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 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 _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 __trigger_mysqld_crash(cls, inst_id):
inst = test_utils.get_builtin_instance( inst_id)
nova_inst = inst.server
ip = test_utils.check_allocate_ip(nova_inst)
test_utils.stop_mysqld(ip, stop_ga=True)
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 __trigger_mysql_data_lost(cls, inst_id):
inst = test_utils.get_builtin_instance( inst_id)
nova_inst = inst.server
ip = test_utils.check_allocate_ip(nova_inst)
test_utils.mysql_data_lost(ip)
def __trigger_vm_remove_tmp_sql_file(cls, inst_id):
inst = test_utils.get_builtin_instance( inst_id)
nova_inst = inst.server
ip = test_utils.check_allocate_ip(nova_inst)
test_utils.remove_tmp_initsql(ip)
def __trigger_vm_oom(cls, inst_id):
inst = test_utils.get_builtin_instance( inst_id)
nova_inst = inst.server
ip = test_utils.check_allocate_ip(nova_inst)
test_utils.reboot_host(ip)
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()
