Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-61792

Extending lock in batcher hurts performance

    XMLWordPrintableJSON

Details

    • Icon: Bug Bug
    • Resolution: Fixed
    • Icon: Major - P3 Major - P3
    • 5.2.0
    • None
    • None
    • None
    • Fully Compatible
    • ALL
    • Replication 2021-12-13
    • 135

    Description

      The solution to avoiding an uninterruptible lock in SERVER-61334 causes performance degradation in the insert_vector test. Experimentation shows that the issue is not the double-locking but rather the extension of the lock duration. This is because taking the global lock also takes the PBWM, which results in synchronizing the OplogBatcher to the OplogApplier.

      We should not take the PBWM nor any WT tickets for this particular lock acquisition, and audit other GlobalLock uses added as part of FileCopyBasedInitialSync to see if they should also not take the PBWM or WT tickets.

      Attachments

        Activity

          People

            matthew.russotto@mongodb.com Matthew Russotto
            matthew.russotto@mongodb.com Matthew Russotto
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: