Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-42385

election timeout callbacks are scheduled too frequently

    XMLWordPrintableJSON

Details

    • Icon: Bug Bug
    • Resolution: Fixed
    • Icon: Major - P3 Major - P3
    • 4.3.1
    • None
    • Replication
    • None
    • Fully Compatible
    • ALL
    • Repl 2019-08-26, Repl 2019-09-09, Repl 2019-09-23, Repl 2019-10-07, Repl 2019-10-21, Repl 2019-11-04

    Description

      In the process of investigating a BF, I discovered this behavior:
      https://logkeeper.mongodb.org/lobster/build/0b38b5af3083b9d185100bc1ce94b3a5/test/5d379645f84ae81c0108232f#bookmarks=0%2C8967&l=1

      If you jump to line 5707 in that log, you can see that "Scheduling election timeout callback at " is happening on average every 2-3 milliseconds, for many seconds. Additionally, even within one thread, the callback is not being repeatedly scheduled for times exclusively in the future of the previous scheduling; sometimes it goes backwards in time. This doesn't seem correct nor efficient.
      At least, can we change this log message to log level 5 so that it doesn't affect performance of the test by swamping the log system?

      Attachments

        Activity

          People

            matthew.russotto@mongodb.com Matthew Russotto
            milkie@mongodb.com Eric Milkie
            Votes:
            0 Vote for this issue
            Watchers:
            7 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: