Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-17468

actionlog should not log every single balancer round

    XMLWordPrintable

    Details

    • Type: Improvement
    • Status: Closed
    • Priority: Major - P3
    • Resolution: Fixed
    • Affects Version/s: 3.0.0
    • Fix Version/s: 3.2.5, 3.3.3
    • Component/s: Sharding
    • Labels:
    • Backwards Compatibility:
      Fully Compatible
    • Backport Completed:
    • Backport Requested:
      v3.0
    • Sprint:
      Sharding 10 (02/19/16), Sharding 11 (03/11/16)

      Description

      The sharding config collection actionlog added for SERVER-8752 logs every single balancer round, including when the balancer acquires the balancer lock and finds absolutely nothing to do.

      If you have several dozen mongos processes, it's conceivable that each of these balancing rounds will be happening every second or two and since actionlog is 2MB capped collection, after about 3-4 hours we will have overwritten all previous actions. Is there value of recording "balancer thread woke up, got the lock, found nothing to do and did nothing"? If there was nothing to do during balancing round, did it really happen? Even if it did, is there any value to recording that over and over again?

        Attachments

          Activity

            People

            • Votes:
              1 Vote for this issue
              Watchers:
              11 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: