def get(self) -> Munch:
asset = self.BASE_ASSET.copy()
self._verify_asset_fields(asset)
with utils.file_lock_context(self._lock_file):
assets_in_use = self._get_assets_in_use_from_assets_file()
self._fill_allocated_ips_and_bridges_from_assets_file(
assets_in_use)
self._fill_allocated_ips_and_bridges_by_interface()
self._fill_virsh_allocated_ips_and_bridges()
self._override_ip_networks_values_if_not_free(asset)
self._override_network_bridges_values_if_not_free(asset)
self._taken_assets.add(str(asset))
assets_in_use.append(asset)
self._dump_all_assets_in_use_to_assets_file(assets_in_use)
self._allocated_bridges.clear()
self._allocated_ips_objects.clear()
log.info("Taken asset: %s", asset)
return Munch.fromDict(asset)
def list_leases(self, network_name):
with utils.file_lock_context():
net = self.libvirt_connection.networkLookupByName(network_name)
leases = net.DHCPLeases(
) # TODO: getting the information from the XML dump until dhcp-leases bug is fixed
hosts = self._get_hosts_from_network(net)
return leases + [
h for h in hosts
if h["ipaddr"] not in [ls["ipaddr"] for ls in leases]
]
def get_network_leases(network_name):
warnings.warn(
"get_network_leases is deprecated. Use LibvirtController.get_network_leases "
"instead.", DeprecationWarning)
with utils.file_lock_context():
net = conn.networkLookupByName(network_name)
leases = net.DHCPLeases(
) # TODO: getting the information from the XML dump until dhcp-leases bug is fixed
hosts = _get_hosts_from_network(net)
return leases + [
h for h in hosts
if h["ipaddr"] not in [ls["ipaddr"] for ls in leases]
]
def clean_virsh_resources(skip_list, resource_filter):
with utils.file_lock_context():
_clean_domains(skip_list, resource_filter)
_clean_pools(skip_list, resource_filter)
_clean_networks(skip_list, resource_filter)
def release_all(self):
with utils.file_lock_context(self._lock_file):
assets_in_use = self._get_assets_in_use_from_assets_file()
self._remove_taken_assets_from_all_assets_in_use(assets_in_use)
self._dump_all_assets_in_use_to_assets_file(assets_in_use)
def start_install_and_wait_for_installed(self):
cluster_name = self.config.day1_cluster_name
# Running twice as a workaround for an issue with terraform not spawning a new node on first apply.
for _ in range(2):
with utils.file_lock_context():
utils.run_command(
f"make _apply_terraform CLUSTER_NAME={cluster_name} PLATFORM={consts.Platforms.BARE_METAL}"
)
time.sleep(5)
num_nodes_to_wait = self.config.day2_workers_count
installed_status = consts.NodesStatus.DAY2_INSTALLED
tfvars = utils.get_tfvars(self.config.tf_folder)
tf_network_name = tfvars["libvirt_network_name"]
config = TerraformConfig()
config.nodes_count = num_nodes_to_wait
libvirt_controller = LibvirtController(config=config,
entity_config=ClusterConfig())
libvirt_controller.wait_till_nodes_are_ready(
network_name=tf_network_name)
# Wait for day2 nodes
waiting.wait(
lambda: self.are_libvirt_nodes_in_cluster_hosts(),
timeout_seconds=consts.NODES_REGISTERED_TIMEOUT,
sleep_seconds=10,
waiting_for="Nodes to be registered in inventory service",
)
self.set_nodes_hostnames_if_needed(tf_network_name)
wait_till_all_hosts_are_in_status(
client=self.api_client,
cluster_id=self.config.cluster_id,
nodes_count=self.config.day2_workers_count,
statuses=[consts.NodesStatus.KNOWN],
interval=30,
)
# Start day2 nodes installation
log.info("Start installing all known nodes in the cluster %s",
self.config.cluster_id)
kubeconfig = utils.get_kubeconfig_path(self.config.day1_cluster_name)
ocp_ready_nodes = self.get_ocp_cluster_ready_nodes_num(kubeconfig)
hosts = self.api_client.get_cluster_hosts(self.config.cluster_id)
[
self.api_client.install_day2_host(self.config.infra_env_id,
host["id"]) for host in hosts
if host["status"] == "known"
]
log.info(
"Waiting until all nodes of cluster %s have been installed (reached added-to-existing-cluster)",
self.config.cluster_id,
)
wait_till_all_hosts_are_in_status(
client=self.api_client,
cluster_id=self.config.cluster_id,
nodes_count=num_nodes_to_wait,
statuses=[installed_status],
interval=30,
)
log.info(
"Waiting until installed nodes has actually been added to the OCP cluster"
)
waiting.wait(
lambda: self.wait_nodes_join_ocp_cluster(
ocp_ready_nodes, self.config.day2_workers_count, kubeconfig),
timeout_seconds=consts.NODES_REGISTERED_TIMEOUT,
sleep_seconds=30,
waiting_for="Day2 nodes to be added to OCP cluster",
expected_exceptions=Exception,
)
log.info("%d worker nodes were successfully added to OCP cluster",
self.config.day2_workers_count)
def day2_nodes_flow(client, terraform_cluster_dir_prefix, tf_folder, cluster,
has_ipv_6, num_worker_nodes, install_cluster_flag,
day2_type_flag, with_static_network_config,
base_cluster_name):
tf_network_name, total_num_nodes = get_network_num_nodes_from_tf(tf_folder)
# Running twice as a workaround for an issue with terraform not spawning a new node on first apply.
for _ in range(2):
with utils.file_lock_context():
utils.run_command(
f'make _apply_terraform CLUSTER_NAME={terraform_cluster_dir_prefix} PLATFORM={consts.Platforms.BARE_METAL}'
)
time.sleep(5)
if day2_type_flag == "ocp":
num_nodes_to_wait = total_num_nodes
installed_status = consts.NodesStatus.INSTALLED
else:
num_nodes_to_wait = num_worker_nodes
installed_status = consts.NodesStatus.DAY2_INSTALLED
wait_till_nodes_are_ready(nodes_count=num_nodes_to_wait,
network_name=tf_network_name)
waiting.wait(
lambda: are_libvirt_nodes_in_cluster_hosts(client, cluster.id,
num_nodes_to_wait),
timeout_seconds=consts.NODES_REGISTERED_TIMEOUT,
sleep_seconds=10,
waiting_for="Nodes to be registered in inventory service",
)
set_nodes_hostnames_if_needed(client, tf_folder,
with_static_network_config, has_ipv_6,
tf_network_name, cluster.id)
wait_till_all_hosts_are_in_status(
client=client,
cluster_id=cluster.id,
nodes_count=num_worker_nodes,
statuses=[consts.NodesStatus.KNOWN],
interval=30,
)
if install_cluster_flag:
log.info("Start installing all known nodes in the cluster %s",
cluster.id)
kubeconfig = utils.get_kubeconfig_path(base_cluster_name)
ocp_orig_ready_nodes = get_ocp_cluster_ready_nodes_num(kubeconfig)
hosts = client.get_cluster_hosts(cluster.id)
[
client.install_day2_host(cluster.id, host['id']) for host in hosts
if host["status"] == 'known'
]
log.info(
"Start waiting until all nodes of cluster %s have been installed( reached added-to-existing-clustertate)",
cluster.id)
wait_till_all_hosts_are_in_status(
client=client,
cluster_id=cluster.id,
nodes_count=num_nodes_to_wait,
statuses=[installed_status],
interval=30,
)
log.info(
"Start waiting until installed nodes has actually been added to the OCP cluster"
)
waiting.wait(lambda: wait_nodes_join_ocp_cluster(
ocp_orig_ready_nodes, num_worker_nodes, day2_type_flag, kubeconfig
),
timeout_seconds=consts.NODES_REGISTERED_TIMEOUT,
sleep_seconds=30,
waiting_for="Day2 nodes to be added to OCP cluster",
expected_exceptions=Exception)
log.info("%d worker nodes were successfully added to OCP cluster",
num_worker_nodes)
def create_nodes(
tf
):
log.info('Start running terraform')
with utils.file_lock_context():
return tf.apply()