rewrite LB for full implementation

[likexian]

Currently LB is not useable.

[lubars]

Trying to make use of new resources available as of provider.tencentcloud v1.15.0 and encountered some issues:

1. Error: [TencentCloudSDKError] Code=InvalidParameterValue, Message=Scheduler should be WRR or LEAST_CONN, RequestId=26db209f-3997-47b8-8b0b-e1ea002374af

According to the documentation for tencentcloud_clb_listener, this parameter is optional and the default is 'WRR':

scheduler - (Optional) Scheduling method of the CLB listener, and available values include 'WRR', 'IP HASH' and 'LEAST_CONN'. The defaule is 'WRR'.

2. Having difficulty coming up with a valid configuration for tencentcloud_clb_listener and tencentcloud_clb_listener rule, e.g:

Error: [TencentCloudSDKError] Code=InvalidParameterValue, Message=This interface only support HTTP/HTTPS listener., RequestId=6bd1e9a2-cb76-4817-b89d-10f689d56617
Error: health_check_http_path can only be set with protocol TCP
Error: health para can only be set with TCP/UDP listener or rule of HTTP/HTTPS listener

3. Not clear what to provide for 'domain' and 'url' of tencentcloud_clb_listener_rule; semantics differ from those of other providers.

4. No attributes exported from tencentcloud_clb_instance, so not possible to determine its IP address.

I suspect these are largely documentation issues.

[likexian]

Hello @lubars
Sorry for the trouble.

[likexian]

1. Please don't define Scheduler in tf file, then it will use the default value WRR.

[likexian]

2. tencentcloud_clb_instance support TCP/UDP or HTTP/HTTPS mode, the arguments can not mixed, please check documents for detail.

Error: health_check_http_path can only be set with protocol TCP, means health_check_http_path only work with TCP mode. what document said:

health_check_http_path - (Optional) Path of health check. NOTES: Only supports listeners of 'HTTPS'/'HTTP' protocol.

refer: https://www.terraform.io/docs/providers/tencentcloud/r/clb_listener_rule.html#health_check_http_path

[likexian]

3. Yes, there is something different with other providers.

resource "tencentcloud_clb_listener_rule" "foo" {
  listener_id                = "lbl-hh141sn9"
  clb_id                     = "lb-k2zjp9lv"
  domain                     = "foo.net"
  url                        = "/bar"
}

be equal to AWS

resource "aws_lb_listener_rule" "static" {
  listener_arn = "${aws_lb_listener.front_end.arn}"

  condition {
    field  = "path-pattern"
    values = ["/bar*"]
  }

  condition {
    field  = "host-header"
    values = ["foo.net"]
  }
}

[likexian]

4. Sorry, we will export it next version.
   Attributes clb_vips, a list of ip string.

[likexian]

The example may help to configure.

https://github.com/terraform-providers/terraform-provider-tencentcloud/tree/master/examples/tencentcloud-clb

[likexian]

Hello @lubars
We have release v1.15.1, for 1, 2, 3 would you please upgrade and have a new try?
for 4th, please use attributes clb_vips, for example: tencentcloud_clb_instance.foo.clb_vips (there is something wrong with documents, we will fix it assp).

[lubars]

Hi @likexian,

I am having success with clb_vips, thank you.

By the way, docs for tencentcloud_clb_attachment indicate that the default weight is 10:

weight - (Optional) Forwarding weight of the backend service, the range of [0, 100], defaults to 10.

However looking at the console, it appears that the default weight is '0'.

[lubars]

Also, have you given any thought to making the target of tencentcloud_clb_attachment into a list of tencentcloud_instance ids, rather than a hard-coded list of targets? The most compact way I was able to add every tencentcloud_instance to the backend pool:

resource "tencentcloud_clb_attachment" "default" {
  listener_id = tencentcloud_clb_listener.default.id
  clb_id      = tencentcloud_clb_instance.default.id

  dynamic "targets" {
    for_each = [for instance in tencentcloud_instance.default: {
      id = instance.id
    }]

    content {
      instance_id = targets.value.id
      port        = var.ForwardPort
    }
  }
}

But maybe I've missed something simpler.

[likexian]

Hi @likexian,

I am having success with clb_vips, thank you.

By the way, docs for tencentcloud_clb_attachment indicate that the default weight is 10:

weight - (Optional) Forwarding weight of the backend service, the range of [0, 100], defaults to 10.

However looking at the console, it appears that the default weight is '0'.

Thank you @lubars
The default weight of console is '10' too, however you can set to '0' manually.

[likexian]

Also, have you given any thought to making the target of tencentcloud_clb_attachment into a list of tencentcloud_instance ids, rather than a hard-coded list of targets? The most compact way I was able to add every tencentcloud_instance to the backend pool:

resource "tencentcloud_clb_attachment" "default" {
  listener_id = tencentcloud_clb_listener.default.id
  clb_id      = tencentcloud_clb_instance.default.id

  dynamic "targets" {
    for_each = [for instance in tencentcloud_instance.default: {
      id = instance.id
    }]

    content {
      instance_id = targets.value.id
      port        = var.ForwardPort
    }
  }
}

But maybe I've missed something simpler.

You are doing the right way, I don't have a simpler too.

However you can make it as follow:

resource "tencentcloud_clb_attachment" "default" {
  listener_id = "${tencentcloud_clb_listener.default.id}"
  clb_id      = "${tencentcloud_clb_instance.default.id}"

  dynamic "targets" {
    for_each = tencentcloud_instance.default

    content {
      instance_id = targets.value.id
      port        = var.ForwardPort
    }
  }
}

[lubars]

The weight comes up in the console as '0' if not specified in the Terraform template, which results in no traffic being directed at that target; as soon as I set it to '10' in the template (or override it in the console), traffic starts flowing. So I think you need to look again at the Terraform default.

[likexian]

Sorry @lubars
There is some misunderstanding, I got what you means now.
There is a problem with the default value, we will fixed it in next version.

[likexian]

Hello @lubars
The default value is fixed now, please try.

[lubars]

@likexian It looks like the behavior now matches the documentation. It would be nice if health_check_http_path could be used with a TCP listener so that compute instances could indicate they are not active, even when they are reachable (for example, during a rolling upgrade). Do you think this makes sense as a feature request, or is this a limitation of TencentCloud itself?

[likexian]

Hello @lubars , thank you for your feedback, health_check_http_path now only works with HTTP, it is the limitation of TencentCloud CLB. Will it possible that you change you TCP listener to HTTP listener?

[likexian]

Hello,
If there is still some problem, please feel free to open a new issue.