When we increase the semaphore, we have already scheduled the strand for
now.
If the label is just waiting for the semaphore to increase, there's no
need for short naps.
Most of our wait labels are just waiting for the semaphore to increase,
so I extended their naps to 6 hours.
It will help decrease the load on the respirate on production
[^1]: 28dacb968b/model/semaphore.rb (L10)
139 lines
4.5 KiB
Ruby
139 lines
4.5 KiB
Ruby
# frozen_string_literal: true
|
|
|
|
class Prog::Kubernetes::KubernetesClusterNexus < Prog::Base
|
|
subject_is :kubernetes_cluster
|
|
|
|
def self.assemble(name:, project_id:, location_id:, version: "v1.32", private_subnet_id: nil, cp_node_count: 3, target_node_size: "standard-2", target_node_storage_size_gib: nil)
|
|
DB.transaction do
|
|
unless (project = Project[project_id])
|
|
fail "No existing project"
|
|
end
|
|
|
|
unless ["v1.32", "v1.31"].include?(version)
|
|
fail "Invalid Kubernetes Version"
|
|
end
|
|
|
|
Validation.validate_kubernetes_name(name)
|
|
Validation.validate_kubernetes_cp_node_count(cp_node_count)
|
|
|
|
subnet = if private_subnet_id
|
|
project.private_subnets_dataset.first(id: private_subnet_id) || fail("Given subnet is not available in the given project")
|
|
else
|
|
subnet_name = name + "-k8s-subnet"
|
|
ps = project.private_subnets_dataset.first(:location_id => location_id, Sequel[:private_subnet][:name] => subnet_name)
|
|
ps || Prog::Vnet::SubnetNexus.assemble(
|
|
project_id,
|
|
name: subnet_name,
|
|
location_id:,
|
|
ipv4_range: Prog::Vnet::SubnetNexus.random_private_ipv4(Location[location_id], project, 18).to_s
|
|
).subject
|
|
end
|
|
|
|
# TODO: Validate location
|
|
# TODO: Move resources (vms, subnet, LB, etc.) into our own project
|
|
# TODO: Validate node count
|
|
|
|
kc = KubernetesCluster.create_with_id(name:, version:, cp_node_count:, location_id:, target_node_size:, target_node_storage_size_gib:, project_id: project.id, private_subnet_id: subnet.id)
|
|
|
|
Strand.create(prog: "Kubernetes::KubernetesClusterNexus", label: "start") { _1.id = kc.id }
|
|
end
|
|
end
|
|
|
|
def before_run
|
|
when_destroy_set? do
|
|
if strand.label != "destroy"
|
|
kubernetes_cluster.active_billing_records.each(&:finalize)
|
|
hop_destroy
|
|
end
|
|
end
|
|
end
|
|
|
|
label def start
|
|
register_deadline("wait", 120 * 60)
|
|
hop_create_load_balancer
|
|
end
|
|
|
|
label def create_load_balancer
|
|
custom_hostname_dns_zone_id = DnsZone[name: Config.kubernetes_service_hostname]&.id
|
|
custom_hostname_prefix = if custom_hostname_dns_zone_id
|
|
"#{kubernetes_cluster.name}-apiserver-#{kubernetes_cluster.ubid.to_s[-5...]}"
|
|
end
|
|
load_balancer_st = Prog::Vnet::LoadBalancerNexus.assemble(
|
|
kubernetes_cluster.private_subnet_id,
|
|
name: "#{kubernetes_cluster.name}-apiserver",
|
|
algorithm: "hash_based",
|
|
src_port: 443,
|
|
dst_port: 6443,
|
|
health_check_endpoint: "/healthz",
|
|
health_check_protocol: "tcp",
|
|
custom_hostname_dns_zone_id:,
|
|
custom_hostname_prefix:,
|
|
stack: LoadBalancer::Stack::IPV4
|
|
)
|
|
kubernetes_cluster.update(api_server_lb_id: load_balancer_st.id)
|
|
|
|
hop_bootstrap_control_plane_vms
|
|
end
|
|
|
|
label def bootstrap_control_plane_vms
|
|
nap 5 unless kubernetes_cluster.endpoint
|
|
|
|
hop_wait_nodes if kubernetes_cluster.cp_vms.count >= kubernetes_cluster.cp_node_count
|
|
|
|
bud Prog::Kubernetes::ProvisionKubernetesNode, {"subject_id" => kubernetes_cluster.id}
|
|
|
|
hop_wait_control_plane_node
|
|
end
|
|
|
|
label def wait_control_plane_node
|
|
reap
|
|
hop_bootstrap_control_plane_vms if leaf?
|
|
donate
|
|
end
|
|
|
|
label def wait_nodes
|
|
nap 10 unless kubernetes_cluster.nodepools.all? { _1.strand.label == "wait" }
|
|
hop_create_billing_records
|
|
end
|
|
|
|
label def create_billing_records
|
|
records =
|
|
kubernetes_cluster.cp_vms.map { {type: "KubernetesControlPlaneVCpu", family: _1.family, amount: _1.vcpus} } +
|
|
kubernetes_cluster.nodepools.flat_map(&:vms).flat_map {
|
|
[
|
|
{type: "KubernetesWorkerVCpu", family: _1.family, amount: _1.vcpus},
|
|
{type: "KubernetesWorkerStorage", family: "standard", amount: _1.storage_size_gib}
|
|
]
|
|
}
|
|
|
|
records.each do |record|
|
|
BillingRecord.create(
|
|
project_id: kubernetes_cluster.project_id,
|
|
resource_id: kubernetes_cluster.id,
|
|
resource_name: kubernetes_cluster.name,
|
|
billing_rate_id: BillingRate.from_resource_properties(record[:type], record[:family], kubernetes_cluster.location.name)["id"],
|
|
amount: record[:amount]
|
|
)
|
|
end
|
|
|
|
hop_wait
|
|
end
|
|
|
|
label def wait
|
|
nap 6 * 60 * 60
|
|
end
|
|
|
|
label def destroy
|
|
reap
|
|
donate unless leaf?
|
|
decr_destroy
|
|
kubernetes_cluster.api_server_lb.incr_destroy
|
|
kubernetes_cluster.cp_vms.each(&:incr_destroy)
|
|
kubernetes_cluster.remove_all_cp_vms
|
|
kubernetes_cluster.nodepools.each { _1.incr_destroy }
|
|
nap 5 unless kubernetes_cluster.nodepools.empty?
|
|
kubernetes_cluster.destroy
|
|
pop "kubernetes cluster is deleted"
|
|
end
|
|
end
|