def create_backup(self):
self.inst_id = self._get_rid()
utils.check_server_status(self.inst_id, type = DBInstanceType.READ_REPLI)
backup = utils.create_backup_byclient(self.inst_id, backup_type='snapshot')
utils.check_backup_status(backup.id)
utils.check_backup_path(backup.id)
def create(self):
inst_id = None
try:
inst_id = utils.get_instance_id_bytenant(DBInstanceType.MASTER)
except Exception as e:
print e
pass
self.inst_id = inst_id
if not self.inst_id:
self.inst_id = utils.create_rds_byclient("HA")
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=240)
self.vip_id = utils.get_vip_id(self.inst_id)
_inst = utils.get_instance(self.inst_id, deleted = False)
self.group_id = _inst.group_id
origin_id = inst_utils.virtual_instid_2_origin_instid(self.inst_id)
assert _inst.id == origin_id, (_inst.id, origin_id)
self.dbslave_id = utils.get_instance_id(self.group_id, DBInstanceType.STANDBY)
LOG.info('master_id:%s slave_id:%s' % (self.inst_id, self.dbslave_id))
self.validate()
def resize(self):
old_rr_inst_id = self._get_rid()
rr_instance = utils.get_builtin_instance(old_rr_inst_id)
old_flavor_id = rr_instance.flavor_id
old_virtual_inst_id = rr_instance.virtual_instance_id
flavor = old_flavor_id + 1 if old_flavor_id < 5 else 5
_ret_inst = utils.resize_flavor_byclient(old_rr_inst_id,flavor)
utils.check_server_status(old_virtual_inst_id, type = DBInstanceType.READ_REPLI,
expected_task = utils.tasks.InstanceTasks.RESIZING)
insts = utils.check_resize_status(self.group_id)
new_rr_id = insts[0]
utils.check_server_status(new_rr_id, type = DBInstanceType.READ_REPLI,
expected_task = utils.tasks.InstanceTasks.NONE,
expected_svr_status = utils.ServiceStatuses.RUNNING,
deleted = False, timeout = 120)
new_rr_instance = utils.get_instance(new_rr_id, deleted = False)
new_flavor_id = new_rr_instance.flavor_id
assert old_flavor_id != new_flavor_id
self._validate(self.master_id, new_rr_id)
def _delete_RRs(self):
rids = self._get_rids()
orig_rid_vips = {}
for rid in rids:
vip_id = utils.get_vip_id(rid, deleted = False)
orig_rid_vips[rid] = vip_id
for rid in rids:
utils.check_server_status(rid,expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.READ_REPLI,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False,timeout=10)
utils.delete_rds_byclient(rid)
utils.check_server_status(rid,expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.READ_REPLI,
expected_svr_status=utils.ServiceStatuses.DELETED,
deleted=True,timeout=CONF.nova_delete_timeout)
vip_id = utils.get_vip_id(rid)
utils.check_vip(rid, vip_id=vip_id, deleted=True)
# Fore. 20150615. Delete One RR Instance will not make change to other RR Instance's VIP-RIP mapping.
orig_rid_vips.pop(rid)
for k, v in orig_rid_vips.items():
utils.check_vip(id = k, vip_id = v, deleted = False)
def migrate(self, instance_id=None):
self.inst_id = instance_id
if not self.inst_id:
self.inst_id = utils.get_instance_id_bytenant(DBInstanceType.SINGLE)
utils.check_server_status(self.inst_id,
expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.SINGLE,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False)
inst_db_info = utils.get_instance(self.inst_id, deleted= False)
old_nova_server_id = inst_db_info.compute_instance_id
ran_count = 45
utils.generate_databases(self.inst_id, count = ran_count)
_ret = rpc.call(utils.get_context(),
"taskmanager",
{"method": "migrate",
"args": {'migrate_id': inst_db_info.id}})
utils.check_server_status(self.inst_id,
expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.SINGLE,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False)
new_nova_server_id = utils.get_instance(self.inst_id, deleted= False).compute_instance_id
assert old_nova_server_id != new_nova_server_id
utils.check_generated_databases(self.inst_id, count = ran_count)
self.validate()
def _lock(self):
self.inst_id = utils.get_instance_id_bytenant(DBInstanceType.MASTER)
utils.lock_byclient(self.inst_id)
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.LOCKED,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=240)
utils.check_instance_is_lock(self.inst_id)
def __prepare_HA_innstance(cls, backup_id):
master_id = test_utils.create_rds_byclient("HA", backup = backup_id)
test_utils.check_server_status(master_id,
expected_task = tasks.InstanceTasks.NONE,
type = DBInstanceType.MASTER,
expected_svr_status= test_utils.ServiceStatuses.RUNNING,
deleted=False, timeout=10)
return master_id
def __prepare_RR_instance(cls, master_id):
rr_inst_id = test_utils.create_rds_byclient("RR", instance_id = master_id)
test_utils.check_server_status(rr_inst_id,
expected_task=tasks.InstanceTasks.NONE,
type=DBInstanceType.READ_REPLI,
expected_svr_status = test_utils.ServiceStatuses.RUNNING,
deleted=False, timeout=300)
return rr_inst_id
def migrate_when_backup_fail(self):
self.inst_id = self._get_rid()
utils.check_server_status(self.inst_id, type = DBInstanceType.READ_REPLI)
backup = utils.create_backup_byclient(self.inst_id)
utils.check_backup_status(backup.id)
bk_info = DBBackup.find_by(utils.get_context(), id = backup.id)
bk_info.state = 'FAILED'
bk_info.save()
self.migrate()
def delete(self):
self.inst_id = utils.get_instance_id_bytenant(DBInstanceType.SINGLE)
_master_vm_id = utils.get_instance(self.inst_id).compute_instance_id
utils.delete_rds_byclient(self.inst_id)
utils.check_vm_is_running(_master_vm_id, expect_status="DELETED", timeout=CONF.nova_delete_timeout)
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
expected_svr_status=utils.ServiceStatuses.DELETED,
deleted=True, timeout=CONF.nova_delete_timeout)
utils.check_vip(self.inst_id, deleted=True)
def upgradeha(self,instance_id=None):
self.inst_id = instance_id
if not self.inst_id:
self.inst_id = utils.get_instance_id_bytenant(DBInstanceType.SINGLE)
self.group_id = utils.get_instance(self.inst_id).group_id
utils.upgrade_ha_byclient(self.inst_id)
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False)
_ha = HAInstance()
_ha.create()
def restore_to_point_in_time(self):
self.create()
old_master_inst = utils.get_builtin_instance(self.inst_id)
master_vid = old_master_inst.virtual_instance_id
db_count = 100
utils.generate_databases(self.inst_id, count = db_count)
utils.check_generated_databases(self.inst_id, count = db_count)
time.sleep(3)
time1 = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
utils.clear_generated_databases(self.inst_id, count = db_count)
time.sleep(3)
dt2 = datetime.datetime.now()
time2 = dt2.strftime('%Y-%m-%d %H:%M:%S')
while True:
restorable_time = utils.get_restorable_time(self.inst_id)
dt = datetime.datetime.strptime(restorable_time.end, '%Y-%m-%d %H:%M:%S')
if dt > dt2:
break
time.sleep(3)
utils.restore_to_point_in_time_byclient(master_vid, time1)
time.sleep(10) # wait for compute instance appear
new_inst1 = utils.get_restore_instance()
utils.check_server_status(new_inst1, DBInstanceType.SINGLE, timeout=RESTORE_TIME_OUT)
utils.check_generated_databases(new_inst1, count = db_count)
utils.delete_rds_byclient(new_inst1)
utils.check_server_status(new_inst1, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.SINGLE,
expected_svr_status=utils.ServiceStatuses.DELETED,
deleted=True, timeout=CONF.trove_delete_timeout)
# delete original instance
utils.delete_rds_byclient(master_vid)
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.DELETED,
deleted=True, timeout=CONF.trove_delete_timeout)
# restore to point in time after deleting
utils.restore_to_point_in_time_byclient(master_vid, time1)
time.sleep(10) # wait for compute instance appear
new_inst2 = utils.get_restore_instance()
utils.check_server_status(new_inst2, DBInstanceType.SINGLE, timeout=RESTORE_TIME_OUT)
utils.check_generated_databases(new_inst2, count = db_count)
utils.delete_rds_byclient(new_inst2)
utils.check_server_status(new_inst2, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.SINGLE,
expected_svr_status=utils.ServiceStatuses.DELETED,
deleted=True, timeout=CONF.trove_delete_timeout)
def _unlock(self):
self.inst_id = utils.get_instance_id_bytenant(DBInstanceType.MASTER)
utils.unlock_byclient(self.inst_id)
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=240)
self.vip_id = utils.get_vip_id(self.inst_id)
_inst = utils.get_builtin_instance(self.inst_id)
self.group_id = _inst.group_id
assert self.inst_id == _inst.virtual_instance_id, (self.inst_id, _inst.virtual_instance_id)
self.dbslave_id = utils.get_instance_id(self.group_id, DBInstanceType.STANDBY)
LOG.info('master_id:%s slave_id:%s' % (self.inst_id, self.dbslave_id))
self.validate()
def _create_with_backup(self, backup, admin_user=None, admin_password=None):
self.inst_id = utils.create_rds_byclient("HA", backup=backup.id, admin_user=admin_user,
admin_password=admin_password)
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=600)
self.vip_id = utils.get_vip_id(self.inst_id)
_inst = utils.get_builtin_instance(self.inst_id)
self.group_id = _inst.group_id
self.dbslave_id = utils.get_instance_id(self.group_id, DBInstanceType.STANDBY)
LOG.info('master_id:%s slave_id:%s' % (self.inst_id, self.dbslave_id))
self.validate()
return self.inst_id
def create_with_master_config(self):
self.master_id = utils.get_instance_id_bytenant(DBInstanceType.MASTER)
utils.check_server_status(self.master_id,expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False,timeout=10)
_inst = utils.get_builtin_instance(self.master_id)
self.group_id = _inst.group_id
if not _inst.db_info.configuration_id:
raise Exception("master configuration_id is None")
self.inst_id = utils.create_rds_byclient("RR", instance_id=self.master_id,
config_id_from_master=_inst.db_info.configuration_id)
self.validate()
utils.check_instancedb_mysql_variables(self.inst_id)
def create_with_config(self):
self.master_id = utils.get_instance_id_bytenant(DBInstanceType.MASTER)
utils.check_server_status(self.master_id,expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False,timeout=10)
_inst = utils.get_builtin_instance(self.master_id)
self.group_id = _inst.group_id
config = utils.get_config_bytenant()
if not config:
raise Exception("not found template config")
self.inst_id = utils.create_rds_byclient("RR",instance_id = self.master_id, config_id = config.id)
self.validate()
utils.check_instancedb_mysql_variables(self.inst_id)
def configuration_patch(self):
self.inst_id = utils.get_instance_id_bytenant(DBInstanceType.MASTER)
if self.inst_id is None:
raise Exception("not found HA instance")
utils.patch_instance_config_and_check(self.inst_id)
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=240)
self.vip_id = utils.get_vip_id(self.inst_id)
_inst = utils.get_builtin_instance(self.inst_id)
self.group_id = _inst.group_id
self.inst_id = utils.get_instance_id(self.group_id, DBInstanceType.MASTER)
self.dbslave_id = utils.get_instance_id(self.group_id, DBInstanceType.STANDBY)
LOG.info('master_id:%s slave_id:%s' % (self.inst_id, self.dbslave_id))
self.validate()
utils.check_instace_config_is_same(self.inst_id)
def _validate(self, master_id, validated_inst_id):
utils.check_server_status(validated_inst_id,expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.READ_REPLI,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False,timeout=300)
utils.check_mysql_adminuser(validated_inst_id)
utils.check_vip(master_id)
vip_id = utils.get_vip_id(validated_inst_id)
utils.check_vip(validated_inst_id,vip_id=vip_id)
utils.check_rpl_delay(validated_inst_id)
master_ip = utils.check_allocate_ip(utils.get_builtin_instance(master_id).server)
rr_ip = utils.check_allocate_ip(utils.get_builtin_instance(validated_inst_id).server)
utils.check_rpl_consist(master_id, [validated_inst_id],master_ip,[rr_ip])
utils.check_rpl_topo_rr(self.group_id)
def migrate(self):
rr_inst_id = self._get_rid()
old_rr_inst = utils.get_builtin_instance(rr_inst_id)
old_rr_id = old_rr_inst.id
rr_vid = old_rr_inst.virtual_instance_id
_ret = rpc.call(utils.get_context(), "taskmanager",
{"method": "migrate",
"args": {'migrate_id': old_rr_inst.id}})
utils.check_server_status(rr_vid, DBInstanceType.READ_REPLI, InstanceTasks.NONE, utils.ServiceStatuses.RUNNING, timeout = 600)
new_rr_inst = utils.get_builtin_instance(rr_vid)
assert new_rr_inst.virtual_instance_id == rr_vid and new_rr_inst.id != old_rr_id, (rr_vid, new_rr_inst.id)
master_id = utils.get_instance_id_bytenant(DBInstanceType.MASTER)
self._validate(master_id, rr_vid)
def validate(self):
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False)
inst_db_info = utils.get_instance(self.inst_id)
self.group_id = inst_db_info.group_id
utils.check_mysql_adminuser(inst_db_info.id)
_ret = utils.get_builtin_instance(inst_db_info.id)
self.nova_instance = _ret.server
# ip = utils.check_allocate_ip(self.nova_instance)
utils.check_mysql_is_running(inst_db_info.id)
utils.check_vip(inst_db_info.id)
self.backup_id = utils.check_backup(self.group_id)
utils.check_backup_status(self.backup_id)
utils.check_backup_path(self.backup_id)
utils.check_rpl_topo_single(self.group_id)
def restart(self, instance_id=None):
self.inst_id = instance_id
if self.inst_id is None:
self.inst_id = utils.get_instance_id_bytenant(DBInstanceType.MASTER)
utils.restart_mysql(self.inst_id)
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.STANDBY,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=240)
self.vip_id = utils.get_vip_id(self.inst_id)
_inst = utils.get_builtin_instance(self.inst_id)
self.group_id = _inst.group_id
assert self.inst_id == _inst.virtual_instance_id, (self.inst_id, _inst.virtual_instance_id)
self.dbslave_id = utils.get_instance_id(self.group_id, DBInstanceType.STANDBY)
LOG.info('master_id:%s slave_id:%s' % (self.inst_id, self.dbslave_id))
utils.check_server_status(self.dbslave_id, expected_task=utils.tasks.InstanceTasks.REBOOTING,
type=DBInstanceType.STANDBY,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=240)
utils.check_server_status(self.dbslave_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.STANDBY,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=240)
self.validate()
def validate(self):
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=600)
utils.check_server_status(self.dbslave_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.STANDBY,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=600)
rr_items = InstanceGroupItem.get_by_gid_type(utils.get_context(), self.group_id, DBInstanceType.READ_REPLI, deleted = False)
slave_ids = []
for rr in rr_items:
slave_ids.append(rr.instance_id)
slave_ids.append(self.dbslave_id)
utils.check_mysql_adminuser(self.inst_id)
utils.check_mysql_adminuser(self.dbslave_id)
for _id in [self.inst_id, self.dbslave_id]:
utils.check_mysql_is_running(self.inst_id)
utils.check_vip(self.inst_id, vip_id=self.vip_id)
self.backup_id = utils.check_backup(self.group_id)
utils.check_backup_status(self.backup_id)
utils.check_backup_path(self.backup_id)
for slave_id in slave_ids:
utils.check_rpl_delay(slave_id)
master_inst = utils.get_builtin_instance(self.inst_id)
slave_inst = utils.get_builtin_instance(self.dbslave_id)
master_ip = utils.check_allocate_ip(master_inst.server)
slave_ip = utils.check_allocate_ip(slave_inst.server)
LOG.info("master_ip:%s slave_ip:%s" % (master_ip, slave_ip))
utils.check_rpl_consist(self.inst_id, slave_ids, master_ip, [slave_ip])
utils.check_rpl_topo_ha(self.group_id)
def _create_RRs(self, rr_count = 2, attach_same_vip = False):
self.master_id = utils.get_instance_id_bytenant(DBInstanceType.MASTER)
utils.check_server_status(self.master_id,expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False,timeout=10)
_inst = utils.get_builtin_instance(self.master_id)
self.group_id = _inst.group_id
rids = self._get_rids()
if len(rids) == 0:
vinst_id1 = utils.create_rds_byclient("RR", instance_id = self.master_id, attached_vip=None)
self._validate(self.master_id, vinst_id1)
origin_inst_id1 = inst_utils.virtual_instid_2_origin_instid(vinst_id1, get_deleted = False)
vip_id = utils.get_vip_id(vinst_id1, deleted = False)
left_to_create = rr_count - 1
while left_to_create > 0:
if attach_same_vip:
utils.create_rds_byclient("RR", instance_id = self.master_id, attached_vip=vip_id)
else:
utils.create_rds_byclient("RR", instance_id = self.master_id, attached_vip=None)
left_to_create -= 1
rids = self._get_rids()
map(lambda x: self._validate(self.master_id, x), rids)
if attach_same_vip:
utils.check_lbs_vip(vip_id, rids)
else:
utils.check_lbs_vip(vip_id, [origin_inst_id1])
else:
map(lambda x: self._validate(self.master_id, x), rids)
def create_with_config(self):
config = utils.get_config_bytenant()
if not config:
raise Exception("not found template config")
self.inst_id = utils.create_rds_byclient("HA",config_id=config.id)
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=240)
self.vip_id = utils.get_vip_id(self.inst_id)
_inst = utils.get_builtin_instance(self.inst_id)
self.group_id = _inst.group_id
self.inst_id =utils.get_instance_id(self.group_id, DBInstanceType.MASTER)
self.dbslave_id = utils.get_instance_id(self.group_id, DBInstanceType.STANDBY)
LOG.info('master_id:%s slave_id:%s' % (self.inst_id, self.dbslave_id))
self.validate()
utils.check_instancedb_mysql_variables(self.inst_id)
utils.check_instancedb_mysql_variables(self.dbslave_id)
utils.check_instace_config_is_same(self.inst_id)
def restart(self):
self.inst_id = self._get_rid()
old_rr_inst = utils.get_builtin_instance(self.inst_id)
rr_vid = old_rr_inst.virtual_instance_id
utils.check_server_status(rr_vid, DBInstanceType.READ_REPLI, InstanceTasks.NONE, utils.ServiceStatuses.RUNNING, timeout = 600)
utils.restart_mysql(rr_vid)
utils.check_server_status(rr_vid, DBInstanceType.READ_REPLI, InstanceTasks.REBOOTING, utils.ServiceStatuses.RUNNING, timeout = 60)
utils.check_server_status(rr_vid, DBInstanceType.READ_REPLI, InstanceTasks.NONE, utils.ServiceStatuses.RUNNING, timeout = 60)
self.validate()
def restorable_time_failover_twice(self):
def failover(id):
rpc.call(utils.get_context(), "taskmanager",
{"method": "failover",
"args": {'instance_id':id}})
self.create()
master = utils.get_builtin_instance(self.inst_id)
master_vid = master.virtual_instance_id
rt_1 = utils.get_restorable_time(master_vid)
# case1: failover only
failover1_time = datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')
failover(master.id)
utils.check_server_status(master_vid, DBInstanceType.MASTER, timeout = 360)
new_slave_id = master.id
utils.check_server_status(new_slave_id, DBInstanceType.STANDBY, timeout = 360)
rt_2 = utils.get_restorable_time(master_vid)
assert rt_2.end > rt_1.end and rt_1.begin <= rt_2.begin, \
("before first failover rt: [%s, %s], after first failover rt: [%s, %s]" % (rt_1.begin, rt_1.end, rt_2.begin, rt_2.end))
# case2: restart & failover
master2 = utils.get_builtin_instance(master_vid)
failover(master2.id)
utils.check_server_status(master_vid, DBInstanceType.MASTER, timeout = 360)
new_slave_id2 = master2.id
utils.check_server_status(new_slave_id2, DBInstanceType.STANDBY, timeout = 360)
rt_3 = utils.get_restorable_time(master_vid)
assert rt_3.begin > failover1_time and rt_3.end > rt_2.end, (rt_3.begin, failover1_time, rt_3.end, rt_2.end)
print rt_1.begin, rt_1.end
print rt_2.begin, rt_2.end
print rt_3.begin, rt_3.end
print "the first failover time: %s" % (failover1_time)
time.sleep(60)
rt_4 = utils.get_restorable_time(master_vid)
print rt_4.begin, rt_4.end
assert rt_4.end > rt_3.end
def _failover(self, stop_mysqld = False, rm_mysql_data = False):
self.inst_id = self._get_rid()
utils.check_server_status(self.inst_id,expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.READ_REPLI,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False,timeout=10)
self.vip_id = utils.get_vip_id(self.inst_id)
_rr_server = utils.get_builtin_instance(self.inst_id)
nova_instance = _rr_server.server
self.group_id = _rr_server.db_info.group_id
instance_id = self.inst_id
ran_count = 56
utils.generate_databases(self.master_id, count = ran_count)
ip = utils.check_allocate_ip(nova_instance)
if stop_mysqld:
utils.stop_mysqld(ip)
if rm_mysql_data:
utils.mysql_data_lost(ip)
utils.check_server_status(self.inst_id,expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.READ_REPLI,
expected_svr_status=utils.ServiceStatuses.SHUTDOWN,
deleted=False,timeout=120)
_ret = rpc.call(utils.get_context(),"taskmanager",
{
"method": "failover",
"args": {'instance_id':instance_id}
}
)
utils.check_server_status(self.inst_id,expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.READ_REPLI,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False,timeout=120)
self.master_id = utils.get_instance_id(self.group_id,DBInstanceType.MASTER)
self.inst_id = self.inst_id
utils.check_generated_databases(self.inst_id, count = ran_count)
self.validate()
def migrate(self, instance_id=None, _strategy=None):
self.inst_id = instance_id
if not self.inst_id:
self.inst_id = utils.get_instance_id_bytenant(DBInstanceType.MASTER)
_master_inst = utils.get_instance(id = self.inst_id, deleted = 0)
self.group_id = _master_inst.group_id
self.dbslave_id = utils.get_instance_id(self.group_id, DBInstanceType.STANDBY)
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=120)
utils.check_server_status(self.dbslave_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.STANDBY,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=120)
self.vip_id = utils.get_vip_id(_master_inst.id)
virtual_instance_id = None
strategy = CONF.migrate_strategy
if strategy == 'master':
_ret = utils.get_builtin_instance(_master_inst.id)
nova_instance = _ret.server
instance_id = _master_inst.id
type = DBInstanceType.MASTER
virtual_instance_id = _ret.virtual_instance_id
elif strategy == 'standby':
_ret = utils.get_builtin_instance(self.dbslave_id)
nova_instance = _ret.server
instance_id = self.dbslave_id
type = DBInstanceType.STANDBY
else:
raise Exception("not found migrate_strategy ss%s" % strategy)
ran_count = random.randint(50, 100)
utils.generate_databases(self.inst_id, count = ran_count)
_ret = rpc.call(utils.get_context(), "taskmanager",
{"method": "migrate",
"args": {'migrate_id':instance_id}})
if strategy == 'master':
raw_instance_id = inst_utils.virtual_instid_2_origin_instid(virtual_instance_id)
new_server_id = utils.get_builtin_instance(raw_instance_id).server_id
utils.check_server_status(raw_instance_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=120)
assert new_server_id != nova_instance.id
self.inst_id = raw_instance_id
elif strategy == "standby":
utils.check_server_status(_ret['id'], expected_task=utils.tasks.InstanceTasks.NONE,
type = DBInstanceType.STANDBY,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout = 123)
self.dbslave_id = _ret['id']
utils.check_generated_databases(self.inst_id, count = ran_count)
utils.check_generated_databases(self.dbslave_id, count = ran_count)
utils.clear_generated_databases(self.inst_id, count = ran_count)
self.validate()
def _failover_test(self, group_id, trigger_inst_id,
do_workload = False, do_prepare = False,
mysqld_killed = False,
host_rebooted = False,
remove_tmp_initsql = False,
mysql_data_lost = False,
check_vip = False,
check_rpl_consist = True,
check_binlog_range = False):
LOG.info("Doing Failover Test, group_id:%s, instance_id:%s, do_workload:%s, do_prepare:%s." %
(group_id, trigger_inst_id, do_workload, do_prepare))
before_group_items = InstanceGroupItem.list_by_gid(test_utils.get_context(), group_id, deleted = False)
before_items = set(map(lambda x: x.type + "_" + x.instance_id, before_group_items))
before_instance = test_utils.get_builtin_instance( trigger_inst_id)
before_rip = test_utils.check_allocate_ip(before_instance.server)
before_origin_instid = before_instance.id
rt_before = rt_after = None
if check_binlog_range:
rt_before = test_utils.get_restorable_time(trigger_inst_id)
if do_workload and before_instance.type == DBInstanceType.MASTER:
FAILOVERInstance.__run_workload(do_prepare = do_prepare)
if remove_tmp_initsql:
FAILOVERInstance.__trigger_vm_remove_tmp_sql_file(trigger_inst_id)
if mysqld_killed:
FAILOVERInstance.__trigger_mysqld_crash(trigger_inst_id)
test_utils.check_server_status(trigger_inst_id, expected_task=tasks.InstanceTasks.NONE,
type=before_instance.type, expected_svr_status=test_utils.ServiceStatuses.SHUTDOWN,
deleted=False, timeout=120)
if host_rebooted:
FAILOVERInstance.__trigger_host_reboot(trigger_inst_id)
# when host-machine rebooted, no guestagent update service's status.
# test_utils.check_server_status(trigger_inst_id, expected_task=tasks.InstanceTasks.NONE,
# type=before_instance.type, expected_svr_status=test_utils.ServiceStatuses.SHUTDOWN,
# deleted=False, timeout=120)
if mysql_data_lost:
FAILOVERInstance.__trigger_mysql_data_lost(trigger_inst_id)
rpc.call(test_utils.get_context(), "taskmanager",
{"method": "failover", "args": {'instance_id':before_origin_instid}}, timeout = 3600)
## check vip <--> rip mapping.
## vip should be changed in 10 seconds.
if before_instance.type == DBInstanceType.MASTER or before_instance.type == DBInstanceType.READ_REPLI:
after_instance = test_utils.get_builtin_instance( trigger_inst_id)
after_nova_inst = after_instance.server
after_rip = test_utils.check_allocate_ip(after_nova_inst)
assert after_instance.vip == before_instance.vip and before_rip != after_rip
if before_instance.type == DBInstanceType.MASTER:
test_utils.check_server_status(before_instance.id,
expected_task = tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status = test_utils.ServiceStatuses.RUNNING,
deleted=False, timeout=120)
## check replication topo
after_group_items = InstanceGroupItem.list_by_gid(test_utils.get_context(), group_id, deleted = False)
after_items = set(map(lambda x: x.type + "_" + x.instance_id, after_group_items))
LOG.info("before " + str(before_items))
LOG.info("after " + str(after_items))
if check_rpl_consist:
diff_items = (before_items - after_items)
# assert len(diff_items) == 0
assert len(before_group_items) == len(after_group_items), "size of mysql cluster should be the same."
for group_item in after_group_items:
if group_item.type == DBInstanceType.STANDBY and group_item.instance_id == before_instance.id:
item = InstanceGroupItem.get_by_instance_id(test_utils.get_context(), group_item.instance_id, deleted = False)
assert item != None
continue
test_utils.check_server_status(group_item.instance_id,
expected_task = tasks.InstanceTasks.NONE,
type = group_item.type,
expected_svr_status = test_utils.ServiceStatuses.RUNNING,
deleted = False, timeout = 120)
if check_binlog_range:
rt_after = test_utils.get_restorable_time(trigger_inst_id)
assert rt_after.end > rt_before.end, (rt_after.end, rt_before.end)
time.sleep(60)
rt_after2 = test_utils.get_restorable_time(trigger_inst_id)
assert rt_after2.end > rt_after.end, (rt_after2.end, rt_after.end)
def failover(self, instance_id=None, _strategy=None):
self.inst_id = instance_id
if not self.inst_id:
self.inst_id = utils.get_instance_id_bytenant(DBInstanceType.MASTER)
_inst = utils.get_builtin_instance(self.inst_id)
self.group_id = _inst.group_id
self.dbslave_id = utils.get_instance_id(self.group_id, DBInstanceType.STANDBY)
utils.check_server_status(self.inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=120)
utils.check_server_status(self.dbslave_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.STANDBY,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=120)
self.vip_id = utils.get_vip_id(self.inst_id)
strategy = CONF.ha_failover_strategy
virtual_instance_id = None
if strategy == 'master':
_ret = utils.get_builtin_instance(self.inst_id)
nova_instance = _ret.server
instance_id = _ret.id
type = DBInstanceType.MASTER
virtual_instance_id = _ret.virtual_instance_id
elif strategy == 'standby':
_ret = utils.get_builtin_instance(self.dbslave_id)
nova_instance = _ret.server
instance_id = _ret.id
type = DBInstanceType.STANDBY
else:
raise Exception("not found ha_failover_strategy %s" % strategy)
rancount = random.randint(50, 100)
utils.generate_databases(self.inst_id, count = rancount)
ip = utils.check_allocate_ip(nova_instance)
utils.stop_mysqld(ip, stop_ga=True)
utils.check_server_status(instance_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=type,
expected_svr_status=utils.ServiceStatuses.SHUTDOWN,
deleted=False, timeout=120)
rpc.call(utils.get_context(), "taskmanager",
{"method": "failover",
"args": {'instance_id':instance_id}})
if strategy == 'master':
origin_inst_id = inst_utils.virtual_instid_2_origin_instid(virtual_instance_id)
self.inst_id = origin_inst_id
utils.check_server_status(origin_inst_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.MASTER,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=120)
new_slave_id = utils.get_instance_id(self.group_id, DBInstanceType.STANDBY)
utils.check_server_status(new_slave_id, expected_task=utils.tasks.InstanceTasks.NONE,
type=DBInstanceType.STANDBY,
expected_svr_status=utils.ServiceStatuses.RUNNING,
deleted=False, timeout=120)
self.dbslave_id = new_slave_id
utils.check_generated_databases(self.inst_id, count = rancount)
utils.clear_generated_databases(self.dbslave_id, count = rancount)
self.validate()