Uploaded image for project: 'Core Server'
  1. Core Server
  2. SERVER-18964

Primary in a replset lost primary status during YCSB 50/50 load

XMLWordPrintableJSON

    • Type: Icon: Bug Bug
    • Resolution: Done
    • Priority: Icon: Critical - P2 Critical - P2
    • None
    • Affects Version/s: 3.0.5, 3.1.4
    • Component/s: Replication
    • Labels:
      None
    • Fully Compatible
    • Linux
    • Hide

      Workload: YCSB workload A (50% update, 50% read), 20 million documents, workload set to run 200 million operation but usually hits this condition way before that (~10 million ops, maybe)
      Configuration:
      3-node replset, each node is an Amazon m3.2xlarge (8 vCPU, 30 GB memory)

      Show
      Workload: YCSB workload A (50% update, 50% read), 20 million documents, workload set to run 200 million operation but usually hits this condition way before that (~10 million ops, maybe) Configuration: 3-node replset, each node is an Amazon m3.2xlarge (8 vCPU, 30 GB memory)
    • RPL 7 08/10/15

      While running a YCSB 50/50 workload on EC2 instances, the connections between the primary and the secondaries timed out and this lead to one of the secondaries taking over as the primary.

        1. mongod.log.10.3.1.102
          280 kB
        2. mongod.log.10.3.1.101
          459 kB
        3. mongod.log.10.3.1.100
          67.85 MB
        4. log_appender_flushing_disabled_3.1.4-mongod_logs.tar.gz
          70.51 MB
        5. data_and_log_same_partition_3.0.5-mongod_logs.tar.gz
          18.45 MB
        6. cluster-config_new_logpath_3.1.4-mongod_logs.tar.gz
          11.11 MB

            Assignee:
            benety.goh@mongodb.com Benety Goh
            Reporter:
            chung-yen.chang Chung-yen Chang
            Votes:
            0 Vote for this issue
            Watchers:
            12 Start watching this issue

              Created:
              Updated:
              Resolved: