Uploaded image for project: 'Indy Node'
  1. Indy Node
  2. INDY-1747

Find out why max node prod time increases during long load test

    Details

    • Type: Task
    • Status: Complete
    • Priority: Medium
    • Resolution: Done
    • Affects Version/s: None
    • Fix Version/s: 1.6.79
    • Component/s: None
    • Labels:
      None

      Description

      Some observations:

      • growth is same for loads of 1 NYM/sec and 10 NYMs/sec (so it doesn't depend on number of transactions written, although it's likely to depend on number of batches ordered)
      • growth is same when using RocksDB and LevelDB as storage
      • maximum node-to-node message size grows as well, but not as fast
      • average node prod time doesn't seem to be affected, which means spikes are in fact rare

      In my opinion prime suspect here is cyclic garbage collector, but more research is needed.

      Acceptance criteria

      • find out why max prod time grows
      • find out if this growth has upper bound
      • create issues for actual fix

        Attachments

          Activity

            People

            • Assignee:
              sergey.khoroshavin Sergey Khoroshavin
              Reporter:
              sergey.khoroshavin Sergey Khoroshavin
              Watchers:
              Alexander Shcherbakov, Sergey Khoroshavin
            • Votes:
              0 Vote for this issue
              Watchers:
              2 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: