batch_create cost a lot of overhead in processing

asked 2018-12-28 00:29:10 -0500

Bruce Han gravatar image

We intend to use batch_create to speed up the heat stack creation procedure. However we see from the resource list that resources in “previous” resource group instance seem to have been processed a second time when taking the next batch, since status is UPDATE_COMPLETE. And we can see the CPU load on Openstack controller is not reduced because of the batch_create. It seems batch_create cost a lot of overhead in processing.

The resource group definition as below:

Sub_ports:
  properties:
    count:
      get_param: sub_port_count
    resource_def:
      properties:
        stack_name:
          get_param: stack_name
      type: subport.yaml
  update_policy:
    batch_create: {max_batch_size: 50, pause_time: 3}
  type: OS::Heat::ResourceGroup

Could anyone please give me some advice about how to check the problem and how to tune openstack?

edit retag flag offensive close merge delete