def post(self):
username = self.get_argument('useranme')
password = self.get_argument('password')
# valid: 强制密码修改为高强度口令（12位以上，需包括大小写字母和数字混合）
def valid_passwd(pw):
MIN_LEN = 12
if(len(pw)<12):raise Exception("password too short, min length is %d" % MIN_LEN)
has_A = False
has_a = False
has_0 = False
for c in password:
if not has_A and c >= 'A' and c <= 'Z':
has_A = True
continue
if not has_a and c >= 'a' and c <= 'z':
has_a = True
continue
if not has_0 and c >= '0' and c <= '9':
has_0 = True
continue
if not has_A: raise Exception("password must at last has one upper char")
if not has_a: raise Exception("password must at last has one little char")
if not has_0: raise Exception("password must at last has one number char")
valid_passwd(password)
session = database.getSession()
user = session.query(User).filter(and_(User.name == username, User.password == password)).first()
session.close()
if user:
self.set_current_user({'name': user.name, 'password': user.password,"type": user.type })
self.login_jump(user.type)
else:
self.redirect('/login')
def get_host_list():
host_list = []
session = database.getSession()
for instance in session.query(Instance).filter(Instance.role == "gmond"):
host_list.append(instance.host)
session.close()
return host_list
def pre_check(self):
'''
覆盖manager的pre_check
每次检查之前都获取判断规则
'''
#获取instance列表,得到所有机器和角色的对应关系
self.role_host_map = {}
self.rule_lists = []
self.group_host_map = {}
session = database.getSession()
for instance in session.query(Instance).filter(
Instance.status == Instance.STATUS_START):
host = instance.host
role = instance.role
if not self.role_host_map.has_key(role):
self.role_host_map[role] = []
self.role_host_map[role].append(host)
for gh in session.query(GroupHost):
group = gh.group
host = gh.hostname
if not self.group_host_map.has_key(group):
self.group_host_map[group] = []
self.group_host_map[group].append(host)
for alarm in session.query(Alarm):
active_lists, except_lists = alarm.get_host_tuple()
self.rule_lists.append(
Rule(alarm.name, alarm.expression, alarm.callback,
active_lists, except_lists))
session.close()
def run():
session = database.getSession(False)
logCount = 0 ;
while True:
cbs = session.query(CallBack).filter( CallBack.status == CallBack.STATUS_CHECK )
idList = []
cbcache = {}
for cb in cbs:
idList.append(cb.taskid)
if not cbcache.has_key(cb.taskid) :
cbcache[cb.taskid]=[]
cbcache[cb.taskid].append(cb)
logCount = logCount + 1
if logCount == 10:
callback_log.info("find " + str( len(idList) ) + " need to callback" )
logCount = 0
if len(idList) != 0 :
for task in session.query(Task).filter(Task.id.in_(idList)):
callback_log.info(str(task.id)+" "+str(task.status))
if task.status == Task.STATUS_FINISH:
remove_callback(session,task.id)
callback_log.info("remove "+str(task.id) )
if cbcache.has_key(task.id) :
for cb in cbcache[task.id]:
callback_log.info("call back "+str(cb.taskid)+" "+cb.func+" "+cb.params )
deal_callback(session, task, cb.func, json.loads(cb.params))
# session.
session.commit()
#session.flush()
time.sleep(3)
def check_instance_state(self):
'''
获取所有的instance状态,报告哪些有问题
'''
service_role_map = {}
#get all instance
session = database.getSession()
for instance in session.query(Instance):
result,msg = self.check_instance(instance)
self.update_result(result,msg)
if not service_role_map.has_key(instance.service) :
service_role_map[instance.service] = []
service_role_map[instance.service].append(instance)
for (service,roles) in service_role_map.items():
if service == "zookeeper" :
zk_leader_port = database.get_service_conf(session,"zookeeper","zookeeper_leader_port")
result,msg = self.check_zk_leader(zk_leader_port,roles)
self.update_result(result,msg)
elif service == "hbase" :
#TODO
hbase_master_info_port = database.get_service_conf(session,"hbase","hbase_master_info_port")
result,msg = self.check_hbase_master(hbase_master_info_port,roles)
self.update_result(result,msg)
session.close()
return (self.state,self.msg)
def get_cluster_name():
cluster_name = ""
session = database.getSession()
cluster_name = database.get_service_conf(session, "ganglia",
"cluster_name")
session.close()
return cluster_name
def signUp():
session = getSession()
method = request.method
if(method == 'POST'):
name = request.form.get('name')
lastname = request.form.get('lastname')
email = request.form.get('email')
dateBirth=request.form.get('dateBirth')
password = toBytes(request.form.get('password'))
gender=request.form.get('gender_id')
try:
user = User()
user.email = email
user.name = name
user.dateBirth=dateBirth
user.gender_id=gender
user.lastname = lastname
hashed = bcrypt.hashpw(
password, bcrypt.gensalt()
)
user.password = hashed.decode("utf-8")
session.add(user)
session.commit()
login_user(user)
return redirect(url_for('main'))
except Exception as error:
session.rollback()
raise error
genders=session.query(Gender).all()
return render_template('signUp.html',genders=genders)
def add_host(id, hosts, user, port, passwd, sudopasswd):
async_set(id, "progressMsg", "begin to run")
async_set(id, "progress", "10")
time.sleep(2)
async_set(id, "progressMsg", "ssh to")
async_set(id, "progress", "20")
time.sleep(2)
async_set(id, "progressMsg", "get info")
async_set(id, "progress", "80")
time.sleep(2)
async_set(id, "progressMsg", "finish")
async_set(id, "progress", "100")
#保留10秒后清除
session = database.getSession()
for hostName in hosts:
host = Host(hostName)
host.ip = "192.168.1.1"
host.cpu = "fade8core"
host.mem = "fade100G"
host.disk = "fade100T"
host.rack = "unknow"
session.add(host)
session.commit()
time.sleep(10)
async_remove(id)
def run_update(task):
session = database.getSession(False)
session.query(Task).filter(Task.id == task.id).update(
{Task.status: task.status, Task.result: task.result, Task.msg: task.msg}
)
session.commit()
session.close()
def create():
method = request.method
session = getSession()
if (method == 'POST'):
date = request.form.get('date')
name = request.form.get('name')
category_id = request.form.get('category')
track = request.form.get('track')
competence = Competence()
competence.date = date
competence.name_competence = name
competence.category_id = category_id
competence.track_id = track
try:
session.add(competence)
session.commit()
userCompetence = UserCompetence()
userCompetence.user_id = current_user.id
userCompetence.competences_id = competence.id
session.add(userCompetence)
session.commit()
return redirect(url_for('competence.index'))
except Exception as error:
session.rollback()
raise error
categories = session.query(Category).all()
tracks = session.query(Track).all()
return render_template('createCompetences.html',
categories=categories,
tracks=tracks)
def add_host(id,hosts,user,port,passwd,sudopasswd):
async_set(id,"progressMsg","begin to run")
async_set(id,"progress","10")
time.sleep(2)
async_set(id,"progressMsg","ssh to")
async_set(id,"progress","20")
time.sleep(2)
async_set(id,"progressMsg","get info")
async_set(id,"progress","80")
time.sleep(2)
async_set(id,"progressMsg","finish")
async_set(id,"progress","100")
#保留10秒后清除
session=database.getSession()
for hostName in hosts:
host = Host(hostName)
host.ip="192.168.1.1"
host.cpu="fade8core"
host.mem="fade100G"
host.disk="fade100T"
host.rack="unknow"
session.add(host)
session.commit()
time.sleep(10)
async_remove(id)
def collectMetrics(self):
# 获取当前集群的状态
metrics = self.getMetrics()
# recordTime指向上个10分钟的开头时间，所以要往后移动一个10分钟
meRecord = MetricsRecord(self.recordTime + config.collect_interval)
recordKey = [
"appsCompleted",
"appsPending",
"appsRunning",
"appsFailed",
"appsKilled",
"totalMB",
"availableMB",
"allocatedMB",
"containersAllocated",
"totalNodes",
"activeNodes",
]
# totalMB=availableMB+allocatedMB
temp = metrics["clusterMetrics"]
temp["totalMB"] = int(temp["availableMB"]) + int(temp["allocatedMB"])
for key in recordKey:
meRecord.set(key, temp[key])
session = database.getSession()
session.merge(meRecord)
session.commit()
def createPost():
session = getSession()
name = request.form.get('name')
location = request.form.get('location')
size = request.form.get('size')
try:
""" connection.execute(
"INSERT INTO tracks(name,location,size) VALUES(%s, %s,%s)",
name, location, size
) """
track = Track()
track.name = name
track.location = location
track.size = size
session.add(track)
session.commit()
files = request.files.getlist("images")
imagesPaths = uploadfiles(files, folderTracks)
for imagesPath in imagesPaths:
trackImage = TrackImage()
trackImage.src = imagesPath
trackImage.track_id = track.id
session.add(trackImage)
session.commit()
return redirect(url_for('track.tracks'))
except Exception as error:
session.rollback()
raise error
return render_template('createTracks.html')
def run(self):
session = getSession()
for item in data:
track = self.db.session.query(Track).filter(
Track.name == item.get("name")
).first()
exist = bool(track)
if not exist: track = Track()
track.name = item.get("name")
track.location = item.get("location")
track.size = item.get("size") or 5000
if not exist: self.db.session.add(track)
self.db.session.commit()
images = item.get("images")
if images:
for image in images:
haveImage = self.db.session.query(TrackImage).filter(
TrackImage.track_id == track.id
).first()
if not haveImage:
ext = Path(image).suffix
namefile = uuid.uuid4().hex + ext
file = open(path + image, 'rb')
s3.put_object(Bucket=bucket, Key=to + namefile, Body=file)
trackImage = TrackImage()
trackImage.src = namefile
trackImage.track_id = track.id
self.db.session.add(trackImage)
def get_host_list():
host_list = []
session = database.getSession()
for instance in session.query(Instance).filter(Instance.role == "gmond"):
host_list.append(instance.host)
session.close()
return host_list
def _get_mail_to():
session = database.getSession()
mail_list = session.query(AlarmAssist).filter(
AlarmAssist.name == "mail_to").first()
log.info("to mail %s" % mail_list.value)
session.close()
return mail_list.value.split(",")
def trainings():
session = getSession()
trainings = session.query(Training, Track, Category).join(
Track, Category
).filter(Training.user_id == current_user.id).order_by(Training.date.desc()).limit(10).all()
return render_template('trainings.html', trainings=trainings)
def check_instance_state(self):
'''
获取所有的instance状态,报告哪些有问题
'''
service_role_map = {}
#get all instance
session = database.getSession()
for instance in session.query(Instance):
#检查实例
key_word,msg = self.check_instance(instance)
if key_word != None:
self.update_result(key_word, msg)
#采集服务
if not service_role_map.has_key(instance.service) :
service_role_map[instance.service] = []
service_role_map[instance.service].append(instance)
for (service,roles) in service_role_map.items():
if service == "zookeeper" :
zk_leader_port = database.get_service_conf(session,"zookeeper","zookeeper_leader_port")
result,msg = self.check_zk_leader(zk_leader_port,roles)
if not result:
key_word = "cluster(zk no leader)"
self.update_result(key_word, msg)
elif service == "hbase" :
#TODO
hbase_master_info_port = database.get_service_conf(session,"hbase","hbase_master_info_port")
result,msg = self.check_hbase_master(hbase_master_info_port,roles)
if not result:
key_word = "cluster(hbase no leader)"
self.update_result(key_word, msg)
session.close()
return self.alarm_list
def collectApp(self):
#获取所有的过去时段完成的app的列表
apps = self.getAppList()
if not apps or not apps["apps"]:
log.info("no appid match")
return
startCollectTime = time.time()
#轮询app列表，获取每个app的详细信息
for app in apps["apps"]["app"]:
startTime = time.time()
appid = app["id"]
if app['state'] == 'FINISHED':
try:
jobid = appidToJobid(appid)
jobHistory = self.getJobHistory(jobid)
if jobHistory:
jobCounter = self.getJobCounter(jobid)
jobTasks = self.getJobAllTask(jobid)
self.updateWithAppid(app, jobHistory, jobCounter)
else:
log.info(
"find some app run success but no history file:" +
appid)
except:
log.exception("get error while doing app " + appid)
endTime = time.time()
else:
self.updateWithNotSuccAppid(app)
log.info("getting appid: %s using %d ms" %
(appid, (endTime - startTime) * 1000))
endCollectTime = time.time()
log.info("using %d ms to collect the data" %
((endCollectTime - startCollectTime) * 1000))
startFlushTime = time.time()
#提交数据
session = database.getSession()
for (appid, appRecordValue) in self.appList.items():
session.merge(appRecordValue)
session.commit()
log.info("push %d appRecord into table" % (len(self.appList)))
for (key, NmRecordValue) in self.nmList.items():
session.merge(NmRecordValue)
session.commit()
log.info("push %d NmRecord into table" % (len(self.nmList)))
for (key, RmRecordValue) in self.rmList.items():
session.merge(RmRecordValue)
session.commit()
log.info("push %d RmRecord into table" % (len(self.rmList)))
endFlushTime = time.time()
log.info("using %d ms to push to the db" %
((endFlushTime - startFlushTime) * 1000))
def indexC():
session = getSession()
connection = session.connection()
results = connection.execute(
text('''
SELECT
competences.id,
competences.date,
tracks.name as track,
categories.NAME AS category,
tracks.size AS track_size
FROM
public.competences
INNER JOIN user_competences ON public.competences.id = public.user_competences.competences_id
INNER JOIN categories ON competences.category_id = categories.id
INNER JOIN tracks ON competences.track_id = tracks.id
WHERE
user_competences.user_id = :user_id;'''), ({
"user_id": current_user.id
}))
competences = query_to_dict(results)
print(competences)
competences = session.query(
Competence, Track, Category, UserCompetence).join(
UserCompetence, Track,
Category).filter(UserCompetence.user_id == current_user.id).all()
print(competences)
return render_template('competence-table.html', competences=competences)
def pre_check(self):
'''
覆盖manager的pre_check
每次检查之前都获取判断规则
'''
#获取instance列表,得到所有机器和角色的对应关系
self.role_host_map = {}
self.rule_lists = []
self.group_host_map = {}
session = database.getSession()
for instance in session.query(Instance).filter(Instance.status==Instance.STATUS_START) :
host = instance.host
role = instance.role
if not self.role_host_map.has_key(role) :
self.role_host_map[role] = []
self.role_host_map[role].append(host)
for gh in session.query(GroupHost):
group = gh.group
host = gh.hostname
if not self.group_host_map.has_key(group) :
self.group_host_map[group] = []
self.group_host_map[group].append(host)
for alarm in session.query(Alarm):
active_lists, except_lists = alarm.get_host_tuple()
self.rule_lists.append( Rule(alarm.name,alarm.expression,alarm.callback,active_lists,except_lists) )
session.close()
def graphicPreviousCompetitions(id):
session = getSession()
competence = session.query(Competence).filter(Competence.id == id).first()
competences = session.query(
Competence, UserCompetence).join(UserCompetence).filter(
Competence.date <= competence.date,
UserCompetence.user_id == current_user.id,
Competence.track_id == competence.track_id).limit(6).all()
result = []
for item in competences:
c = item.Competence
uc = item.UserCompetence
result.append({
'turns': uc.turns,
'duration': uc.duration,
'competence': {
'id': c.id,
'date': c.date.isoformat(),
'name_competence': c.name_competence,
'track_id': c.track_id,
'category_id': c.category_id,
},
})
return jsonify(result)
def fade_add_del_service(addRunningId,delRunningId,msg=""):
time.sleep(5)
session=database.getSession()
for id in addRunningId:
database.update_task(session,id,Task.STATUS_FINISH,Task.RESULT_SUCCESS,msg)
for id in delRunningId:
database.update_task(session,id,Task.STATUS_FINISH,Task.RESULT_SUCCESS,msg)
session.close()
def namenode_web(self):
session = database.getSession()
nm_web_port = database.get_service_conf(session,"hdfs","dfs_namenode_http_address_port")
nms = []
for inst in session.query(Instance).filter(Instance.role == "namenode"):
nms.append(inst.host)
session.close()
return namenode_web.namenode_web(nms, nm_web_port)
def init_alarm_autofix():
session = database.getSession()
session.merge(AlarmAutofix(".*(datanode down)",15,"hdfs","datanode","restart"))
session.merge(AlarmAutofix(".*(nodemanager down)",15,"yarn","nodemanager","restart"))
session.commit()
session.close()
def one(id):
session = getSession()
competence = session.query(
Competence, Track, Category,
UserCompetence).join(UserCompetence, Track, Category).filter(
Competence.id == id,
UserCompetence.user_id == current_user.id).first()
return render_template('competence.html', competence=competence)
def run_update(instance):
session = database.getSession(False)
session.query(Instance).filter(Instance.id==instance.id) \
.update({Instance.health: instance.health,
Instance.monitor_time: instance.monitor_time,
Instance.msg: instance.msg})
session.commit()
session.close()
def update_msg(task, line):
session = database.getSession(False)
msg = task.msg or ""
msg = msg + line
task.msg = msg
session.query(Task).filter(Task.id == task.id).update({Task.msg: task.msg})
session.commit()
session.close()
def run_update(task):
session = database.getSession(False)
session.query(Task).filter(Task.id==task.id) \
.update({Task.status: task.status,
Task.result: task.result,
Task.msg: task.msg})
session.commit()
session.close()
def init_alarm_assist():
session = database.getSession()
#alarm var
session.merge(AlarmAssist("mail_to","*****@*****.**"))
session.merge(AlarmAssist("ignore_key_word",""))
session.commit()
session.close()
def init_alarm_assist():
session = database.getSession()
#alarm var
session.merge(AlarmAssist("mail_to", "*****@*****.**"))
session.merge(AlarmAssist("ignore_key_word", ""))
session.commit()
session.close()
def update_msg(task, line):
session = database.getSession(False)
msg = task.msg or ""
msg = msg + line
task.msg = msg
session.query(Task).filter(Task.id==task.id) \
.update({Task.msg: task.msg})
session.commit()
session.close()
def run_update(instance):
session = database.getSession(False)
try:
session.query(Instance).filter(Instance.id==instance.id) \
.update({Instance.health: instance.health,
Instance.monitor_time: instance.monitor_time,
Instance.msg: instance.msg})
session.commit()
finally:
session.close()
def delete(id):
session = getSession()
training = session.query(Training).filter(
Training.id == id, Training.user_id == current_user.id
)
if(training.first()):
training.delete()
session.commit()
return redirect(url_for('training.trainings'))
return "No tienes acceso"
def delete(id):
session = getSession()
competence = session.query(UserCompetence).filter(
UserCompetence.competences_id == id,
UserCompetence.user_id == current_user.id)
if competence.first():
competence.delete()
session.commit()
return redirect(url_for('competence.index'))
return "No tienes acceso"
def collectApp(self):
#获取所有的过去时段完成的app的列表
apps = self.getAppList()
if not apps or not apps["apps"]:
log.info("no appid match")
return
startCollectTime = time.time()
#轮询app列表，获取每个app的详细信息
for app in apps["apps"]["app"]:
startTime = time.time()
appid = app["id"]
if app['state'] == 'FINISHED':
try:
jobid = appidToJobid(appid)
jobHistory = self.getJobHistory(jobid)
if jobHistory:
jobCounter = self.getJobCounter(jobid)
jobTasks = self.getJobAllTask(jobid)
self.updateWithAppid(app,jobHistory,jobCounter)
else:
log.info("find some app run success but no history file:"+appid)
except:
log.exception("get error while doing app "+appid)
endTime = time.time()
else:
self.updateWithNotSuccAppid(app)
log.info("getting appid: %s using %d ms" % (appid, (endTime - startTime)*1000))
endCollectTime = time.time()
log.info("using %d ms to collect the data" % ((endCollectTime - startCollectTime)*1000) )
startFlushTime = time.time()
#提交数据
session = database.getSession()
for (appid,appRecordValue) in self.appList.items():
session.merge(appRecordValue)
session.commit()
log.info("push %d appRecord into table" % (len(self.appList)))
for (key,NmRecordValue) in self.nmList.items():
session.merge(NmRecordValue)
session.commit()
log.info("push %d NmRecord into table" % (len(self.nmList)))
for (key,RmRecordValue) in self.rmList.items():
session.merge(RmRecordValue)
session.commit()
log.info("push %d RmRecord into table" % (len(self.rmList)))
endFlushTime = time.time()
log.info("using %d ms to push to the db" % ((endFlushTime - startFlushTime)*1000))
def init_alarm_autofix():
session = database.getSession()
session.merge(
AlarmAutofix(".*(datanode down)", 15, "hdfs", "datanode", "restart"))
session.merge(
AlarmAutofix(".*(nodemanager down)", 15, "yarn", "nodemanager",
"restart"))
session.commit()
session.close()
def conf_to_db(dir):
group_var_dir = os.path.join(dir, "group_vars")
host_var_dir = os.path.join(dir, "host_vars")
session = database.getSession()
#将机器和组的关系写入数据库
#TODO
#读取组变量
for group in os.listdir(group_var_dir):
if group.startswith("."):
continue
stream = open(os.path.join(group_var_dir, group), "r")
vars = yaml.load(stream)
for name in vars:
value = vars[name]
if name.find("__") > 0:
(service, new_name) = name.split("__")
else:
service = ""
new_name = name
if isinstance(value, list):
gv = GroupVar(group, service, new_name, ",".join(value), 1, "")
else:
gv = GroupVar(group, service, new_name, value, 0, "")
session.merge(gv)
#读取机器变量
if os.path.exists(host_var_dir):
for host in os.listdir(host_var_dir):
if host.startswith("."):
continue
stream = open(os.path.join(host_var_dir, host), "r")
vars = yaml.load(stream)
for name in vars:
value = vars[name]
if name.find("__") > 0:
(service, new_name) = name.split("__")
else:
service = ""
new_name = name
if isinstance(value, list):
hv = HostVar(host, service, new_name, ",".join(value), 1,
"")
else:
hv = HostVar(host, service, new_name, value, 0, "")
session.merge(hv)
session.commit()
session.close()
def loginPost():
email = request.form.get('email')
password = toBytes(request.form.get('password'))
session = getSession()
user = session.query(User).filter_by(email=email).first()
hashed = toBytes(user.password)
if bcrypt.checkpw(password, hashed):
login_user(user)
return redirect(url_for('main'))
else:
return "Constraseña incorrecta"
def getdb():
session = database.getSession()
task = Task("1", "2", "3")
session.add(task)
session.flush()
print task.id
session.commit()
objects = session.query(Task)
for object in objects:
print object.id, object.task
print objects
def getdb():
session=database.getSession()
task = Task("1","2","3")
session.add(task)
session.flush()
print task.id
session.commit()
objects = session.query(Task)
for object in objects:
print object.id,object.task
print objects
def getPerformance(training: Training,
track: Track = None,
category: Category = None):
session = getSession()
if not track:
track = session.query(Track).get(training.track_id)
if not category:
category = session.query(Category).get(training.category_id)
size = int(track.size) / 1000
turns = (7 / (size * 60)) / (1 / (3 * category.duration_max))
return (training.turns * 100) / turns
def post(self):
username = self.get_argument('useranme')
password = self.get_argument('password')
session = database.getSession()
user = session.query(User).filter(and_(User.name == username, User.password == password)).first()
session.close()
if user:
self.set_current_user({'name': user.name, 'password': user.password,"type": user.type })
self.login_jump(user.type)
else:
self.redirect('/login')
def resourcemanager_web(self):
session = database.getSession()
rm_port = database.get_service_conf(session,"yarn","yarn_rm_webapp_port")
rms = []
for inst in session.query(Instance).filter(Instance.role == "resourcemanager"):
rms.append(inst.host)
session.close()
if len(rms) == 0 :
return (contants.ALARM_ERROR, u"%s 检查不到有resourcemanager" % self.rule.name )
if len(rms) != 1 :
return (contants.ALARM_ERROR, u"%s 检查到有多个resourcemanager %s" % (self.rule.name, ",".join(rms) ) )
return resourcemanager_web.resourcemanager_web(rms[0],rm_port)
def conf_to_db(dir):
group_var_dir=os.path.join(dir,"group_vars")
host_var_dir=os.path.join(dir,"host_vars")
session=database.getSession()
#将机器和组的关系写入数据库
#TODO
#读取组变量
for group in os.listdir(group_var_dir):
if group.startswith("."):
continue
stream=open(os.path.join(group_var_dir,group),"r")
vars=yaml.load(stream)
for name in vars:
value = vars[name]
if name.find("__") > 0:
(service,new_name) = name.split("__")
else:
service=""
new_name=name
if isinstance(value,list):
gv = GroupVar(group,service,new_name,",".join(value),1,"")
else:
gv = GroupVar(group,service,new_name,value,0,"")
session.merge(gv)
#读取机器变量
if os.path.exists(host_var_dir):
for host in os.listdir(host_var_dir):
if host.startswith("."):
continue
stream=open(os.path.join(host_var_dir,host),"r")
vars=yaml.load(stream)
for name in vars:
value = vars[name]
if name.find("__") > 0:
(service,new_name) = name.split("__")
else:
service=""
new_name=name
if isinstance(value,list):
hv = HostVar(host,service,new_name,",".join(value),1,"")
else:
hv = HostVar(host,service,new_name,value,0,"")
session.merge(hv)
session.commit()
session.close()
def before_request():
if 'access_key' in session:
try:
g.sess = database.getSession(session['id'])
g.user = g.sess.user
except:
g.user = None
g.sess = None
else:
g.user = None
g.sess = None
def load_db_conf():
conf = {}
session=database.getSession()
for gv in session.query(GroupVar):
if gv.group == "all" :
if gv.service !=None and len(gv.service) > 0 :
name = "%s__%s" % (gv.service,gv.name)
else:
name = gv.name
conf[name] = { "value": gv.value , "type": gv.type }
session.close()
return conf
def load_db_conf():
conf = {}
session = database.getSession()
for gv in session.query(GroupVar):
if gv.group == "all":
if gv.service != None and len(gv.service) > 0:
name = "%s__%s" % (gv.service, gv.name)
else:
name = gv.name
conf[name] = {"value": gv.value, "type": gv.type}
session.close()
return conf
def fade_add_del_service(addRunningId, delRunningId, msg=""):
'''
已经弃用
'''
time.sleep(5)
session = database.getSession()
for id in addRunningId:
database.update_task(session, id, Task.STATUS_FINISH,
Task.RESULT_SUCCESS, msg)
for id in delRunningId:
database.update_task(session, id, Task.STATUS_FINISH,
Task.RESULT_SUCCESS, msg)
session.close()
def getTrending(key: str, model: Union[Track, Category], limit=8):
session = getSession()
sumColumn = (
func.count(getattr(Training, key)) +
func.count(getattr(Competence, key))
).label("count")
query = session.query(
sumColumn,
model
).join(Training, Competence, UserCompetence, isouter=True)\
.filter(or_(Training.user_id == current_user.id, UserCompetence.user_id == current_user.id))\
.group_by(model.id).order_by(sumColumn.desc())\
.limit(limit)
return query.all()
def query_all_instance():
"""query all instance
like sql : "SELECT id, service, role, hostname FROM instance"
return: [ instance ]
"""
session = database.getSession(False)
try:
rows = session.query(Instance)
finally:
session.close()
return rows
def query_tasks(ids=None):
"""query special id tasks
param:
ids: [ taskid ]
return: [ task ]
"""
session = database.getSession(False)
tasks = session.query(Task).filter(
and_(Task.status.in_([Task.STATUS_INIT]), Task.taskType.in_(["auto_fix", "ansible", "shell"]))
)
if ids:
tasks = tasks.filter(Task.id.in_(ids))
session.close()
return tasks
def init_alarm():
session = database.getSession()
for alarm in session.query(Alarm):
session.delete(alarm)
session.commit()
#base machine resource check
session.add(Alarm("check load one","max(load_one,cpu_num,cpu_num*2)","send_mail","all"))
session.add(Alarm("check disk use","disk_use(90,95)","send_mail","all"))
session.add(Alarm("check swap","max((swap_total-swap_free)/mem_total,0.5,0.7)","send_mail","all"))
session.add(Alarm("check memory","min(mem_free+mem_cached,2048*1024,1024*1024)","send_mail","all"))
session.add(Alarm("check bytesin","max(bytes_in,50*1024*1024,80*1024*1024)","send_mail","all"))
session.add(Alarm("check bytesout","max(bytes_out,50*1024*1024,80*1024*1024)","send_mail","all"))
session.add(Alarm("check swapadd","max(swap_free-PRE_swap_free,10*1024*1024,100*1024*1024)","send_mail","all"))
#instance check state
session.add(Alarm("check instance","check_instance_state()","send_mail","cluster"))
#zookeeper check
session.add(Alarm("check zk heap","max(zookeeper_memory_memHeapUsed/zookeeper_memory_memHeapCommitted,0.8,0.9)","send_mail","zookeeper"))
#hdfs check
session.add(Alarm("check missing blocks","max(dfs.FSNamesystem.MissingBlocks,0,5)","send_mail","namenode"))
session.add(Alarm("check lost heartbeat","max(dfs.FSNamesystem.ExpiredHeartbeats-PRE_dfs.FSNamesystem.ExpiredHeartbeats,0,0)","send_mail","namenode"))
session.add(Alarm("check nmweb","namenode_web()","send_mail","cluster"))
session.add(Alarm("check nn heap","max(jvm.JvmMetrics.ProcessName=NameNode.MemHeapUsedM/jvm.JvmMetrics.ProcessName=NameNode.MemHeapCommittedM,0.8,0.9)","send_mail","namenode"))
session.add(Alarm("check dn heap","max(jvm.JvmMetrics.ProcessName=DataNode.MemHeapUsedM/jvm.JvmMetrics.ProcessName=DataNode.MemHeapCommittedM,0.8,0.9)","send_mail","datanode"))
#yarn check
session.add(Alarm("check rmweb","resourcemanager_web()","send_mail","cluster"))
session.add(Alarm("check rm heap","max(jvm.JvmMetrics.ProcessName=ResourceManager.MemHeapUsedM/jvm.JvmMetrics.ProcessName=ResourceManager.MemHeapCommittedM,0.8,0.9)","send_mail","resourcemanager"))
session.add(Alarm("check nm heap","max(jvm.JvmMetrics.ProcessName=NodeManager.MemHeapUsedM/jvm.JvmMetrics.ProcessName=NodeManager.MemHeapCommittedM,0.8,0.9)","send_mail","nodemanager"))
#hbase
session.add(Alarm("check hm heap","max(hbasemaster_memory_memHeapUsed/hbasemaster_memory_memHeapCommitted,0.8,0.9)","send_mail","hbasemaster"))
session.add(Alarm("check rs heap","max(regionserver_memory_memHeapUsed/regionserver_memory_memHeapCommitted,0.8,0.9)","send_mail","regionserver"))
#hive
session.add(Alarm("check metastore heap","max(hivemetastore_memory_memHeapUsed/hivemetastore_memory_memHeapCommitted,0.8,0.9)","send_mail","hivemetastore"))
session.add(Alarm("check hiveserver heap","max(hiveserver_memory_memHeapUsed/hiveserver_memory_memHeapCommitted,0.8,0.9)","send_mail","hiveserver"))
session.add(Alarm("check hiveserver2 heap","max(hiveserver2_memory_memHeapUsed/hiveserver2_memory_memHeapCommitted,0.8,0.9)","send_mail","hiveserver2"))
session.commit()
session.close()
def resourcemanager_web(self):
session = database.getSession()
rm_port = database.get_service_conf(session,"yarn","yarn_rm_webapp_port")
rms = []
for inst in session.query(Instance).filter(Instance.role == "resourcemanager"):
rms.append(inst.host)
session.close()
alarm_list = []
if len(rms) == 0 :
key_word = "cluster(no rm)"
msg = u"%s 检查不到有resourcemanager" % self.rule.name
return [{"key_word":key_word,"msg":msg}]
if len(rms) != 1 :
key_word = "cluster(too much rm)"
msg = u"%s 检查到有多个resourcemanager %s" % (self.rule.name, ",".join(rms) )
return [{"key_word":key_word,"msg":msg}]
return resourcemanager_web.resourcemanager_web(rms[0],rm_port)
def _save_to_db(self):
session = database.getSession()
groups = self.groups
# 组变量
for (gname, group) in groups.items():
if gname == self.META: continue
if self.VARS not in group: continue
gname = gname.lower()
for (name, value) in group[self.VARS].items():
if name.find("__") > 0:
(service, new_name) = name.split("__")
else:
service=""
new_name=name
if isinstance(value, list):
gv = GroupVar(gname, service, new_name, ",".join(value), 1, "")
else:
gv = GroupVar(gname, service, new_name, value, 0, "")
session.merge(gv)
session.commit()
#读取机器变量
host_vars = {}
if self.META in groups and self.HVARS in groups[self.META]:
host_vars = groups[self.META][self.HVARS]
for (host, vars) in host_vars.items():
for (name, value) in vars.items():
if name.find("__") > 0:
(service, new_name) = name.split("__")
else:
service=""
new_name=name
if isinstance(value,list):
hv = HostVar(host,service,new_name,",".join(value),1,"")
else:
hv = HostVar(host,service,new_name,value,0,"")
session.merge(hv)
session.commit()
session.close()
def init_config(self):
'''
初始化关于获取端口等的内容
默认从数据库中获取,由于可能只使用监控
所以本函数可以直接修改为固定值
'''
session = database.getSession()
insts = session.query(Instance).filter(Instance.role=="resourcemanager")
for inst in insts:
self.rmhost = inst.host;
self.rmport=database.get_service_conf(session,"yarn","yarn_rm_webapp_port")
insts = session.query(Instance).filter(Instance.role=="historyserver")
for inst in insts:
self.hshost = inst.host;
self.hsport = database.get_service_conf(session,"yarn","mapreduce_jobhistory_webapp_port")
session.close()
def update_db(diff_result):
session = database.getSession()
group = "all"
file_only = diff_result["file_only"]
for temp in file_only:
name = temp["name"]
value = temp["value"]
type = temp["type"]
print "import name:%s value: %s" % (name, value)
if name.find("__") > 0:
(service, new_name) = name.split("__")
else:
service = ""
new_name = name
if type == 1:
gv = GroupVar(group, service, new_name, value, 1, "")
else:
gv = GroupVar(group, service, new_name, value, 0, "")
session.add(gv)
session.commit()
session.close()
def conf_to_db(file):
print "import %s into db" % file
session=database.getSession()
stream=open(file,"r")
vars=yaml.load(stream)
group="all"
for name in vars:
value = vars[name]
if name.find("__") > 0:
(service,new_name) = name.split("__")
else:
service=""
new_name=name
if isinstance(value,list):
gv = GroupVar(group,service,new_name,",".join(value),1,"")
else:
gv = GroupVar(group,service,new_name,value,0,"")
session.merge(gv)
session.close()