Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-67029

Make aggregate command in internal_transactions* workloads robust against failover/shutdown or collection rename

    • Type: Icon: Bug Bug
    • Resolution: Fixed
    • Priority: Icon: Major - P3 Major - P3
    • 6.0.0-rc9, 6.1.0-rc0
    • Affects Version/s: None
    • Component/s: None
    • Labels:
      None
    • Fully Compatible
    • ALL
    • v6.0
    • Sharding NYC 2022-06-13
    • 60

      As shown in BF-25473, the aggregate command used to get a random document involves running getMore commands which can fail with a QueryPlanKilled error in the following cases:

      1. The node steps down or restarts between when the cursor is opened and when a getMore command arrives at the node.
      2. The node renames the collection due to resharding between when the cursor is opened and when a getMore command arrives at the node.

      This is expected and the internal_transactions* workloads should retry the aggregate command in these cases. Alternatively, the aggregate command should be run with a large batchSize to avoid getMore commands.

            Assignee:
            cheahuychou.mao@mongodb.com Cheahuychou Mao
            Reporter:
            cheahuychou.mao@mongodb.com Cheahuychou Mao
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

              Created:
              Updated:
              Resolved: