Terraform GCP HTTPS load balancer google_compute_url_map with variable length of host_rule blocks

I'm a Terraform novice creating GCP external HTTP/HTTPS load balancers that route different hostnames to different backends. In a working proof of concept, I have two hostnames supported, like this:

locals { hostnames = [ "", "" ]
}
resource "google_compute_url_map" "MY_URL_MAP_HTTPS" { name = "mylb-https" description = "External HTTPS load balancer for multiple hostnames" host_rule { hosts = [ local.hostnames[0] ] path_matcher = "site1-path-matcher" } host_rule { hosts = [ local.hostnames[1] ] path_matcher = "site2-path-matcher" } path_matcher { name = "site1-path-matcher" default_service = "projects/myproject/global/backendServices/site1-backend-service" } path_matcher { name = "site2-path-matcher" default_service = "projects/myproject/global/backendServices/site2-backend-service" }
}

I know I can do a for loop when Terraform expects a list, such as associating multiple SSL certificates with a target proxy:

resource "google_compute_ssl_certificate" "MY_SSL_CERT" { count = length(local.hostnames) name = replace(local.hostnames[count.index], ".", "-")
}
resource "google_compute_target_https_proxy" "MY_TARGET_PROXY_HTTPS" { ssl_certificates = [ for cert in google_compute_ssl_certificate.MY_SSL_CERT cert.id ]
}

But how can a do a for loop on the host_rule and path_matcher blocks? Is this possible?

1 Answer

Found a bug report which helped me see dynamic blocks are the solution:

variable "hostnames" { type = list(string) default = null
}
resource "google_compute_url_map" "URL_MAP_HTTPS" { name = "urlmap-https" default_service = google_compute_backend_bucket.default.id dynamic "host_rule" { for_each = var.hostnames content { hosts = [ host_rule.value] path_matcher = "path-matcher-${host_rule.key}" } } dynamic "path_matcher" { for_each = var.hostnames content { name = "path-matcher-${path_matcher.key}" default_service = google_compute_backend_service.default[path_matcher.key].id } }
}

This is also the solution to having a variable number of backend blocks to use with google_compute_backend_service and google_compute_region_backend_service.