def rebuild(self, context, server):
"""Rebuild/redeploy a server.
:param context: The security context.
:param server: The server object.
"""
LOG.debug('Rebuild called for server', server=server)
node_uuid = server.node_uuid
node = self._get_node(node_uuid)
self._add_server_info_to_node(node, server)
# trigger the node rebuild
try:
self.ironicclient.call("node.set_provision_state", node_uuid,
ironic_states.REBUILD)
except (ironic_exc.InternalServerError, ironic_exc.BadRequest) as e:
msg = (_("Failed to request Ironic to rebuild server "
"%(server)s: %(reason)s") % {
'server': server.uuid,
'reason': six.text_type(e)
})
raise exception.ServerDeployFailure(msg)
# Although the target provision state is REBUILD, it will actually go
# to ACTIVE once the redeploy is finished.
timer = loopingcall.FixedIntervalLoopingCall(self._wait_for_active,
server)
timer.start(interval=CONF.ironic.api_retry_interval).wait()
LOG.info('Server was successfully rebuilt', server=server)
def _wait_for_active(self, server):
"""Wait for the node to be marked as ACTIVE in Ironic."""
server.refresh()
if server.status in (states.DELETING, states.ERROR, states.DELETED):
raise exception.ServerDeployAborted(
_("Server %s provisioning was aborted") % server.uuid)
node = self._validate_server_and_node(server)
if node.provision_state == ironic_states.ACTIVE:
# job is done
LOG.debug("Ironic node %(node)s is now ACTIVE",
dict(node=node.uuid),
server=server)
raise loopingcall.LoopingCallDone()
if node.target_provision_state in (ironic_states.DELETED,
ironic_states.AVAILABLE):
# ironic is trying to delete it now
raise exception.ServerNotFound(server=server.uuid)
if node.provision_state in (ironic_states.NOSTATE,
ironic_states.AVAILABLE):
# ironic already deleted it
raise exception.ServerNotFound(server=server.uuid)
if node.provision_state == ironic_states.DEPLOYFAIL:
# ironic failed to deploy
msg = (_("Failed to provision server %(server)s: %(reason)s") % {
'server': server.uuid,
'reason': node.last_error
})
raise exception.ServerDeployFailure(msg)
_log_ironic_polling('become ACTIVE', node, server)
def _wait_for_servers_status(cls,
server_id,
wait_interval,
wait_timeout,
status=None,
power_state=None,
locked=None):
"""Waits for a Server to reach the given status, power_state,
lock state.
"""
server_status = None
server_power_state = None
server_locked = None
start = int(time.time())
def _condition():
compare_pairs = ((status, server_status), (power_state,
server_power_state),
(locked, server_locked))
return all([r == a for r, a in compare_pairs if r is not None])
while not _condition():
time.sleep(wait_interval)
try:
body = cls.baremetal_compute_client.show_server(server_id)
server_status = body['status']
server_power_state = body['power_state']
server_locked = body['locked']
except lib_exc.NotFound:
if status == 'deleted':
break
else:
raise
if server_status == 'error' and status != 'error':
msg = ('Failed to provision server %s' % server_id)
raise exception.ServerDeployFailure(msg)
if int(time.time()) - start >= wait_timeout:
message = ('Server %s failed to reach %s status '
'(current %s) within the required time (%s s).' %
(server_id, status, server_status, wait_timeout))
raise lib_exc.TimeoutException(message)
def manage(self, server, node_uuid):
"""Manage an existing bare metal node.
:param server: The bare metal server object.
:param node_uuid: The manageable bare metal node uuid.
"""
# Associate the node with a server
patch = [{'path': '/instance_uuid', 'op': 'add', 'value': server.uuid}]
try:
self.ironicclient.call('node.update',
node_uuid,
patch,
retry_on_conflict=False)
except ironic_exc.BadRequest:
msg = (_("Failed to update parameters on node %(node)s "
"when provisioning the server %(server)s") % {
'node': node_uuid,
'server': server.uuid
})
LOG.error(msg)
raise exception.ServerDeployFailure(msg)
def _add_server_info_to_node(self, node, server):
patch = list()
# Associate the node with a server
patch.append({
'path': '/instance_uuid',
'op': 'add',
'value': server.uuid
})
# Add the required fields to deploy a node.
patch.append({
'path': '/instance_info/image_source',
'op': 'add',
'value': server.image_uuid
})
# TODO(zhenguo) Add partition support
patch.append({
'path': '/instance_info/root_gb',
'op': 'add',
'value': str(node.properties.get('local_gb', 0))
})
try:
# FIXME(lucasagomes): The "retry_on_conflict" parameter was added
# to basically causes the deployment to fail faster in case the
# node picked by the scheduler is already associated with another
# server due bug #1341420.
self.ironicclient.call('node.update',
node.uuid,
patch,
retry_on_conflict=False)
except ironic_exc.BadRequest:
msg = (_("Failed to add deploy parameters on node %(node)s "
"when provisioning the server %(server)s") % {
'node': node.uuid,
'server': server.uuid
})
LOG.error(msg)
raise exception.ServerDeployFailure(msg)
def _add_server_info_to_node(self,
node,
server,
preserve_ephemeral=None,
partitions=None):
patch = list()
# Associate the node with a server
patch.append({
'path': '/instance_uuid',
'op': 'add',
'value': server.uuid
})
# Add the required fields to deploy a node.
patch.append({
'path': '/instance_info/image_source',
'op': 'add',
'value': server.image_uuid
})
root_gb = node.properties.get('local_gb', 0)
if preserve_ephemeral is not None:
patch.append({
'path': '/instance_info/preserve_ephemeral',
'op': 'add',
'value': str(preserve_ephemeral)
})
if partitions:
patch.append({
'path': '/instance_info/ephemeral_gb',
'op': 'add',
'value': str(partitions.get('ephemeral_gb', 0))
})
patch.append({
'path': '/instance_info/swap_mb',
'op': 'add',
'value': str(partitions.get('swap_mb', 0))
})
# Local boot support with partition images, **must** contain
# ``grub2`` installed within it
patch.append({
'path': '/instance_info/capabilities',
'op': 'add',
'value': '{"boot_option": "local"}'
})
# If partitions is not None, use the root_gb in partitions instead
root_gb = partitions.get('root_gb', root_gb)
# root_gb is required not optional
patch.append({
'path': '/instance_info/root_gb',
'op': 'add',
'value': str(root_gb)
})
try:
# FIXME(lucasagomes): The "retry_on_conflict" parameter was added
# to basically causes the deployment to fail faster in case the
# node picked by the scheduler is already associated with another
# server due bug #1341420.
self.ironicclient.call('node.update',
node.uuid,
patch,
retry_on_conflict=False)
except ironic_exc.BadRequest:
msg = (_("Failed to add deploy parameters on node %(node)s "
"when provisioning the server %(server)s") % {
'node': node.uuid,
'server': server.uuid
})
LOG.error(msg)
raise exception.ServerDeployFailure(msg)