Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-86817

ClusterChunksResizePolicy does not fully clean its state upon completion

    • Type: Icon: Bug Bug
    • Resolution: Fixed
    • Priority: Icon: Minor - P4 Minor - P4
    • 6.0.15
    • Affects Version/s: 6.0.14
    • Component/s: None
    • Labels:
    • Catalog and Routing
    • Fully Compatible
    • ALL
    • CAR Team 2024-02-19, CAR Team 2024-03-04

      SERVER-64171 added a new step to the FCV downgrade process from 6.0 to 5.x to enforce the maximum chunk size on each sharded collection of the cluster, implemented within ClusterChunksResizePolicyImpl.

      This is a best-effort and potentially long-running procedure, which per-collection progress its persisted on disk by setting an optional field on the related config.collections document and cleared upon completion.

      The clear logic is nevertheless flawed, since the clear operation (which is meant to target all the entries matching the find clause) gets translated into an update operation with the multi option set to false.

            paolo.polato@mongodb.com Paolo Polato
            paolo.polato@mongodb.com Paolo Polato
            0 Vote for this issue
            3 Start watching this issue