Anyone still maintaining this?

     ticker            = time.NewTicker(rc.getKeepAliveTimeout())

...

if err := rc.writeControlPingMessage(); err != nil {
    log.Println(err)
}

<-ticker.C
if time.Since(keepAliveResponse.getLastResponse()) > rc.getKeepAliveTimeout() {
    rc.CloseAndReconnect()
    return
}

Shouldn't time.Since(keepAliveResponse.getLastResponse()) always be greater than keepAliveTimeout because it's waiting on the timer which is set to keepAliveTimeout???

I'm encountering the situation where the time since the last response is ever so slightly longer than keepAliveTimeout.

Hi,

Is there a way for ignoring SSL/TLS Certificats error ? My websocket dev server use TLS with a self signed certificate.

Note that after retrying after 2 seconds, recws enter into a loop for 2 seconds before closing ws, see #38

https://github.com/gorilla/websocket/releases/tag/v1.4.1

KeepAliveInterval should be less than KeepAliveTimeout (ideally less than 1/2). so the code should be change from:

to:
ticker = time.NewTicker(rc.getKeepAliveTimeout() / 2)

This will solve occasionally keep alive timeout problem.

// CloseAndReconnect will try to reconnect.
func (rc *RecConn) CloseAndReconnect() {
	rc.Close()
	go rc.connect()
}

the Close() function send to the close chan

rc.close <- true

connect() will return if it receives from close chan

func (rc *RecConn) connect() {
	b := rc.getBackoff()
	rand.Seed(time.Now().UTC().UnixNano())

	for {
		select {
		case <-rc.close:
			return
....

in the end calling connect() right after Close() will not work.

Am I missing something here or it is a bug?

In got an infinite loop once a day, after profiling the infinite loop is here at around line 375.

		for {
			if !rc.IsConnected() {
				continue
			}

It doesn't go away until I restart my application.

version: v1.3.1
I am also using

&recws.RecConn{
		KeepAliveTimeout: 60 * time.Second,
	}

It kept reconnecting after the first connection reset, I also couldn't receive new messages afterwards.

after socket is connected, restarted websocket server.

version 1.2.1

2020/12/30 14:33:38 read tcp [::1]:61619->[::1]:3000: read: connection reset by peer
2020/12/30 14:33:38 Dial: will try again in 2s seconds.
2020/12/30 14:33:40 dial tcp [::1]:3000: connect: connection refused
2020/12/30 14:33:40 Dial: will try again in 2.891027974s seconds.
2020/12/30 14:33:43 dial tcp [::1]:3000: connect: connection refused
2020/12/30 14:33:43 Dial: will try again in 2.909185715s seconds.
2020/12/30 14:33:46 dial tcp [::1]:3000: connect: connection refused
2020/12/30 14:33:46 Dial: will try again in 2.856300582s seconds.
2020/12/30 14:33:48 dial tcp [::1]:3000: connect: connection refused
2020/12/30 14:33:48 Dial: will try again in 7.599538362s seconds.

retry and connecting is succeed, but

version 1.2.2

not retry and just fail error message

I need to send a login message when the connection is reestablished. How do I do this?

Code (from example, but use ws rather than wss)

The server is a raw ws nodejs app, working perfectly with a golang client using gorilla websocket.

package main

import (
	"context"
	"log"
	"time"

	"github.com/recws-org/recws"
)

func main() {
	ctx, cancel := context.WithCancel(context.Background())
	ws := recws.RecConn{
		KeepAliveTimeout: 10 * time.Second,
	}
	ws.Dial("ws://192.168.56.101", nil)

	go func() {
		time.Sleep(2 * time.Second)
		cancel()
	}()

	for {
		select {
		case <-ctx.Done():
			go ws.Close()
			log.Printf("Websocket closed %s", ws.GetURL())
			return
		default:
			if !ws.IsConnected() {
				log.Printf("Websocket disconnected %s", ws.GetURL())
				continue
			}

			if err := ws.WriteMessage(1, []byte("Incoming")); err != nil {
				log.Printf("Error: WriteMessage %s", ws.GetURL())
				return
			}

			_, message, err := ws.ReadMessage()
			if err != nil {
				log.Printf("Error: ReadMessage %s", ws.GetURL())
				return
			}

			log.Printf("Success: %s", message)
		}
	}
}

14:25:30 => 14:25:32 = OK, waiting
14:25:32 => 14:25:34 = disconnected loop ?
14:25:34 = closing

2021/03/17 14:25:30 websocket: bad handshake
2021/03/17 14:25:30 Dial: will try again in 2s seconds.
2021/03/17 14:25:32 Websocket disconnected ws://192.168.56.101
2021/03/17 14:25:32 Websocket disconnected ws://192.168.56.101
2021/03/17 14:25:32 Websocket disconnected ws://192.168.56.101
.....
2021/03/17 14:25:34 Websocket disconnected ws://192.168.56.101
2021/03/17 14:25:34 Websocket disconnected ws://192.168.56.101
2021/03/17 14:25:34 Websocket disconnected ws://192.168.56.101
2021/03/17 14:25:34 Websocket disconnected ws://192.168.56.101
2021/03/17 14:25:34 Websocket disconnected ws://192.168.56.101
2021/03/17 14:25:34 Websocket disconnected ws://192.168.56.101
2021/03/17 14:25:34 Websocket closed ws://192.168.56.101

Hello,

I don't understand why the SubscribeHandler is executed only when getNonVerbose is false.
Is mean that we can't use this feature in NonVerbose mode. Makes no sense for me.
Is there any reason ?

Hi,

I encountered the following error in my application:

panic: runtime error: invalid memory address or nil pointer dereference[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x7d96d0]goroutine 59856 [running]:github.com/gorilla/websocket.(*Conn).SetPongHandler(0x0, 0xc003712960) /go/pkg/mod/github.com/gorilla/[email protected]/conn.go:1160 +0x20github.com/recws-org/recws.(*RecConn).keepAlive(0x1399a00) /go/pkg/mod/github.com/recws-org/[email protected]/recws.go:365 +0xc5github.com/recws-org/recws.(*RecConn).connect(0x1399a00) /go/pkg/mod/github.com/recws-org/[email protected]/recws.go:422 +0x3cacreated by github.com/recws-org/recws.(*RecConn).CloseAndReconnect /go/pkg/mod/github.com/recws-org/[email protected]/recws.go:64 +0x4d

Looks to happen because the error on this line is not being handled, but I might be wrong:

Any ideas how to go about this?

Thank you.

Currently I do it this way, but find it rather unclean:

for {
  if !ws.IsConnected() {
    time.Sleep(time.Second * 2)
    continue
  }

  // Do stuff with connection
}

One idea I have is to export the getBackoff() method to wait for the correct time until trying to read a new message. Another idea is to have a method WaitConnection() / EnsureConnection(), which blocks until the connection is reestablished. I am happy to create a PR for either of the proposals.

After abnormal closure, automatic reconnection does not happen:

2023/07/24 19:26:15 Connected to server:ws://localhost:8081/connect
2023/07/24 19:26:15 websocket read error: websocket: close 1006 (abnormal closure): unexpected EOF
2023/07/24 19:26:15 websocket read error: websocket: not connected
2023/07/24 19:26:15 websocket read error: websocket: not connected
2023/07/24 19:26:15 websocket read error: websocket: not connected
...

	for _, conn := range connections {
		go func(conn *recws.RecConn) {
			for {
			start:
				_, msg, err := conn.ReadMessage()
				if err != nil {
					log.Println("websocket read error:", err)
					continue
				}
			...
			}
		}(conn)
	...

If you're blocking on ReadMessage(), while the goroutine for KeepAliveTimeout fires a closeAndReconnect, the ReadMessage will fire a second closeAndReconnect. This causes more and more connections to be created and orphaned.

I adapted the basic example to just listen for incoming messages, if the websocket server goes down, recws correctly retries to connect but then the messages are no longer received.

Expected behaviour, the messages continue to be received OK. You can see the messages received before disconnect but not afterwards.

Adapted example

package main

import (
	"context"
	"github.com/recws-org/recws"
	"log"
	"time"
)

func main() {
	ctx, cancel := context.WithCancel(context.Background())
	ws := recws.RecConn{
		KeepAliveTimeout: 10 * time.Second,
	}
	ws.Dial("wss://echo.websocket.org", nil)

	go func() {
		time.Sleep(2 * time.Second)
		cancel()
	}()

	for {
		select {
		default:
			if !ws.IsConnected() {
				log.Printf("Websocket disconnected %s", ws.GetURL())
				continue
			}

			_, message, err := ws.ReadMessage()
			if err != nil {
				log.Printf("Error: ReadMessage %s", ws.GetURL())
				return
			}

			log.Printf("Success: %s", message)
		}
	}
}

I am a beginner to golang, thank you. :p

When trying to get the new v1.3.0:

go: github.com/recws-org/[email protected]: verifying go.mod: github.com/recws-org/[email protected]/go.mod: reading https://sum.golang.org/lookup/github.com/recws-org/[email protected]: 410 Gone
        server response: not found: create zip: makefile: case-insensitive file name collision: "Makefile" and "makefile"

I believe this is due to upstream issue : golang/go#33778

Can you please collapse the 2 Makefiles into 1 and release a v1.3.1?

Thanks!

Was playing around with this lib trying to have a fault tolerant ws connection. Tried patches from different prs and forks, still there was a present problem with race - multiple reconnects occurring from time to time. Also making a custom heartbeat is a headache.

And realised, why not just wrap the ws listener in infinite cycle?

	for {
		select {
		case <-ctx.Done():
			return
		default:
			err := InitAndListenWs()
			if err != nil {
				....
			}
		}
	}

No concurrency problems and easy to implement.

Seems arbitrary and pointless?

Using the basic example, the code below will flood the console and cause 100% CPU usage on a single core;

if !ws.IsConnected() {
	log.Printf("Websocket disconnected %s", ws.GetURL())
	continue
}

It's not the most elegant solution, but an easy fix is to do something like this:

if !rm.conn.IsConnected() {
	<-time.After(rm.conn.RecIntvlMin)
	continue
}

Is there a better way to do this? It doesn't look like there's a connected channel to block on.

go get fetches the 1.0.0 tag, which includes the wrong import path. Thus a new tagged release is needed to fix go get.

Thank you so much for this package! 🎉 👍