def post(self):
title = '调度任务列表'
#需要先从azkaban登陆
session_id = self.get_argument('session_id', '')
login_user = self.get_argument('login_user', '')
if session_id == '' or login_user == '':
self.render('to_login.html')
return
#参数
query_name = self.get_argument('query_name', '')
query_project_name = self.get_argument('query_project_name', '')
query_server_host = self.get_argument('query_server_host', '')
query_user = self.get_argument('query_user', '')
#列表
jobs = Job.get_alljobs(query_name, query_project_name,
query_server_host, query_user, login_user)
query_dict = {
'query_name': query_name,
'query_project_name': query_project_name,
'query_server_host': query_server_host,
'query_user': query_user,
'session_id': session_id,
'login_user': login_user
}
logging.info('query job list [%s]' % query_dict)
self.render('list.html', title=title, jobs=jobs, query_dict=query_dict)
def get_alljobs(self):
job_list = Job.get_alljobs()
jobs = map(lambda x: {
'name': x.name,
'project_name': x.project_name
}, job_list)
self.write(json.dumps(jobs))
def post(self):
op = self.get_argument('op', '')
if op == 'show':
name = self.get_argument('name', '')
job = Job.get_job_fromdb(name)
self.write('<pre>%s</pre>' % str(job))
return
title = '任务配置'
#需要先从azkaban登陆
session_id = self.get_argument('session_id', '')
login_user = self.get_argument('login_user', '')
if session_id == '' or login_user == '':
self.render('to_login.html')
return
#参数
query_name = self.get_argument('query_name', '')
query_project_name = self.get_argument('query_project_name', '')
query_server_host = self.get_argument('query_server_host', '')
query_user = self.get_argument('query_user', '')
#任务
name = self.get_argument('name', '')
is_copy = self.get_argument('is_copy', 'false')
job = Job()
if name != '':
job = Job.get_job_fromdb(name)
if is_copy == 'true':
job.name = ''
job.loc = ''
projects = Job.get_projects(login_user)
jobs = Job.get_alljobs(login_user=login_user)
all_jobs = Job.get_alljobs()
query_dict = {
'query_name': query_name,
'query_project_name': query_project_name,
'query_server_host': query_server_host,
'query_user': query_user,
'session_id': session_id,
'login_user': login_user
}
logging.info('to update job [%s]' % name)
self.render('to_update.html',
title=title,
job=job,
jobs=jobs,
all_jobs=all_jobs,
projects=projects,
query_dict=query_dict)
def post(self):
#需要先从azkaban登陆
op = self.get_argument('op', '')
login_user = self.get_argument('login_user', '')
project_name = self.get_argument('project_name', '')
if login_user == '':
self.render('to_login.html')
return
if op == 'edit':
#项目
jobs = Job.get_alljobs(project_name=project_name,
login_user=login_user)
has_dependce_jobs = set()
is_dependced_jobs = set()
for job in jobs:
if job.dependencies != '':
has_dependce_jobs.add(job.name)
is_dependced_jobs.update(job.dependencies.split(','))
for job in jobs:
job.has_dependce = True if job.name in has_dependce_jobs else False
job.is_dependced = True if job.name in is_dependced_jobs else False
logging.info('to edit dag [%s][%s]' % (project_name, login_user))
self.render('dag_edit.html',
project_name=project_name,
jobs=jobs,
login_user=login_user)
elif op == 'save':
try:
nodes = self.get_argument('nodes', '')
links = self.get_argument('links', '')
ns = json.loads(nodes)
ls = json.loads(links)
Job.update_dag(login_user, ns, ls)
logging.info('edit dag [%s][%s]' % (project_name, login_user))
self.write("保存成功")
except Exception, e:
logging.info(e)
self.write('保存失败[%s]' % str(e))
def post(self):
#任务
name = self.get_argument('name', '')
logging.info('----name:[%s]' % name)
job = None
try:
job = Job.get_job_fromdb(name)
except:
pass
if job == None:
self.write('non_exist')
else:
self.write('exist')
def exe_job(name,execid,start_time,param_dict_str=''):
job = Job.get_job_fromdb(name)
try:
error_flag = process(name,execid,start_time,param_dict_str)
except:
traceback.print_exc()
exe_alarm(job,execid,start_time)
sysout('-------------以下为系统主动报错信息,忽略-------------------')
raise MyScheduleException('脚本执行过程出错')
if error_flag:
exe_alarm(job,execid,start_time)
sysout('-------------以下为系统主动报错信息,忽略-------------------')
raise MyScheduleException('脚本执行过程出错')
def post(self):
title = '任务配置'
#需要先从azkaban登陆
session_id = self.get_argument('session_id','')
login_user = self.get_argument('login_user','')
if session_id=='' or login_user=='':
self.render('to_login.html')
return
#参数
query_name = self.get_argument('query_name','')
query_project_name = self.get_argument('query_project_name','')
query_server_host = self.get_argument('query_server_host','')
query_user = self.get_argument('query_user','')
#任务删除
name = self.get_argument('name','')
job = Job.get_job_fromdb(name)
#job = Job()
#job.name = name
job.updater = login_user
job.unschedule_flow()
job.delete_dependencies()
job.delete_job()
#列表
jobs = Job.get_alljobs(query_name,query_project_name,query_server_host,query_user,login_user)
query_dict = {
'query_name':query_name,
'query_project_name':query_project_name,
'query_server_host':query_server_host,
'query_user':query_user,
'session_id':session_id,
'login_user':login_user
}
logging.info('[%s] delete job [%s]' % (login_user,name))
self.render('list.html',title=title,jobs=jobs,query_dict=query_dict)
def generate_files(username='',session_id=''):
#1/清空所有文件
dir_path = '%s/../files/' % CURRENTPATH
os.system('rm -rf %s' % dir_path)
os.system('mkdir %s' % dir_path)
#2/生成文件目录
#3/生成文件
job_list = Job.get_alljobs(login_user=username)
for job in job_list:
#2/生成文件目录
project_path = dir_path+job.project_name+'/'
zip_path = dir_path+job.project_name+'.zip'
if not os.path.exists(project_path):
os.system('mkdir %s' % project_path)
#3/生成文件
file_name = job.name+'.job'
file_path = project_path+file_name
f = open(file_path,'wb')
f.write(job.get_file_str())
f.close()
#4/zip包
zf = zipfile.ZipFile(zip_path,'w',zipfile.ZIP_DEFLATED)
#遍历所有文件
for dirpath, dirnames, filenames in os.walk(project_path):
for filename in filenames:
zf.write(os.path.join(dirpath,filename))
zf.close()
print '任务%s.job文件已生成' % job.name
result_list = []
#5/上传zip包
for filename in os.listdir(dir_path):
if filename.endswith('.zip'):
zip_path = dir_path+'/'+filename
project_name = filename.replace('.zip','')
result = upload_zip(session_id,project_name,zip_path)
result_list.append(result)
print result_list
return result_list
def delete_job(self):
login_user = self.username
name = self.get_argument('name')
try:
job = Job.get_job_fromdb(name)
except:
raise Exception('not fonud job[%s]' % name)
job.updater = login_user
flag, mes = job.has_job_permission()
logging.info('check job permission [%s] [%s]' % (flag, mes))
if not flag:
raise Exception(mes)
job.unschedule_flow(self.session_id)
job.delete_dependencies()
job.delete_job()
logging.info('[%s]delete job [%s]' % (login_user, name))
def schedule_flow(session_id,project_name,flow_name,schedule_time,period):
url = 'https://localhost:8443/schedule'
project_id = Job.get_projectid_byname(project_name)
sstime = datetime.datetime.strptime(schedule_time, "%Y-%m-%d %H:%M")
sdate = sstime.strftime('%m/%d/%Y')
stime = sstime.strftime('%I,%M,%p,')
params = 'is_recurring=on&period=%s&projectName=%s&flow=%s&projectId=%s&scheduleTime=%s&scheduleDate=%s' % (period,project_name,flow_name,project_id,stime, sdate)
command = '''curl -k %s -d "ajax=scheduleFlow&%s" -b azkaban.browser.session.id=%s''' % (url,params,session_id)
logging.info(command)
status,result = commands.getstatusoutput(command)
#
resp = json.loads(result[result.find('{'):])
if "error" in resp.keys():
raise Exception(resp['error'])
if 'status' in resp.keys() and resp['status']=='error':
raise Exception(resp['message'])
return resp
def get_extdep_jobs(base_time,ext_deps):
ext_jobs = json.loads(ext_deps)
query_jobs = []
for ext_job in ext_jobs['jobs']:
execute_time = get_exetime(base_time,ext_job['time_type'],ext_job['hour_diff'])
#判断小时级任务天级依赖
scripts = Job.get_job_fromdb(ext_job['name']).server_script
if ext_job['time_type']=='day' and (scripts.find('&[last_hour]')>=0 or scripts.find('&[cur_hour]')>=0):
sysout('>>>>>>>>>>>>%s为小时级任务天级依赖，生成所有任务依赖' % ext_job['name'])
for i in range(24):
job_status = JobStatus()
job_status.job_name = ext_job['name']
job_status.execute_time = '%s%s' % (execute_time,'0%s' % i if i<10 else i)
sysout('>>>>>>>>>>>>添加外部依赖任务检查项[%s][%s]' % (job_status.job_name,job_status.execute_time))
query_jobs.append(job_status)
else:
job_status = JobStatus()
job_status.job_name = ext_job['name']
job_status.execute_time = execute_time
sysout('>>>>>>>>>>>>添加外部依赖任务检查项[%s][%s]' % (job_status.job_name,job_status.execute_time))
query_jobs.append(job_status)
return query_jobs
def exe_job(name,execid,start_time,param_dict_str=''):
job = Job.get_job_fromdb(name)
if job.server_host == '' or job.server_user=='' or job.server_script=='':
raise MyScheduleException('远程配置信息不全')
#1/初始化配置
ssh = paramiko.SSHClient()
ssh.load_system_host_keys()
ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
ssh.connect(job.server_host,22,job.server_user, "")
#2/初始化默认参数,替换脚本
param_dict = get_param_dict(start_time,param_dict_str)
#拼接脚本命令,cd目录,执行脚本,获取脚本执行状态
command = ' source /etc/bashrc;'
if job.server_dir !='':
work_dir,tmp_time = replace_param(job.server_dir,param_dict)
print 'work_dir:%s' % work_dir
command = 'cd %s;' % work_dir
command = command + job.server_script
if not job.server_script.endswith(';'):
command = command + ';'
command = command+'echo "script execute status ["$?"]";'
command,execute_time = replace_param(command,param_dict)
print '>>>>>>>>>>>>%s' % (command)
#####更新job执行状态,启动######
update_jobstatus(name,execute_time,execid,0)
#3/检查跨项目或时间维度的依赖任务,未完成的两分钟检查一次,并实时输出到日志中
file_path = '%s/../executor/executions/%s/home/hadoop/service/azkaban/files/%s/_job.%s.%s.log' % (CURRENTPATH,execid,job.project_name,execid,job.name)
if job.ext_dependencies != '':
query_jobs = get_extdep_jobs(execute_time,job.ext_dependencies)
f = open(file_path,'a')
f.write('>>>>>>>>>>>>当前任务的基准时间:%s\n' % execute_time)
f.write('>>>>>>>>>>>>依赖的外部任务配置:%s\n' % job.ext_dependencies)
f.close()
while True:
flag,mes = JobStatus.is_ready(query_jobs)
f = open(file_path,'a')
f.write('>>>>>>>>>>>>依赖的外部任务准备情况:%s\n' % mes)
f.close()
if flag:
break
else:
time.sleep(120)
#4/脚本执行,输出信息
stdin, stdout, stderr = ssh.exec_command(command)
error_flag = True
print '---------------out----------------'
for line in stdout.readlines():
print line.replace('\n','')
if line.find('script execute status [0]')>=0:
error_flag=False
print '---------------err-----------------'
for line in stderr.readlines():
print line.replace('\n','')
print '---------------err end-----------------'
ssh.close()
#5/出错发邮件及短信,并抛出异常
if error_flag:
#####更新job执行状态,失败######
update_jobstatus(name,execute_time,execid,-1)
exe_alarm(job,execid,start_time)
raise MyScheduleException('脚本执行过程出错')
#####更新job执行状态,成功######
update_jobstatus(name,execute_time,execid,1)
def update_job(self):
login_user = self.username
#必需参数
required_args = [
'name', 'project_name', 'server_host', 'server_user',
'server_script', 'server_dir'
]
for arg in required_args:
self.get_argument(arg)
#生成job
attr_list = Job.get_attr_list()
#dependencies_box = self.get_argument('dependencies_box','')
job = Job()
#动态加载字段,默认均为字符串
for attr in attr_list:
value = str(self.get_argument(attr, '')).strip()
if value != '':
setattr(job, attr, value)
logging.info(attr + ':' + value)
#默认设置
job.name = job.name.replace('.', '-')
job.updater = login_user
job.update_time = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())
if job.creator == '':
job.creator = job.updater
job.create_time = job.update_time
#更新
flag, mes = job.has_job_permission()
logging.info('check job permission [%s] [%s]' % (flag, mes))
if not flag:
raise Exception(mes)
job.update_job()
logging.info('[%s] update job [%s]' % (login_user, job.name))
def process(name,execid,start_time,param_dict_str=''):
sysout('+++++++++++++任务初始化+++++++++++++++')
sysout('开始加载远程任务 %s' % name)
job = Job.get_job_fromdb(name)
if job.server_host == '' or job.server_user=='' or job.server_script=='':
raise MyScheduleException('远程配置信息不全')
sysout('任务加载成功 \n=====================\n%s\n=====================' % job)
#1/初始化配置
ssh = paramiko.SSHClient()
ssh.load_system_host_keys()
ssh.set_missing_host_key_policy(paramiko.AutoAddPolicy())
ssh.connect(job.server_host,22,job.server_user, "")
#2/初始化默认参数,替换脚本
param_dict = get_param_dict(execid,param_dict_str)
#拼接脚本命令,cd目录,执行脚本,获取脚本执行状态
command = ' source /etc/bashrc;'
if job.server_dir !='':
work_dir,tmp_time = replace_param(job.server_dir,param_dict)
sysout('配置脚本执行目录 work_dir:%s' % work_dir)
command = 'cd %s;' % work_dir
#给每个命令设备stdout不缓存的参数
scripts = ';'.join(map(lambda x:'' if x.strip()=='' else 'stdbuf -o0 %s' % x , job.server_script.split(';')))
command = command + scripts
if not job.server_script.endswith(';'):
command = command + ';'
command = command+'echo "script execute status ["$?"]";'
command,execute_time = replace_param(command,param_dict)
sysout('生成完整执行脚本:%s' % (command))
sysout('+++++++++++++任务初始化完成+++++++++++++++')
#####更新job执行状态,启动######
update_jobstatus(name,execute_time,execid,0)
#3/检查跨项目或时间维度的依赖任务,未完成的两分钟检查一次,并实时输出到日志中
if job.ext_dependencies != '':
sysout('>>>>>>>>>>>>检查外部依赖完成情况')
sysout('>>>>>>>>>>>>当前任务的基准时间:%s' % execute_time)
sysout('>>>>>>>>>>>>依赖的外部任务配置:%s' % job.ext_dependencies)
query_jobs = get_extdep_jobs(execute_time,job.ext_dependencies)
wait_count = 1
while True:
flag,mes = JobStatus.is_ready(query_jobs)
sysout('>>>>>>>>>>>>第%s次检查,依赖的外部任务准备情况:%s\n' % (wait_count,mes))
if wait_count*120 > 86400:
update_jobstatus(name,execute_time,execid,-1)
raise MyScheduleException('依赖任务等待时间超过1天,请检查')
if flag:
update_jobstatus(name,execute_time,execid,None,True)
break
else:
wait_count += 1
time.sleep(120)
#4/脚本执行,输出信息
sysout('开始执行远程脚本:\n%s' % scripts)
stdin, stdout, stderr = ssh.exec_command(command)
error_flag = True
sysout('---------------out 远程脚本日志输出----------------')
while True:
line = stdout.readline()
if len(line) == 0:
break
sysout(line.strip())
#sysout('%s,%s' % (time.time(),line))
if line.find('script execute status [0]')>=0:
error_flag=False
sysout('---------------err 远程脚本错误输出-----------------')
for line in stderr.readlines():
sysout(line.replace('\n',''))
sysout('---------------err end-----------------')
ssh.close()
#5/出错发邮件及短信,并抛出异常
if error_flag:
#####更新job执行状态,失败######
update_jobstatus(name,execute_time,execid,-1)
return True
#####更新job执行状态,成功######
exe_success(job,execid,start_time)
update_jobstatus(name,execute_time,execid,1)
return False
def post(self):
#更新完跳转到列表页
title = '调度任务列表'
#需要先从azkaban登陆
session_id = self.get_argument('session_id','')
login_user = self.get_argument('login_user','')
if session_id=='' or login_user=='':
self.render('to_login.html')
return
#参数
query_name = self.get_argument('query_name','')
query_project_name = self.get_argument('query_project_name','')
query_server_host = self.get_argument('query_server_host','')
query_user = self.get_argument('query_user','')
#生成job
attr_list = Job.get_attr_list()
dependencies_box = self.get_argument('dependencies_box','')
logging.info('>>>>>>>>>>>'+str(type(dependencies_box)))
logging.info('>>>>>>>>>>>'+str(dependencies_box))
job = Job()
#动态加载字段,默认均为字符串
for attr in attr_list:
value = str(self.get_argument(attr,'')).strip()
if value!='':
setattr(job,attr,value)
logging.info(attr+':'+value)
#默认设置
job.name = job.name.replace('.','-')
job.updater = login_user
job.update_time = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())
if job.creator == '':
job.creator = job.updater
job.create_time = job.update_time
#更新
job.update_job()
#列表
jobs = Job.get_alljobs(query_name,query_project_name,query_server_host,query_user,login_user)
query_dict = {
'query_name':query_name,
'query_project_name':query_project_name,
'query_server_host':query_server_host,
'query_user':query_user,
'session_id':session_id,
'login_user':login_user
}
logging.info('[%s] update job [%s]' % (login_user,job.name))
self.render('list.html',title=title,jobs=jobs,query_dict=query_dict)
