Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-2058

introduce a metric that shows collection scans taking place

    XMLWordPrintable

    Details

    • Type: New Feature
    • Status: Closed
    • Priority: Major - P3
    • Resolution: Duplicate
    • Affects Version/s: None
    • Fix Version/s: None
    • Component/s: Diagnostics
    • Labels:

      Description

      If an application scans by index, then serverStatus().indexCounters.btree.accesses shows the counts and activity against the index. However, if your application is performing scans not using an index (perhaps sub-optimally) there is no metric that shows anything. If a metric was exposed then it would be possible to spot poor plans/scans as well as do ratios of scans vs index accesses. Just a typical pattern for DBA's to see what the DB is doing at any given time. Perhaps the metric could be named: serverStatus().collections.scans. This is also good for an application where scans are needed (like the reads are most of the table) to show the DB activity during these periods. It's also a good measure of overall workload of the DB (scans/sec) and helps to differentiate different I/O patterns/performance. For instance how it changes when faults > 0, and also how it changes platform to platform (aka: I can scan more blocks with platform X).

        Attachments

          Issue Links

            Activity

              People

              Assignee:
              backlog-server-query Backlog - Query Team
              Reporter:
              kgorman Kenny Gorman
              Participants:
              Votes:
              0 Vote for this issue
              Watchers:
              8 Start watching this issue

                Dates

                Created:
                Updated:
                Resolved: