定时任务配合es实现高可用查询

定时任务统计查询

     1.无论是定时任务还是es本身都有可能被异常停止服务

         当服务恢复正常的时候进行补偿查询  把服务中断期间的数据进行统计查询

     2.单次统计也可能由于程序本身和网络连接问题发生异常

         把异常的统计时间段存入异常表 每次查询都从表中取出所有异常查询点 不断查询直到返回成功

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import datetime
import time
import sys
import os
import json
sys.path.append('../')

from common import esConfig
from common import timetools
from common import calastTime
from common.loggerConfig import loggerConfig
from common import constConfig
from common import dbtools


indexname_prefix="sms-service-"
logfile="../log/sms_service.log"
logpath=os.path.abspath(logfile)
log=loggerConfig("smsService",logpath)
logger=log.create_logger()



body={"aggs":{"3":{"terms":{"field":"APP_ID","size":5000,"order":{"1":"desc"}},"aggs":{"1":{"cardinality":{"field":"RECORD_ID"}},"4":{"terms":{"field":"code","size":5000,"order":{"1":"desc"}},"aggs":{"1":{"cardinality":{"field":"RECORD_ID"}},"5":{"terms":{"field":"CHANNEL_ID","size":5000,"order":{"1":"desc"}},"aggs":{"1":{"cardinality":{"field":"RECORD_ID"}},"6":{"terms":{"field":"SWJG_DM","size":5000,"order":{"1":"desc"}},"aggs":{"1":{"cardinality":{"field":"RECORD_ID"}},"7":{"terms":{"field":"STATUS","size":5000,"order":{"1":"desc"}},"aggs":{"1":{"cardinality":{"field":"RECORD_ID"}},"8":{"date_histogram":{"field":"mydate","interval":"1h","time_zone":"Asia/Shanghai","min_doc_count":1},"aggs":{"1":{"cardinality":{"field":"RECORD_ID"}}}}}}}}}}}}}}},"size":0,"stored_fields":["*"],"docvalue_fields":[{"field":"@timestamp","format":"date_time"},{"field":"mydate","format":"date_time"}],"query":{"bool":{"must":[{"match_phrase":{"metricsName":{"query":"SmsService"}}},{"range":{"mydate":{"gte":0,"lte":0,"format":"epoch_millis"}}}]}},"timeout":"30000ms"}

def handle_res(res,index_date):
    outlist=[]
    dnow=datetime.datetime.now().strftime('%Y-%m-%d %H')+":00:00"
    for i3 in res["aggregations"]["3"]["buckets"]:
      for i4 in i3["4"]["buckets"]:
        for i5 in i4["5"]["buckets"]:
          for i6 in i5["6"]["buckets"]:
            for i7 in i6["7"]["buckets"]:
                for i8 in i7["8"]["buckets"]:
                    timestr = i8["key_as_string"][:-6]
                    newtime = timetools.formartTime(timestr)
                    outlist.append({"appId":i3["key"],"code":i4["key"],"channelId":i5["key"],"swjdDm":i6["key"],"status":i7["key"],"fszl":i8["doc_count"],"fsl":i8["1"]["value"],"createTime":newtime,"statisticalTime":dnow})
    indexname = indexname_prefix + index_date
    if esConfig.es.indices.exists(index=indexname):
        logger.info("%s 索引已经存在" % (indexname), extra={'result': '', "querytime": ""})
    else:
        esConfig.es.indices.create(index=indexname)
        logger.info("%s 索引被成功创建" % (indexname), extra={'result': '', "querytime": ""})

    for data in outlist:
        res2=esConfig.es.index(index=indexname, doc_type="doc", body=data)
        logger.info("%s被成功添加到%s索引当中" %(json.dumps(data),indexname),extra={'result': 'data', "querytime": ""})

def handler_hours(hours):
    retryCount = constConfig.MAX_RETRY
    cald=calastTime.CalcDaysHours(beginTime=None,endTime=None,logger=logger)
    for i in hours:
        while retryCount >= 1:
            try:
                index_date = i.strftime("%Y-%m-%d")
                index_hour = i.strftime("%H")
                res = cald.get_data_from_index(i,body)
                handle_res(res, index_date)
                # 日志中记录最后一次成功统计的时间 最后统计成功时间不一定是当前时间
                logger.info("成功统计%d条记录" % (res['hits']['total']), extra={'result': 'success', "querytime": i})
                successtask = dbtools.Task(name='smsService',time=i,status=1)
                dbtools.TaskOps().delete_FailTask(name='smsService',time=i)
                dbtools.TaskOps().Add_Task(successtask)
                time.sleep(10)
                break
            except Exception as e:
                retryCount -= 1
                if retryCount > 0:
                    logger.error("%d次统计发生异常%s,还剩%d次重试" % (retryCount + 1, str(e), retryCount),
                                 extra={'result': 'fail', "querytime": i})
                    time.sleep(60)
                    continue
                else:
                    failtask = dbtools.Task(name='smsService', time=i, status=0)
                    dbtools.TaskOps().delete_FailTask(name='smsService',time=i)
                    dbtools.TaskOps().Add_Task(failtask)
                    logger.error("%d次统计全部失败,结束异常统计" % (constConfig.MAX_RETRY), extra={'result': 'fail', "querytime": i})
                    retryCount = constConfig.MAX_RETRY
                    break

pro=calastTime.ProccessData("smsService",body,logger)
logger.info("开始统计补偿任务",extra={'result': 'start', "querytime": ""})
lost_hours=pro.run_buchang()
handler_hours(lost_hours)
logger.info("补偿任务统计结束",extra={'result': '', "querytime": ""})
logger.info("开始重新统计失败任务",extra={'result': '', "querytime": ""})
error_hours=pro.run_failtask()
handler_hours(error_hours)
logger.info("失败任务统计结束",extra={'result': 'end', "querytime": ""})

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import datetime
import os
from common import timetools
from common import esConfig
from common import dbtools


class CalcTime:
    def __init__(self,name,logger):
        self.name=name
        self.logger=logger

    def getDifferHours(self):
        # logf=LogFileRead(self.filename,self.logger)
        lastSuccessTime=Querydb().getLastSucessTask(self.name).strftime('%Y-%m-%d %H:%M:%S')
        lastSuccessTime=timetools.logtime_to_datetime(lastSuccessTime)
        #最后执行成功的时间 说明下一个小时的统计没有被成功执行 从下一个小时开始计算
        #开始时间=最后执行成功时间+1h-3h 这段时间开始范围内的数据都没有被统计
        nowTime=datetime.datetime.now()
        #结束时间=当前时间-3h
        #补偿异常期间内所有的数据
        self.logger.info("上次统计的成功时间段是%s" % (lastSuccessTime.strftime('%Y-%m-%d %H')),extra={'result': '', "querytime": ""})
        #说明当前时间-3h之前的数据都被成功统计
        self.logger.info("当前时间%s" %(nowTime),extra={'result': '', "querytime": ""})
        #本次的实际统计开始时间是从上次成功统计后的1h开始
        real_startTime=(lastSuccessTime+datetime.timedelta(hours=1)).replace(minute=0,second=0).strftime('%Y-%m-%d %H:%M:%S.%f')[:-3]
        theory_startTime=(nowTime+datetime.timedelta(hours=-3)).replace(minute=0,second=0).strftime('%Y-%m-%d %H:%M:%S.%f')[:-3]
        endTime = (nowTime + datetime.timedelta(hours=-3)).replace(minute=59, second=59).strftime('%Y-%m-%d %H:%M:%S.%f')[:-3]
        self.logger.info("补偿时间范围%s-%s" %(real_startTime,endTime),extra={'result': '', "querytime": ""})
        self.logger.info("此次理论查询时间范围%s-%s" % (theory_startTime, endTime),extra={'result': '', "querytime": ""})
        self.logger.info("此次实际查询时间范围%s-%s" %(real_startTime,endTime),extra={'result': '', "querytime": ""})
        return real_startTime,endTime

class LogFileRead:

    def __init__(self,filename,logger):
        self.filename=filename
        self.logger=logger

    def getLastSuccessTime(self):
        # res = os.popen("tac %s | grep 'success'| head -n 1" % (self.filename))
        res = os.popen("tac %s | sed -n '/^success/p'| head -n 1" % (self.filename))
        # res = os.popen("sed -n '/^success/p' %s | head -n 1" %(self.filename))
        con = None
        for re in res.readlines():
            con = re
        if con:
            cons = con.split("_")
            self.logger.info("%s日志记录中最近的成功统计时间是%s" %(self.filename,cons[2]),extra={'result': '',"querytime":""})
            return cons[2]
        else:
            self.logger.info("%s日志文件中读取不到任何成功执行记录" %(self.filename),extra={'result': '',"querytime":""})
            #如果日志文件中读取不到任何记录  则说明统计的是当前时间-3h的记录
            return (datetime.datetime.now()+datetime.timedelta(hours=-3)).strftime('%Y-%m-%d %H:%M:%S')+",000"

class CalcDaysHours:

    def __init__(self,beginTime,endTime,logger):
        self.beginTime=beginTime
        self.endTime=endTime
        self.logger=logger

    def calcDays(self):
        date1 = datetime.datetime.strptime(self.beginTime[:-13], "%Y-%m-%d")
        date2 = datetime.datetime.strptime(self.endTime[:-13], "%Y-%m-%d")
        date_list = []
        while date1 <= date2:
            date_str = date1
            date_list.append(date_str)
            date1 += datetime.timedelta(days=1)
        return date_list

    def calcHours(self):
        date1 = datetime.datetime.strptime(self.beginTime[:-10], "%Y-%m-%d %H")
        date2 = datetime.datetime.strptime(self.endTime[:-10], "%Y-%m-%d %H")
        hour_list=[]
        while date1 <= date2:
            date_str = date1
            hour_list.append(date_str)
            date1 += datetime.timedelta(hours=1)
        return hour_list

    def getIndexes(self):
        datelist=self.calcDays()
        indexes=[]
        for i in datelist:
            indexes.append(esConfig.index_prefix+i.strftime("%Y-%m-%d"))
        self.logger.info("此次查询的原始索引为%s" %(indexes),extra={'result': '',"querytime":""})
        return indexes

    def getOriginalIndex(self,startTime):
        indexes=[]
        index1=esConfig.index_prefix+startTime.strftime("%Y-%m-%d")
        #es服务异常恢复后会把n天前的数据导入到当前日期的索引中 而不会导入对应日期的索引中
        index2=esConfig.index_prefix+datetime.datetime.now().strftime("%Y-%m-%d")
        indexes.append(index1)
        indexes.append(index2)
        self.logger.info("此次查询的原始索引为%s" % (indexes), extra={'result': '', "querytime": ""})
        return indexes

    def get_data_from_index(self,startTime,body):
        sTime=startTime.strftime('%Y-%m-%d %H:%M:%S')
        eTime=startTime.replace(minute=59,second=59).strftime('%Y-%m-%d %H:%M:%S')
        stime = str(timetools.strtime_to_timestamp(sTime))
        etime = str(timetools.strtime_to_timestamp(eTime))
        self.logger.info("本次统计的时间范围是%s-%s|%s-%s" %(sTime,eTime,stime,etime),extra={'result': '',"querytime":""})
        bodydate=body.get("query").get("bool").get("must")[1].get("range").get("mydate")
        bodydate["gte"]=stime
        bodydate["lte"]=etime
        body.get("query").get("bool").get("must")[1].get("range")["mydate"]=bodydate

        res = esConfig.es.search(body=body,index=self.getOriginalIndex(startTime))
        return res

class Querydb:

    def getFailTasks(self,name):
        res=dbtools.TaskOps().query_FailTask(name)
        return res

    def getLastSucessTask(self,name):
        res =dbtools.TaskOps().query_SucTask(name)
        nowTime = datetime.datetime.now()
        if res:
            return res.time
        else:
            return nowTime+datetime.timedelta(hours=-4)

class ProccessData:

    def __init__(self,name,body,logger):
        self.name=name
        self.body=body
        self.logger=logger


    #执行补偿task
    def run_buchang(self):
        cal=CalcTime(self.name,self.logger)
        startTime,endTime=cal.getDifferHours()
        cday=CalcDaysHours(startTime,endTime,self.logger)
        return cday.calcHours()


    #执行统计失败的任务
    def run_failtask(self):
        res = Querydb().getFailTasks(self.name)
        hours_list=[]
        for re in res:
            # re.time=re.time.strftime('%Y-%m-%d %H')
            hours_list.append(re.time)
        return hours_list

#!/usr/bin/env python
# -*- coding: utf-8 -*-

from sqlalchemy import create_engine,Column,Integer,String,DateTime
from sqlalchemy.orm import sessionmaker
from sqlalchemy.ext.declarative import declarative_base
import datetime
import os

parent_dir=os.path.abspath(".")
if parent_dir.endswith("auto-task"):
    pass
else:
    parent_dir=os.path.abspath("../")

# engine = create_engine('sqlite:///./db/estasks.db',echo=True) #echo要求打印sql语句等调试信息
engine = create_engine('sqlite:///%s/estasks.db' %(parent_dir+"/db"))
session_maker = sessionmaker(bind=engine)
session = session_maker()
Base = declarative_base()

class Task(Base):
  __tablename__ = 'Task'
  id = Column('TaskID',Integer,primary_key=True)
  name = Column('TaskType',String(100),nullable=False)
  status=Column('status',Integer,nullable=False)
  time = Column('TaskTime',DateTime,nullable=False)
  addtime=Column('TaskaddTime',DateTime,default=datetime.datetime.now)

  def __repr__(self):
    return '<Task(id:%s, name:%s, time:%s)>' % (self.id,self.name,self.time)

class TaskRecord(Base):
    __tablename__ = 'TaskRecord'
    id = Column('TaskID',Integer,primary_key=True)
    name = Column('TaskName',String(100),nullable=False)
    failtime = Column('TaskTime',DateTime,nullable=False)
    fixtime=Column('TaskSuccessTime',DateTime,default=datetime.datetime.now)

    def __repr__(self):
        return '<TaskRecord(id:%s,name:%s,time:%s,fixtime:%s)>' %(self.id,self.name,self.time,self.fixtime)

class TaskOps:

    def Add_Task(self,task):
        session.add(task)
        session.commit()

    def query_FailTask(self,name):
        res=session.query(Task).filter(Task.name==name,Task.status==0).all()
        return res

    def query_SucTask(self,name):
        res=session.query(Task).filter(Task.name==name,Task.status==1).order_by(Task.time.desc()).first()
        return res

    def delete_FailTask(self,name,time):
        session.query(Task).filter(Task.name==name,Task.time==time,Task.status==0).delete()
        session.commit()

class TaskRecordOps:

    def Add_TaskRecord(self,task):
        session.add(task)
        session.commit()

    def query_TaskRecord(self):
        res=session.query(TaskRecord).all()
        return res

Base.metadata.create_all(engine) #若存在表则不做,不存在则创建

 

定时任务设置

  添加定时任务 : crontab -e

  基本格式 :
   *　　*　　*　　*　　*　　command
  分　 时　 日　 月　 周　 命令
  解 释:
   第1列表示分钟1～59 每分钟用或者 /1表示
   第2列表示小时1～23（0表示0点）
   第3列表示日期1～31
   第4列表示月份1～12
   第5列标识号星期0～6（0表示星期天）
    第6列要运行的命令

[root]# crontab -l
    */60 * * * * sh /root/auto-task/shell/hostcron.sh           每个整点时间点执行一次
     1 0 * * * sh /root/auto-task/shell/hostcronday.sh        每天的0点过1分执行一次