2016-10-07 09:41
浏览 240


Try subscribing ActiveMQ(Apollo) using Go-Stomp, but I am having read timeout error. My app should be alive 24 hours per day to process incoming message.

Question :

  1. Is there a way to keep the subcription although there is no more message exist in the queue? Trying to put ConnOpt.HeartBeat also does not seems to work
  2. Why after the readTimeout, it seems that I still accept one more message ?

Below is my steps :

  • I put 1000 messages for testing in the inputQueue
  • Run a subscriber, code provided below
  • Subscriber finished reading 1000 messages After 2-3 seconds, saw error " 2016/10/07 17:12:44 Subscription 1: /queue/hflc-in: ERROR message:read timeout".
  • Put another 1000 messages, but it seems the subscription is already down, therefore no message is not being processed

My Code :

   serverAddr   = flag.String("server", "", "STOMP server    endpoint")
   messageCount = flag.Int("count", 10, "Number of messages to send/receive")
   inputQ       = flag.String("inputq", "/queue/hflc-in", "Input queue")

var options []func(*stomp.Conn) error = []func(*stomp.Conn) error{
   stomp.ConnOpt.Login("userid", "userpassword"),
   stomp.ConnOpt.HeartBeat(360*time.Second, 360*time.Second), // I put this but seems no impact

func main() {
  jobschan := make(chan bean.Request, 10)
  //my init setup
  go getInput(1, jobschan)

func getInput(id int, jobschan chan bean.Request) {
   conn, err := stomp.Dial("tcp", *serverAddr, options...)

   if err != nil {
      println("cannot connect to server", err.Error())
   fmt.Printf("Connected %v 
", id)

   sub, err := conn.Subscribe(*inputQ, stomp.AckClient)
   if err != nil {
     println("cannot subscribe to", *inputQ, err.Error())

   fmt.Printf("Subscribed %v 
", id)
   var messageCount int
   for {
    msg := <-sub.C
    //expectedText := fmt.Sprintf("Message #%d", i)
    if msg != nil {

        actualText := string(msg.Body)

        var req bean.Request
        if actualText != "SHUTDOWN" {
            messageCount = messageCount + 1
            var err2 = easyjson.Unmarshal([]byte(actualText), &req)
            if err2 != nil {
                log.Error("Unable unmarshall", zap.Error(err))
                println("message body %v", msg.Body) // what is [0/0]0x0 ?
            } else {
                fmt.Printf("Subscriber %v received message, count %v 
  ", id, messageCount)
                jobschan <- req
        } else {
            logchan <- "got some issue"

Error :

2016/10/07 17:12:44 Subscription 1: /queue/hflc-in: ERROR message:read timeout
[E] 2016-10-07T09:12:44Z Unable unmarshall
message body %v [0/0]0x0

  • 写回答
  • 好问题 提建议
  • 关注问题
  • 收藏
  • 邀请回答

1条回答 默认 最新

  • douyakao5308 2016-10-10 03:47

    Solved by adding these lines :

    in Apollo, noticed that the queue deleted after being empty after several secs, so put auto_delete_after to several hours in apollo.xml , eg :

    <queue id="hflc-in" dlq="dlq-in" nak_limit="3" auto_delete_after="7200"/>
    <queue id="hflc-log" dlq="dlq-log" nak_limit="3" auto_delete_after="7200"/>
    <queue id="hflc-out" dlq="dlq-out" nak_limit="3" auto_delete_after="7200"/>

    in Go, noticed that go-stomp will just give up right after it can't found any message in queue, so in the conn options, add HeartBeat Error

    var options []func(*stomp.Conn) error = []func(*stomp.Conn) error{
       //.... original configuration
       stomp.ConnOpt.HeartBeatError(360 * time.Second),

    However still confused about the question no 2.

    解决 无用
    打赏 举报

相关推荐 更多相似问题