Fault Tolerance

• Fault Tolerance

  • We have implemented an HPA Horizontal pod autoscaler, this will help us achieve fault tolerance.

  • We are creating replicas if the memory util hits 50% of the allocated memory limit.

  • Min replica is 1 and max replica is 5.

  • For tracking CPU, memory utilization we are using krew extension with resource capacity as an extension.

  • kubectl resource-capacity --sort cpu.util --namespace=default --util --pods

• HPA simulation

  • Currently, default replicas of user service are 3, but the HPA range is 1 to 5. So it downsized to save resources to 1 instance.
  • In the first image, we see that the CPU limit is 500m, and utilization is 16m.
  • We have set the replica to create if it hits 50% CPU utilization which is 250M.