Problems caused by Mongo flowcontrol


mongo flowcontrol

  1. When Mongo replica sets are deployed, Mongo writes to the default write priority. If the slave node lags too far behind or there are hung replicas, and the number of replicas is less than half of the replica set data, Mongo will actively appear the flow control.Problems caused by Mongo flowcontrol
  2. View the flowcontrol configuration of the cluster
    db.admin.runCommand(getParameters: “*”); // Get all configuration parameters of mongodb.mongo status

    "flowControlDecayConstant" : 0.5,
    "flowControlFudgeFactor" : 0.95,
    "flowControlMaxSamples" : 1000000,
    "Flowcontrolminticketspersecond": 100, // the parameters that ultimately affect the speed, with a minimum of 100 ticklets per second
    "flowControlSamplePeriod" : 1000,
    "flowControlTargetLagSeconds" : 10,
    "flowControlThresholdLagPercentage" : 0.5,
    "flowControlTicketAdderConstant" : 1000,
    "flowControlTicketMultiplierConstant" : 1.05,
    "flowControlWarnThresholdSeconds" : 10,
  3. Slow log form of flowcontrol
    Problems caused by Mongo flowcontrol
  4. Mongo arbitration node deployment vulnerability
    In order to save server resources online, the master-slave arbitration method is adopted to deploy the replica set. Because the slave node memory of Mongo is too large, it is killed by the kernel, resulting in only the master node and arbitration node left in the cluster. Because the arbitration node only participates in voting. There is only one node (primary node) for cluster writes. Less than majority (2). This causes flowcontrol to occur in the whole cluster. It seriously affects the performance of Mongo service.