Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-37952

Restart index builds from persisted recovery points

    • Type: Icon: Task Task
    • Resolution: Won't Do
    • Priority: Icon: Major - P3 Major - P3
    • None
    • Affects Version/s: None
    • Component/s: Storage
    • None
    • Storage Execution
    • Storage NYC 2019-01-28

      Index builds must be able to vote for commit at some point that is recoverable after server restart. Re the design refactor, index builds will have two phases, scanning/building and draining. Once the index build reaches draining, it should be recoverable. The draining phase will include constraint violations checking prior to commit. Therefore, index builds must be able to move from draining the side table, to constraint checking or committing, back to draining again.

      SERVER-37645 adds parsing and verification logic for the new fields in the index catalog entries to facilitate recovery.
      SERVER-38986 figures out how to acquire a timestamp at which to update the index catalog entry fields separate from the initial and final writes for start and commit index build, respectively.

            Assignee:
            backlog-server-execution [DO NOT USE] Backlog - Storage Execution Team
            Reporter:
            dianna.hohensee@mongodb.com Dianna Hohensee (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            5 Start watching this issue

              Created:
              Updated:
              Resolved: