Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-23663

New primary syncs from chosen node to catch up with timeout

    • Type: Icon: Task Task
    • Resolution: Done
    • Priority: Icon: Major - P3 Major - P3
    • 3.3.11
    • Affects Version/s: None
    • Component/s: Replication
    • Labels:
    • Fully Compatible
    • Repl 15 (06/03/16), Repl 16 (06/24/16), Repl 17 (07/15/16), Repl 18 (08/05/16)

      To avoid unnecessary rollback after a PV1 election, the new primary finds the node with the highest lastOpTime, then tries to sync all the oplog from that node to catch up its opTime. This ticket is for the catch-up phase.

      If the catch-up timeout expires before we exhaust all the sync source’s oplog entries, the new primary ends the catch-up phase and proceeds to exit drain mode like normal.
      If the query from the remote node fails, treat the same as if the catch-up timeout expired. Do not attempt to choose a new sync source.

            siyuan.zhou@mongodb.com Siyuan Zhou
            crystal.horn@mongodb.com Crystal Horn
            0 Vote for this issue
            15 Start watching this issue