Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-47258

Refactor the WiredTigerOplogManager

    • Type: Icon: Improvement Improvement
    • Resolution: Fixed
    • Priority: Icon: Major - P3 Major - P3
    • 4.4.0-rc8, 4.7.0
    • Affects Version/s: None
    • Component/s: Storage
    • None
    • Fully Compatible
    • v4.4
    • Execution Team 2020-04-06, Execution Team 2020-04-20
    • 0

      Remove the small delay to update the oplogReadTimestamp when there are no callers (waitForAllEarlierOplogWritesToBeVisible) awaiting updates to the oplogReadTimestamp. This code is extra logic added in order not to overwhelm the system with too frequent I/O when the thread used to call waitUntilDurable() (flush to disk). The waiting logic was added in this commit in v3.6, to handle the waitUntilDurable() call also added in v3.6. There is no longer a waitUntilDurable() call in the code since the Replicate Before Journaling project completed.

      We also need to rename the functions, etc., to better reflect that journal flushing no longer happens.

      UPDATE:

      This is going to be a purely non-functional change now. The performance results were not good removing the delay. See the CR for details.

            Assignee:
            dianna.hohensee@mongodb.com Dianna Hohensee (Inactive)
            Reporter:
            dianna.hohensee@mongodb.com Dianna Hohensee (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            1 Start watching this issue

              Created:
              Updated:
              Resolved: