Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-84306

checkMetadataConsistency default maxTimeMS might be too small

    XMLWordPrintableJSON

Details

    • Icon: Bug Bug
    • Resolution: Duplicate
    • Icon: Major - P3 Major - P3
    • None
    • None
    • Sharding
    • Catalog and Routing
    • ALL
    • CAR Team 2024-01-22, CAR Team 2024-02-05
    • 5
    • 1

    Description

      SERVER-74251 added a new hook to check if the cluster have consistent metadata while running concurrency suites. This adds an extra layer of checks to ensure that in the presence of a busy cluster with simultaneous operations the metadata remains consistent.

      The problem with this approach is that we're running the check on a really busy cluster, and in the slowest machines in our test infrastructure, we might end up timing out the default 30 seconds before finishing the check. We don't want to fully get rid of this timeout, because this could also help us find deadlocks.

      We should manually set the maxTimeMS in the checkMetadataConsistency command, instead of allowing the default value to be set, to a more reasonable time to allow the check to finish even if the cluster is loaded, something like 5 minutes.

      Attachments

        Activity

          People

            yujin.kang@mongodb.com Yujin Kang Park
            marcos.grillo@mongodb.com Marcos José Grillo Ramirez
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: