Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-50500

Increment topologyVersion for state change errors in failCommand

    • Type: Icon: Improvement Improvement
    • Resolution: Won't Do
    • Priority: Icon: Major - P3 Major - P3
    • None
    • Affects Version/s: None
    • Component/s: None
    • Labels:
      None
    • Repl 2020-09-07, Repl 2020-09-21

      This was originally proposed in SERVER-45686 but was not implemented because at that time, the topology version did not change on mongos or standalones. This is no longer true as of 4.5.1 because mongos increments topologyVersion as part of the quiesce mode project (SERVER-46957). Can we revisit the original request? It would be helpful for drivers testing because the drivers API does not allow heartbeat frequency to be set below 500ms (non-configurable minHeartbeatFrequency), so even tuning it as low as possible means that tests still wait ~0.5s to recover from state change errors when using failCommand.

      EDIT: My comment about the 0.5s lag only apply to a subset of state change errors on 4.4 where the response does not contain a topologyVersion field. The drivers tests addressed by this ticket run on 4.4+, so it'd be helpful to have this functionality backported. I think we'd need to first backport SERVER-47017 and SERVER-47018, though.

            Assignee:
            tess.avitabile@mongodb.com Tess Avitabile (Inactive)
            Reporter:
            divjot.arora@mongodb.com Divjot Arora (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

              Created:
              Updated:
              Resolved: