Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-27230

Poor compression of diagnostic data when replica set is unhealthy

    XMLWordPrintable

    Details

    • Backwards Compatibility:
      Fully Compatible
    • Operating System:
      ALL
    • Backport Requested:
      v3.6, v3.4, v3.2
    • Sprint:
      Repl 2018-01-15, Repl 2018-01-29, Repl 2018-02-26, Repl 2018-03-12, Repl 2018-03-26, Repl 2018-04-09

      Description

      Under some circumstances involving network or other cluster health issues we can see frequent diagnostic data schema changes due to "lastHeartbeatMessage" and "syncingTo" fields that disappear and reappear frequently in replica set status. In one case this was observed to result in about 5x larger compressed ftdc data (~1300 bytes per sample vs the normal ~250 bytes per sample) and therefore about 5x shorter retention period, which is unfortunate because one use case for ftdc is diagnosing the cause of an unhealthy cluster.

      Since the fields in question have string values they don't contribute to ftdc data so shouldn't participate in determining whether there is a schema change.

        Attachments

          Issue Links

            Activity

              People

              • Votes:
                0 Vote for this issue
                Watchers:
                9 Start watching this issue

                Dates

                • Created:
                  Updated:
                  Resolved: