IE9 error - connection timeout

[svscorp]

Hi.

I try to figure out why noVNC doesn't work in IE9. First of all I had a problem when SWF wasn't loaded. Just saw the message: 'canvas is not supported'.

I've found that it is because I didn't use meta tag with IE=9. Probably it will helps someone:

But after that I went to the new issue. Connection timeout. With IE10 - it works. With other browsers it works as well. IE9 not.

Here us log. May be someone knows where the trick?

LOG: Forcing uncached load of WebSocketMain.swf
LOG: >> RFB.constructor
LOG: >> Display.constructor
LOG: User Agent: Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; WOW64; Trident/5.0)
LOG: Browser: trident 6
LOG: No scaling support
LOG: Setting viewport to full display region
LOG: Prefering javascript operations
LOG: Data URI scheme cursor not supported
LOG: << Display.constructor
LOG: Using web-socket-js bridge. Flash version: 11
LOG: >> Keyboard.allKeysUp
LOG: << Keyboard.allKeysUp
LOG: No scaling support
LOG: Setting viewport to full display region
LOG: New state 'loaded', was 'disconnected'. Msg: noVNC ready: WebSockets emulation, canvas rendering
LOG: << RFB.constructor
LOG: >> Keyboard.allKeysUp
LOG: << Keyboard.allKeysUp
LOG: New state 'connect', was 'loaded'.
LOG: >> RFB.connect
LOG: connecting to wss://host:port/hash
LOG: << RFB.connect
LOG: >> Keyboard.allKeysUp
LOG: << Keyboard.allKeysUp
LOG: Closing WebSocket connection
LOG: New state 'failed', was 'connect'. Msg: Connect timeout
LOG: Clearing connect timer
LOG: >> WebSock.onclose
LOG: Received onclose while disconnected
LOG: << WebSock.onclose
LOG: >> Keyboard.allKeysUp
LOG: << Keyboard.allKeysUp
LOG: New state 'failed', was 'failed'.

[AudriusButkevicius]

I have the same issue.

IE10 supports native websockets, thats why it works, though IE9 uses the SWF emulation which I haven't been able to force to work.

[svscorp]

JarOfJan, read this: https://github.com/kanaka/websockify/tree/master/include/web-socket-js

The problem was with 843 port and crpossdomain policy (because SWF needs to have this to connect).

[smithkl42]

I'm having the same trouble. Among the things I've done to troubleshoot:

• Set Util._log_level = "debug". More information, but nothing that seems to help.
• Set WEB_SOCKET_DEBUG = 1. So far as I can tell, no additional messages at all.
• Run Flash policy server on port 843, not just on the websockets port (for me, 5901). I can connect to this from the machine in question, and it responds appropriate to policy server requests, but it never gets called by the WebSocketMain.swf file.
• Switch to WebSocketMainInsecure.swf. No difference, so far as I can tell.

Here are the logs:

LOG: Trying to connect to RFB server 
LOG: >> RFB.constructor 
LOG: >> Display.constructor 
LOG: User Agent: Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; WOW64; Trident/5.0; SLCC2; .NET CLR 2.0.50727; .NET CLR 3.5.30729; .NET CLR 3.0.30729; Media Center PC 6.0; .NET4.0C) 
LOG: Browser: trident 6 
LOG: No scaling support 
LOG: Setting viewport to full display region 
Data URI scheme cursor not supported 
LOG: << Display.constructor 
Using web-socket-js bridge. Flash version: 11 
LOG: >> Keyboard.allKeysUp 
LOG: << Keyboard.allKeysUp 
LOG: No scaling support 
LOG: Setting viewport to full display region 
New state 'loaded', was 'disconnected'. Msg: noVNC ready: WebSockets emulation, canvas rendering 
LOG: RFB state: loaded, msg: noVNC ready: WebSockets emulation, canvas rendering 
LOG: << RFB.constructor 
LOG: >> Keyboard.allKeysUp 
LOG: << Keyboard.allKeysUp 
New state 'connect', was 'loaded'. 
LOG: >> RFB.connect 
LOG: connecting to ws://app.dev.alanta.com:5901/ 
Skipping unsupported WebSocket binary sub-protocol 
LOG: << RFB.connect 
LOG: RFB state: connect, msg: undefined 
LOG: >> Keyboard.allKeysUp 
LOG: << Keyboard.allKeysUp 
LOG: Closing WebSocket connection 
New state 'failed', was 'connect'. Msg: Connect timeout after 5 seconds 
LOG: Clearing connect timer 

Any additional troubleshooting steps?

[smithkl42]

I should also note that the issue may have something to do with the Flash initialization.

For starters, the swfobject callback below does get called, with e.success==true.

swfobject.embedSWF(
  WEB_SOCKET_SWF_LOCATION,
  "webSocketFlash",
  "1" /* width */,
  "1" /* height */,
  "10.0.0" /* SWF version */,
  null,
  null,
  {hasPriority: true, swliveconnect : true, allowScriptAccess: "always"},
  null,
  function(e) {
    if (!e.success) {
      logger.error("[WebSocket] swfobject.embedSWF failed");
    }
  }
);

Nevertheless, I don't think it's getting initialized correctly. One of the first things the WebSocketMain.swf file is supposed to do is call WebSocket.__onFlashInitialized, like so:

  public function WebSocketMain() {
    // various stuff....
    ExternalInterface.call("WebSocket.__onFlashInitialized");
  }

However, this doesn't seem to happen - when I set a breakpoint in the JS, the WebSocket.__onFlashInitialized never seems to get called. So presumably that's where the problem is located. Still digging...

[infernix]

This should really be reopened. I'm having the exact same issue.

Things I did:

• Install debug flash player. Doesn't even create a log.
• Updated the swf. No change
• Run ruby web_socket.rb on port 843, with various options for to-ports and domain (even "*"). Though it shouldn't matter in my case since everything is running on the same server.

Running wireshark shows that there's actually a connection being made to the target port, and it's correctly requesting /websockify. It also shows that it tries to connect to port 843.

Here's my log:

LOG: >> RFB.constructor
LOG: >> Display.constructor
LOG: User Agent: Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; WOW64; Trident/5.0)
LOG: Browser: trident 6
LOG: No scaling support
LOG: Setting viewport to full display region
Data URI scheme cursor not supported
LOG: << Display.constructor
Using web-socket-js bridge. Flash version: 11
LOG: >> Keyboard.allKeysUp
LOG: << Keyboard.allKeysUp
LOG: No scaling support
LOG: Setting viewport to full display region
New state 'loaded', was 'disconnected'. Msg: noVNC ready: WebSockets emulation, canvas rendering
LOG: << RFB.constructor
LOG: >> Keyboard.allKeysUp
LOG: << Keyboard.allKeysUp
New state 'connect', was 'loaded'.
LOG: >> RFB.connect
LOG: connecting to ws://xxxxxxxx:40000/websockify
Skipping unsupported WebSocket binary sub-protocol
LOG: << RFB.connect
LOG: >> Keyboard.allKeysUp
LOG: << Keyboard.allKeysUp
LOG: Closing WebSocket connection
New state 'failed', was 'connect'. Msg: Connect timeout
LOG: Clearing connect timer
LOG: >> Keyboard.allKeysUp
LOG: << Keyboard.allKeysUp
New state 'disconnected', was 'failed'.
LOG: >> WebSock.onopen
LOG: Server chose sub-protocol: base64
LOG: Already in state 'failed', ignoring.
LOG: << WebSock.onopen

[infernix]

Ha! after a lot of digging, I found the problem.

in rfb.js, when I change the conf.connectTimeout * 1000 to conf.connectTimeout * 4000, the problem goes away.

So it looks like it's a timing issue, and the default timeout should just be increased upwards from the 2 seconds if the swf is being used.

I don't really understand why though, because I had already changed "def_con_timeout = Websock_native ? 7 : 7" but it seems that that wasn't being used, which can be verified if you print the timer value in fail("Connect timeout").

[kanaka]

websock.js must be getting loaded after rfb.js (Websock_native is set true or false by websock.js). IE9 is probably doing something weird with the order of dynamic loading of scripts.

[DirectXMan12]

closing due to age and answer

[sanaeefar-saeed]

thx infernix .your soloution really worked.:D