Skip to main content

dtcloud_lb_pool (Resource)

Manages the back-end pool attached to a listener, and how it spreads connections across its members.

Example Usage​

resource "dtcloud_lb_pool" "web" {
  lb_id       = dtcloud_lb.public.id
  listener_id = dtcloud_lb_listener.https.listener_id

  backend_protocol      = "HTTP"
  backend_protocol_port = 8080
  lb_algorithm          = "ROUND_ROBIN"

  # Pin a client to the member it first reached.
  sticky_session = true
}

Schema​

Required​

  • backend_protocol (String) Protocol used to reach the members. One of: HTTP, HTTPS, TCP, UDP. Changing this forces a new resource to be created.
  • backend_protocol_port (Number) Port used to reach the members. Changing this forces a new resource to be created.
  • lb_algorithm (String) Balancing algorithm. One of: LEAST_CONNECTIONS, ROUND_ROBIN, SOURCE_IP. Can be changed in place.
  • lb_id (String) ID of the load balancer this pool belongs to. Changing this forces a new resource to be created.
  • listener_id (String) ID of the listener that feeds this pool. Changing this forces a new resource to be created.

Optional​

  • sticky_session (Boolean) Whether to pin a client to the member it first reached. Can be changed in place.
  • timeouts (Block, Optional) (see below for nested schema)

Read-Only​

  • created_at (String) Creation timestamp as reported by the API.
  • health_monitor_id (String) ID of the health monitor attached to this pool, if any.
  • id (String) The ID of this resource.
  • member_ids (List of String) IDs of the members currently in the pool.
  • name (String) Name assigned by the platform.
  • operating_status (String) Operating status.
  • pool_id (String) ID of the pool on its own, without the load balancer prefix.
  • provisioning_status (String) Provisioning status.
  • updated_at (String) Last-update timestamp as reported by the API.

Nested Schema for timeouts​

Optional:

  • create (String)
  • delete (String)
  • update (String)

Behaviour worth knowing​

  • Only lb_algorithm and sticky_session change in place. The protocol, the port and the owning listener are fixed at creation.
  • sticky_session is write-only. The pool list does not report it, so drift in it is invisible and it does not survive an import.
  • The algorithm comes back as a display name. The API answers Round robin where it was given ROUND_ROBIN; the provider maps it back so an imported pool does not look like it needs replacing.

Timeouts​

create, update and delete each default to 15 minutes.

Import​

# <load-balancer-id>:<pool-id>
terraform import dtcloud_lb_pool.web 3d372446-4c75-4327-9511-98442456162f:5b2a9c31-7e04-4d88-9c6e-44a2f1d0e7b3