package main import ( "encoding/json" "esindex/config" "esindex/oss" "fmt" "github.com/robfig/cron" "go.uber.org/zap" "io/ioutil" util "jygit.jydev.jianyu360.cn/data_processing/common_utils" "jygit.jydev.jianyu360.cn/data_processing/common_utils/elastic" "jygit.jydev.jianyu360.cn/data_processing/common_utils/log" "jygit.jydev.jianyu360.cn/data_processing/common_utils/mongodb" "jygit.jydev.jianyu360.cn/data_processing/common_utils/mysqldb" "jygit.jydev.jianyu360.cn/data_processing/common_utils/udp" "net" "net/http" "sync" "time" ) var ( MgoB *mongodb.MongodbSim MgoP *mongodb.MongodbSim MgoQ *mongodb.MongodbSim MgoS *mongodb.MongodbSim Mysql *mysqldb.Mysql Es, Es1 *elastic.Elastic UdpClient udp.UdpClient UdpTaskMap = &sync.Map{} JyUdpAddr *net.UDPAddr EsBulkSize = 100 // es批量保存大小 updateBiddingPool = make(chan []map[string]interface{}, 5000) //更新bingding数据 updateBiddingSp = make(chan bool, 5) saveEsPool = make(chan map[string]interface{}, 5000) //保存binding数据到es saveEsSp = make(chan bool, 5) saveProjectEsPool = make(chan map[string]interface{}, 5000) //保存project数据到es saveProjectSp = make(chan bool, 5) saveEsAllPool = make(chan map[string]interface{}, 5000) //存储单机版es,爬虫采集判重使用 saveEsAllSp = make(chan bool, 5) saveErrBidPool = make(chan map[string]interface{}, 5000) saveBidSp = make(chan bool, 5) //detailLength = 50000 // es保存detail长度 fileLength = 200000 // es保存附件文本长度,大于20万 时做一个日志记录 //pscopeLength = 32766 // projectscope长度 specialNames = make(map[string]bool, 0) //存储181 凭安提供的特企,爬虫采购单位 responselock sync.Mutex BiddingLastNodeResponse int64 //上次节点接受数据时间 ProjectLastNodeResponse int64 //上次节点接受数据时间 ) func init() { config.Init("./common.toml") InitLog() InitMgo() InitEs() InitField() InitEsBiddingField() oss.InitOss() verifyESFields() //检测es 字段类型 JyUdpAddr = &net.UDPAddr{ IP: net.ParseIP(config.Conf.Udp.JyAddr), Port: util.IntAll(config.Conf.Udp.JyPort), } BiddingLastNodeResponse = time.Now().Unix() ProjectLastNodeResponse = time.Now().Unix() log.Info("init success") } func main() { go checkMapJob() //udp 发送邮件 go task_index() //定时同步更新winner_enterprise、buyer_enterprise ES索引;这个功能很少变动,几乎不需要维护 go UpdateBidding() //更新bidding表数据 go SaveEsMethod() go SaveAllEsMethod() go SaveProjectEs() go SaveBidErr() go LastUdpJob() //监听半小时内有无数据 UdpClient = udp.UdpClient{Local: config.Conf.Udp.LocPort, BufSize: 1024} UdpClient.Listen(processUdpMsg) log.Info("Udp服务监听", zap.String("port:", config.Conf.Udp.LocPort)) ch := make(chan bool, 1) <-ch } var pool = make(chan bool, 20) func processUdpMsg(act byte, data []byte, ra *net.UDPAddr) { switch act { case udp.OP_TYPE_DATA: var mapInfo map[string]interface{} err := json.Unmarshal(data, &mapInfo) log.Info("processUdpMsg", zap.Any("mapInfo:", mapInfo)) if err != nil { UdpClient.WriteUdp([]byte("err:"+err.Error()), udp.OP_NOOP, ra) } else if mapInfo != nil { key, _ := mapInfo["key"].(string) if key == "" { key = "udpok" } go UdpClient.WriteUdp([]byte(key), udp.OP_NOOP, ra) tasktype, _ := mapInfo["stype"].(string) switch tasktype { case "index-by-id": //单个索引 pool <- true go func() { defer func() { <-pool }() biddingTaskById(mapInfo) }() case "bidding": BiddingLastNodeResponse = time.Now().Unix() pool <- true go func() { defer func() { <-pool }() biddingTask(mapInfo) }() case "biddingall": //补充存量数据 pool <- true go func() { defer func() { <-pool }() biddingAllTask(mapInfo) }() case "bidding_all_data": //根据biddingall配置文件,存量迁移数据 pool <- true go func() { defer func() { <-pool }() biddingAllDataTask() }() case "bidding_history": pool <- true go func() { defer func() { <-pool }() biddingTask(mapInfo) }() case "project": ProjectLastNodeResponse = time.Now().Unix() pool <- true go func() { defer func() { <-pool }() projectTask(data, mapInfo) }() case "project_all_data": //存量 projectset 数据 pool <- true go func() { defer func() { <-pool }() projectAllData() }() case "biddingdata": //es 单机版,采集判重 pool <- true go func() { defer func() { <-pool }() biddingDataTask(data, mapInfo) }() case "biddingdelbyextracttype": //根据bidding表extracttype=-1,删除es中重复数据 pool <- true go func() { defer func() { <-pool }() biddingDelByExtracttype(data, mapInfo) }() case "buyer_once": // 采购单位昨天增量数据 pool <- true go func() { defer func() { <-pool }() buyerOnce() }() default: pool <- true go func() { defer func() { <-pool }() log.Info("err", zap.Any("mapInfo", mapInfo)) }() } } case udp.OP_NOOP: //下个节点回应 ok := string(data) if ok != "" { log.Info("udp re", zap.String("data:", ok)) UdpTaskMap.Delete(ok) } } } func task_index() { c := cron.New() _ = c.AddFunc("0 0 0 * * ?", func() { task_winneres() }) //每天凌晨执行一次winner生索引 _ = c.AddFunc("0 0 1 * * ?", func() { task_buyeres() }) //每天1点执行一次buyer生索引 c.Start() } func task_winneres() { log.Info("定时任务,winneres") winnerEsTaskOnce() } func task_buyeres() { log.Info("定时任务,buyeres") buyerOnce() } type UdpNode struct { data []byte addr *net.UDPAddr timestamp int64 retry int } //UpdateBidding 更新bidding表数据 func UpdateBidding() { arru := make([][]map[string]interface{}, 200) indexu := 0 for { select { case v := <-updateBiddingPool: arru[indexu] = v indexu++ if indexu == 200 { updateBiddingSp <- true go func(arru [][]map[string]interface{}) { defer func() { <-updateBiddingSp }() MgoB.UpdateBulk(config.Conf.DB.MongoB.Coll, arru...) }(arru) arru = make([][]map[string]interface{}, 200) indexu = 0 } case <-time.After(1000 * time.Millisecond): if indexu > 0 { updateBiddingSp <- true go func(arru [][]map[string]interface{}) { defer func() { <-updateBiddingSp }() MgoB.UpdateBulk(config.Conf.DB.MongoB.Coll, arru...) }(arru[:indexu]) arru = make([][]map[string]interface{}, 200) indexu = 0 } } } } //SaveBidErr 记录错误信息,暂时记录 附件过长的 func SaveBidErr() { arru := make([]map[string]interface{}, 200) indexu := 0 for { select { case v := <-saveErrBidPool: arru[indexu] = v indexu++ if indexu == 200 { saveBidSp <- true go func(arru []map[string]interface{}) { defer func() { <-saveBidSp }() MgoB.SaveBulk("bidding_es_err_record", arru...) }(arru) arru = make([]map[string]interface{}, 200) indexu = 0 } case <-time.After(1000 * time.Millisecond): if indexu > 0 { saveBidSp <- true go func(arru []map[string]interface{}) { defer func() { <-saveBidSp }() MgoB.SaveBulk("bidding_es_err_record", arru...) }(arru[:indexu]) arru = make([]map[string]interface{}, 200) indexu = 0 } } } } //SaveEsMethod 保存到es func SaveEsMethod() { arru := make([]map[string]interface{}, EsBulkSize) indexu := 0 for { select { case v := <-saveEsPool: arru[indexu] = v indexu++ if indexu == EsBulkSize { saveEsSp <- true go func(arru []map[string]interface{}) { defer func() { <-saveEsSp }() Es.BulkSave(config.Conf.DB.Es.IndexB, arru) if config.Conf.DB.Es.IndexTmp != "" { Es.BulkSave(config.Conf.DB.Es.IndexTmp, arru) } }(arru) arru = make([]map[string]interface{}, EsBulkSize) indexu = 0 } case <-time.After(1000 * time.Millisecond): if indexu > 0 { saveEsSp <- true go func(arru []map[string]interface{}) { defer func() { <-saveEsSp }() Es.BulkSave(config.Conf.DB.Es.IndexB, arru) if config.Conf.DB.Es.IndexTmp != "" { Es.BulkSave(config.Conf.DB.Es.IndexTmp, arru) } }(arru[:indexu]) arru = make([]map[string]interface{}, EsBulkSize) indexu = 0 } } } } func SaveAllEsMethod() { arru := make([]map[string]interface{}, EsBulkSize) indexu := 0 for { select { case v := <-saveEsAllPool: arru[indexu] = v indexu++ if indexu == EsBulkSize { saveEsAllSp <- true go func(arru []map[string]interface{}) { defer func() { <-saveEsAllSp }() Es1.BulkSave("biddingall", arru) }(arru) arru = make([]map[string]interface{}, EsBulkSize) indexu = 0 } case <-time.After(1000 * time.Millisecond): if indexu > 0 { saveEsAllSp <- true go func(arru []map[string]interface{}) { defer func() { <-saveEsAllSp }() Es1.BulkSave("biddingall", arru) }(arru[:indexu]) arru = make([]map[string]interface{}, EsBulkSize) indexu = 0 } } } } func SaveProjectEs() { arru := make([]map[string]interface{}, EsBulkSize) indexu := 0 for { select { case v := <-saveProjectEsPool: arru[indexu] = v indexu++ if indexu == EsBulkSize { saveProjectSp <- true go func(arru []map[string]interface{}) { defer func() { <-saveProjectSp }() Es.BulkSave(config.Conf.DB.Es.IndexP, arru) }(arru) arru = make([]map[string]interface{}, EsBulkSize) indexu = 0 } case <-time.After(1000 * time.Millisecond): if indexu > 0 { saveProjectSp <- true go func(arru []map[string]interface{}) { defer func() { <-saveProjectSp }() Es.BulkSave(config.Conf.DB.Es.IndexP, arru) }(arru[:indexu]) arru = make([]map[string]interface{}, EsBulkSize) indexu = 0 } } } } func checkMapJob() { if config.Conf.Mail.Send { log.Info("checkMapJob", zap.String("to:", config.Conf.Mail.To)) for { UdpTaskMap.Range(func(k, v interface{}) bool { now := time.Now().Unix() node, _ := v.(*UdpNode) if now-node.timestamp > 120 { node.retry++ if node.retry > 5 { UdpTaskMap.Delete(k) res, err := http.Get(fmt.Sprintf("%s?to=%s&title=%s&body=%s", config.Conf.Mail.Api, config.Conf.Mail.To, "field-sync-send-fail", k.(string))) if err == nil { defer res.Body.Close() read, err := ioutil.ReadAll(res.Body) log.Info("send mail ...", zap.String("r:", string(read)), zap.Any("err:", err)) } } else { log.Info("udp重发", zap.Any("k:", k)) UdpClient.WriteUdp(node.data, udp.OP_TYPE_DATA, node.addr) } } else if now-node.timestamp > 10 { log.Info("udp任务超时中..", zap.Any("k:", k)) } return true }) time.Sleep(60 * time.Second) } } } func task() { sess := MgoB.GetMgoConn() defer MgoB.DestoryMongoConn(sess) ch := make(chan bool, 10) wg := &sync.WaitGroup{} query := sess.DB("qfw").C("result_replace_repair_log").Find(nil).Iter() count := 0 for tmp := make(map[string]interface{}); query.Next(tmp); count++ { if count%1000 == 0 { util.Debug("current ---", count) } ch <- true wg.Add(1) go func(tmp map[string]interface{}) { defer func() { <-ch wg.Done() }() if id := util.ObjToString(tmp["replace_id"]); mongodb.IsObjectIdHex(id) { biddingTaskById(map[string]interface{}{"infoid": id, "stype": "bidding"}) } }(tmp) tmp = make(map[string]interface{}) } wg.Wait() util.Debug("over ---", count) } //LastUdpJob 处理UDP 没有接受数据 func LastUdpJob() { for { responselock.Lock() if time.Now().Unix()-BiddingLastNodeResponse >= 1800 { BiddingLastNodeResponse = time.Now().Unix() //重置时间 sendErrMailApi("索引程序异常", fmt.Sprintf("半小时左右 无bidding据进入 ...相关人员检查...")) } if time.Now().Unix()-ProjectLastNodeResponse >= 1800 { ProjectLastNodeResponse = time.Now().Unix() //重置时间 sendErrMailApi("索引程序异常", fmt.Sprintf("半小时左右 无project数据进入 ...相关人员检查...")) } responselock.Unlock() time.Sleep(300 * time.Second) } } //sendErrMailApi 发送邮件 func sendErrMailApi(title, body string) { var tomail, api string if config.Conf.Mail.Send { tomail = config.Conf.Mail.To api = config.Conf.Mail.Api } log.Info("sendErrMailApi", zap.Any(tomail, api)) res, err := http.Get(fmt.Sprintf("%s?to=%s&title=%s&body=%s", api, tomail, title, body)) if err == nil { defer res.Body.Close() read, err := ioutil.ReadAll(res.Body) if err != nil { log.Info("邮件发送成功", zap.String("read", string(read))) } } else { log.Info("sendErrMailApi", zap.String("邮件发送失败", err.Error())) } }