • Subject changed from Inputs hang each other up to Inputs hang each other up (especially ones that timeout)

#4 Updated by Luke Murphey over 7 years ago

  • Target version set to 2.0

#5 Updated by Luke Murphey over 7 years ago

Things that need to be changed to support multi-threading:

#7 Updated by Luke Murphey over 7 years ago

Uncontesting access to locks doesn't appear to have much of a performance issue : http://stackoverflow.com/questions/11966471/python-cost-of-locking-vs-performance-does-multithreading-make-sense

#8 Updated by Luke Murphey over 7 years ago

  • Status changed from New to In Progress
  • Assignee set to Luke Murphey

#9 Updated by Luke Murphey over 7 years ago

  • Status changed from In Progress to Closed

Also available in: Atom PDF