def update_check_plan(app_config_dict, queue):
logger = logging.getLogger("task.generator")
pipe = REDIS_DB.pipeline()
change_list = pop_multi(pipe, '_app_change_list', 30)
logger.info(u'发现应用配置发生变动,应用列表:%s', change_list)
for appname in set(change_list):
# 获取此应用的最新配置
new_config = REDIS_DB.hgetall(appname + '_config')
# 1. 此应用已经被删除了
if new_config is None:
app_config_dict.pop(appname, None)
continue
item = app_config_dict.get(appname)
# 2. 此应用是新添加的应用
if item is None:
host_list = new_config['host_list']
for host in host_list.split(','):
task = CheckTask(appname, host,
int(new_config['check_interval']))
queue.put_task(task)
else:
old_host_list = item['host_list'].split(',')
new_host_list = new_config['host_list'].split(',')
for host in (set(new_host_list) - set(old_host_list)):
task = CheckTask(appname, host,
int(new_config['check_interval']))
queue.put_task(task)
# 更新配置
app_config_dict[appname] = new_config
def get_config():
# 1. 获取应用配置
app_list_str = REDIS_DB.get('_all_app_list')
app_list = []
if app_list_str:
app_list = app_list_str.split(',')
app_config_dict = {}
pipe = REDIS_DB.pipeline()
for appname in app_list:
pipe.hgetall(appname + '_config')
res_list = pipe.execute()
for i in range(len(res_list)):
app_config_dict[app_list[i]] = res_list[i]
return app_config_dict
def record_app_status(task, status, info, check_latency, duration):
pipe = REDIS_DB.pipeline()
# 1. 获取应用的部分配置
pipe.hget(task.appname + '_config', 'max_check_attempts')
pipe.hget(task.appname + '_config', 'notify_interval')
pipe.hget(task.appname + '_config', 'check_interval')
pipe.hget(task.appname + '_config', 'parent_app_list')
res_list = pipe.execute()
max_check_attempts = int(res_list[0])
notify_interval = int(res_list[1])
check_interval = int(res_list[2])
parent_app_list = res_list[3]
# 2. 获取上一次状态信息
key = 'ah_' + task.appname + '_' + task.host
if not REDIS_DB.exists(key):
pipe.hset(key, 'current_status', 'OK')
pipe.hset(key, 'status_info', 'OK - ')
pipe.hset(key, 'current_attempt', 0)
pipe.hset(key, 'last_state_change',
datetime.now().strftime('%Y-%m-%d %H:%M:%S'))
pipe.hset(key, 'last_notification',
datetime.now().strftime('%Y-%m-%d %H:%M:%S'))
pipe.execute()
status_dict = REDIS_DB.hgetall(key)
current_status = status_dict.pop('current_status')
current_attempt = status_dict.pop('current_attempt')
current_attempt = int(current_attempt)
last_state_change = status_dict.pop('last_state_change')
last_state_change = datetime.strptime(last_state_change,
'%Y-%m-%d %H:%M:%S')
last_notification = status_dict.pop('last_notification')
last_notification = datetime.strptime(last_notification,
'%Y-%m-%d %H:%M:%S')
if current_status != status or status != 'OK':
current_attempt = current_attempt + 1
pipe.hset(key, 'current_attempt', current_attempt)
if current_attempt >= max_check_attempts:
current_status = status
pipe.hset(key, 'current_status', current_status)
pipe.hset(key, 'status_info', info)
pipe.hset(key, 'current_attempt', 0)
pipe.hset(key, 'last_state_change',
datetime.now().strftime('%Y-%m-%d %H:%M:%S'))
# recovery 事件不受notify_interval影响
if status == 'OK':
notify_app.apply_async(
(task.appname, task.host, 'RECOVERY', status, info))
margin = total_seconds(datetime.now() - last_notification)
if status != 'OK' and margin > notify_interval * 60:
pipe.hset(key, 'last_notification',
datetime.now().strftime('%Y-%m-%d %H:%M:%S'))
notify_app.apply_async(
(task.appname, task.host, 'PROBLEM', status, info))
else:
pipe.hset(key, 'current_attempt', 0)
pipe.hset(key, 'duration', duration)
pipe.hset(key, 'check_latency', check_latency)
next_scheduled_check = task.plan_time + timedelta(seconds=check_interval *
60)
next_scheduled_check = next_scheduled_check.strftime('%Y-%m-%d %H:%M:%S')
pipe.hset(key, 'next_scheduled_check', next_scheduled_check)
# -------- 其它特殊情况 (为了方便判断主机是否存活) ------------
if parent_app_list is not None:
if parent_app_list.find('HOST_STATUS') != -1:
key = 'ah_HOST_STATUS_' + task.host
pipe.hset(key, 'current_status', current_status)
if parent_app_list.find('SYS_PING') != -1:
key = 'ah_SYS_PING_' + task.host
pipe.hset(key, 'current_status', current_status)
pipe.execute()
def record_app_status(task, status, info, check_latency, duration):
pipe = REDIS_DB.pipeline()
# 1. 获取应用的部分配置
pipe.hget(task.appname + '_config', 'max_check_attempts')
pipe.hget(task.appname + '_config', 'notify_interval')
pipe.hget(task.appname + '_config', 'check_interval')
pipe.hget(task.appname + '_config', 'parent_app_list')
res_list = pipe.execute()
max_check_attempts = int(res_list[0])
notify_interval = int(res_list[1])
check_interval = int(res_list[2])
parent_app_list = res_list[3]
# 2. 获取上一次状态信息
key = 'ah_' + task.appname + '_' + task.host
if not REDIS_DB.exists(key):
pipe.hset(key, 'current_status', 'OK')
pipe.hset(key, 'status_info', 'OK - ')
pipe.hset(key, 'current_attempt', 0)
pipe.hset(key, 'last_state_change', datetime.now().strftime('%Y-%m-%d %H:%M:%S'))
pipe.hset(key, 'last_notification', datetime.now().strftime('%Y-%m-%d %H:%M:%S'))
pipe.execute()
status_dict = REDIS_DB.hgetall(key)
current_status = status_dict.pop('current_status')
current_attempt = status_dict.pop('current_attempt')
current_attempt = int(current_attempt)
last_state_change = status_dict.pop('last_state_change')
last_state_change = datetime.strptime(last_state_change, '%Y-%m-%d %H:%M:%S')
last_notification = status_dict.pop('last_notification')
last_notification = datetime.strptime(last_notification, '%Y-%m-%d %H:%M:%S')
if current_status != status or status != 'OK':
current_attempt = current_attempt + 1
pipe.hset(key, 'current_attempt', current_attempt)
if current_attempt >= max_check_attempts:
current_status = status
pipe.hset(key, 'current_status', current_status)
pipe.hset(key, 'status_info', info)
pipe.hset(key, 'current_attempt', 0)
pipe.hset(key, 'last_state_change', datetime.now().strftime('%Y-%m-%d %H:%M:%S'))
# recovery 事件不受notify_interval影响
if status == 'OK':
notify_app.apply_async((task.appname, task.host, 'RECOVERY', status, info))
margin = total_seconds(datetime.now() - last_notification)
if status !='OK' and margin > notify_interval * 60:
pipe.hset(key, 'last_notification', datetime.now().strftime('%Y-%m-%d %H:%M:%S'))
notify_app.apply_async((task.appname, task.host, 'PROBLEM', status, info))
else:
pipe.hset(key, 'current_attempt', 0)
pipe.hset(key, 'duration', duration)
pipe.hset(key, 'check_latency', check_latency)
next_scheduled_check = task.plan_time + timedelta(seconds=check_interval * 60)
next_scheduled_check = next_scheduled_check.strftime('%Y-%m-%d %H:%M:%S')
pipe.hset(key, 'next_scheduled_check', next_scheduled_check)
# -------- 其它特殊情况 (为了方便判断主机是否存活) ------------
if parent_app_list is not None:
if parent_app_list.find('HOST_STATUS') != -1:
key = 'ah_HOST_STATUS_' + task.host
pipe.hset(key, 'current_status', current_status)
if parent_app_list.find('SYS_PING') != -1:
key = 'ah_SYS_PING_' + task.host
pipe.hset(key, 'current_status', current_status)
pipe.execute()
