Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-27036

WT takes longer to start than MMAPv1 on EBS volumes that have not been intialized

    • Type: Icon: Bug Bug
    • Resolution: Done
    • Priority: Icon: Major - P3 Major - P3
    • None
    • Affects Version/s: None
    • Component/s: WiredTiger
    • None
    • Fully Compatible
    • ALL

      Hey,

      We recently upgraded from MMAP to WT. For the same data amount (~500GB), on an AWS EBS volume which is "cold" (restored from a snapshot) - size 2TB -
      MMAP took ~5 minutes to start
      WT takes 50 minutes, log mentioning this:

      Taking 1005 samples and assuming that each section of oplog contains approximately 5672862 records totaling to 1073741955 bytes
      

      It apparently scans the oplog (we've set its size to 100GB) - and we see the iowait on 100%. But still, 50 minutes is a lot...

      Anything we can do to speed things up?

        1. log
          13 kB
        2. tmp.zip
          898 kB

            Assignee:
            Unassigned Unassigned
            Reporter:
            yonido Yoni Douek
            Votes:
            0 Vote for this issue
            Watchers:
            18 Start watching this issue

              Created:
              Updated:
              Resolved: