def __init__(self, service_name=None):
self.dfa_cfg = {}
self._load_default_opts()
args = sys.argv[1:]
try:
opts = [(args[i], args[i + 1]) for i in range(0, len(args), 2)]
except IndexError:
opts = []
cfgfile = cfg.find_config_files(service_name)
for k, v in opts:
if k == '--config-file':
cfgfile.append(v)
multi_parser = cfg.MultiConfigParser()
read_ok = multi_parser.read(cfgfile)
if len(read_ok) != len(cfgfile):
raise cfg.Error(_LE("Failed to parse config file."))
for parsed_file in multi_parser.parsed:
for parsed_item in parsed_file.keys():
if parsed_item not in self.dfa_cfg:
self.dfa_cfg[parsed_item] = {}
for key, value in parsed_file[parsed_item].items():
self.dfa_cfg[parsed_item][key] = value[0]
# Convert it to object.
self._cfg = utils.Dict2Obj(self.dfa_cfg)
def test_topo_disc_cb(self):
"""Test the topology discovery CB function. """
topo_dict = self._get_topo_disc_arg()
topo_dict_final = self._get_topo_dict_final()
topo_obj = common_utils.Dict2Obj(topo_dict)
with mock.patch('time.ctime') as ctime_fn:
ctime_fn.return_value = 'Jan 1'
self.dfa_vdp_mgr.topo_disc_cb(self.uplink, topo_obj)
self.rpc_client.make_msg.assert_called_with(
'save_topo_disc_params', {}, msg=jsonutils.dumps(topo_dict_final))
def failure_recovery(self, fail_info):
"""Failure recovery task.
In case of failure in projects, network and VM create/delete, this
task goes through all failure cases and try the request.
"""
# Read failed entries from project database and send request
# (create/delete - depends on failure type) to DCNM
# 1. Try failure recovery for create project.
LOG.info(_LI("Started failure_recovery."))
projs = self.get_fialed_projects_entries(constants.CREATE_FAIL)
for proj in projs:
LOG.debug("Failure recovery for project %(name)s.",
({
'name': proj.name
}))
# Try to create the project in DCNM
try:
self.dcnm_client.create_project(
proj.name, self.cfg.dcnm.default_partition_name,
proj.dci_id)
except dexc.DfaClientRequestFailed as e:
LOG.error(
_LE("failure_recovery: Failed to create %(proj)s "
"on DCNM : %(reason)s"), {
'proj': proj.name,
'reason': str(e)
})
else:
# Request is sent successfully, update the database.
self.update_project_info_cache(proj.id,
dci_id=proj.dci_id,
name=proj.name,
opcode='update')
LOG.debug(
'Success on failure recovery for '
'project %(name)s', {'name': proj.name})
# 1.1 Try failure recovery for update project.
projs = self.get_fialed_projects_entries(constants.UPDATE_FAIL)
for proj in projs:
LOG.debug("Failure recovery for project %(name)s.",
({
'name': proj.name
}))
# This was failure of updating DCI id of the project in DCNM.
try:
self.dcnm_client.update_project(
proj.name, self.cfg.dcnm.default_partition_name,
proj.dci_id)
except dexc.DfaClientRequestFailed as exc:
LOG.error(
_LE("failure_recovery: Failed to update %(proj)s "
"on DCNM : %(reason)s"), {
'proj': proj.name,
'reason': str(exc)
})
else:
# Request is sent successfully, update the database.
self.update_project_info_cache(proj.id,
dci_id=proj.dci_id,
name=proj.name,
opcode='update')
LOG.debug(
'Success on failure recovery update for '
'project %(name)s', {'name': proj.name})
# 2. Try failure recovery for create network.
nets = self.get_all_networks()
for net in nets:
if (net.result == constants.CREATE_FAIL
and net.source.lower() == 'openstack'):
net_id = net.network_id
try:
subnets = self.neutron_event.nclient.list_subnets(
network_id=net_id).get('subnets')
except dexc.ConnectionFailed:
LOG.exception(_LE('Failed to get subnets list.'))
continue
for subnet in subnets:
tenant_name = self.get_project_name(subnet['tenant_id'])
snet = utils.Dict2Obj(subnet)
try:
# Check if config_profile is not NULL.
if not net.config_profile:
cfgp, fwd_mod = (self.dcnm_client.
get_config_profile_for_network(
net.name))
net.config_profile = cfgp
net.fwd_mod = fwd_mod
self.dcnm_client.create_network(
tenant_name, net, snet, self.dcnm_dhcp)
except dexc.DfaClientRequestFailed:
# Still is failure, only log the error.
LOG.error(_LE('Failed to create network %(net)s.'),
{'net': net.name})
else:
# Request is sent to DCNM, update the database
params = dict(
columns=dict(config_profile=net.config_profile,
fwd_mod=net.fwd_mod,
result=constants.RESULT_SUCCESS))
self.update_network(net_id, **params)
LOG.debug(
"Success on failure recovery to create "
"%(net)s", {'net': net.name})
# 3. Try Failure recovery for VM create and delete.
instances = self.get_vms()
for vm in instances:
vm_info = dict(status=vm.status,
vm_mac=vm.mac,
segmentation_id=vm.segmentation_id,
host=vm.host,
port_uuid=vm.port_id,
net_uuid=vm.network_id,
oui=dict(ip_addr=vm.ip,
vm_name=vm.name,
vm_uuid=vm.instance_id,
gw_mac=vm.gw_mac,
fwd_mod=vm.fwd_mod,
oui_id='cisco'))
if vm.result == constants.CREATE_FAIL:
try:
self.neutron_event.send_vm_info(str(vm.host), str(vm_info))
except Exception as e:
# Failed to send info to the agent. Keep the data in the
# database as failure to send it later.
LOG.error(
_LE('Failed to send VM info to agent. '
'Reason %s'), str(e))
else:
params = dict(columns=dict(
result=constants.RESULT_SUCCESS))
self.update_vm_db(vm.port_id, **params)
LOG.info(_LI('Created VM %(vm)s.'), {'vm': vm.name})
for vm in instances:
if vm.result == constants.DELETE_FAIL:
vm_info['status'] = 'down'
try:
self.neutron_event.send_vm_info(str(vm.host), str(vm_info))
except Exception as e:
LOG.error(
_LE('Failed to send VM info to agent. '
'Reason %s'), str(e))
else:
self.delete_vm_db(vm.instance_id)
LOG.info(_LI('Deleted VM %(vm)s from DB.'),
{'vm': vm.name})
# 4. Try failure recovery for delete network.
for net in nets:
if (net.result == constants.DELETE_FAIL
and net.source.lower() == 'openstack'):
net_id = net.network_id
segid = net.segmentation_id
tenant_name = self.get_project_name(net.tenant_id)
try:
self.dcnm_client.delete_network(tenant_name, net)
except dexc.DfaClientRequestFailed:
# Still is failure, only log the error.
LOG.error(_LE('Failed to delete network %(net)s.'),
{'net': net.name})
else:
# Request is sent to DCNM, delete the entry
# from database and return the segmentation id to the
# pool.
self.delete_network_db(net_id)
self.segmentation_pool.add(segid)
LOG.debug(
"Success on failure recovery to deleted "
"%(net)s", {'net': net.name})
# 5. Try failure recovery for delete project.
projs = self.get_fialed_projects_entries(constants.DELETE_FAIL)
for proj in projs:
LOG.debug("Failure recovery for project %(name)s.",
({
'name': proj.name
}))
# Try to delete the project in DCNM
try:
self.dcnm_client.delete_project(
proj.name, self.cfg.dcnm.default_partition_name)
except dexc.DfaClientRequestFailed as e:
# Failed to delete project in DCNM.
# Save the info and mark it as failure and retry it later.
LOG.error(
_LE("Failure recovery is failed to delete "
"%(project)s on DCNM : %(reason)s"), {
'project': proj.name,
'reason': str(e)
})
else:
# Delete was successful, now update the database.
self.update_project_info_cache(proj.id, opcode='delete')
LOG.debug(
"Success on failure recovery to deleted "
"%(project)s", {'project': proj.name})
# 6. DHCP port consistency check for HA.
if self.need_dhcp_check():
nets = self.get_all_networks()
for net in nets:
net_id = net.network_id
LOG.debug("dhcp consistency check for net id %s", net_id)
self.correct_dhcp_ports(net_id)
self.decrement_dhcp_check()
LOG.info(_LI("Finished failure_recovery."))