Yikes. Messages alone should be enough to keep the connection alive.
It shouldn't require explicit ping/pongs.

William
ᐧ

On Thu, May 21, 2015 at 10:13 AM, Stuart Henderson <[email protected]>
wrote:

> Further to the below, today we discovered that the issue was down to a
> Palo Alto firewall which the websockets were traversing.
>
> It seems that the firewall was not interpreting actual content (JSON)
> being exchanged over the websocket as keepalives. We don't actually send
> any bona-fide keepalives as such (no ping/pong)... but we do sent content.
> Seems that the firewall was set up to monitor ping/pong and as it could not
> see any it was closing the connections.
>
> Just thought I would send this in, just in case anyone found it useful now
> or later.
>
> Stuart
>
>
> On 19 May 2015 at 23:14, Stuart Henderson <[email protected]> wrote:
>
>> Hi,
>>
>> I'd be deeply grateful for any assistance anyone could offer with this
>> problem, which is causing me extreme pain.
>>
>> We have a Java test which instantiates websockets over which we are
>> sending JSON requests to an Apache webserver. The webserver is reading the
>> JSON which in turn sets the state for various virtual users.
>>
>> After the initial websocket set up and a couple of JSON exchanges the
>> session can often go quiet for some time and the next communication might
>> be as much as 50 minutes away. When this happens and our test sends a
>> simple state change after about 50 minutes we get a websocket.closed() and
>> an event code of 1006. Obviously the state change fails and our test is
>> trashed.
>>
>> The 'client' test is running on a Windows Server 2008R2. We are running
>> out of ideas about how to further troubleshoot this. A simulated session
>> running in Chrome on the same server and talking to the same Apache
>> webserver works just great, its just the Java test with the Jetty
>> implementation that fails. The test, when run on a W7 workstation is
>> okay... it seems to be the combination of Jetty 9.xx AND the Windows Server
>> 2008R2 that breaks it.
>>
>> I've wiresharked the connections... but as we are using wss:// it's
>> almost impossible to see what's going on on the wire. All I can see is that
>> at the 50min mark the client tries to send the request and gets nothing
>> back from the server. There are four retransmissions 1.2s apart then the
>> session crashes out and the websocket.closed() is printed.
>>
>> To say this problem is causing us pain is putting it very lightly. Can
>> anyone point me in the right direction with with how to troubleshoot this
>> or get more information?
>>
>> Thanks,
>> Stuart
>>
>>
>
> _______________________________________________
> jetty-users mailing list
> [email protected]
> To change your delivery options, retrieve your password, or unsubscribe
> from this list, visit
> https://dev.eclipse.org/mailman/listinfo/jetty-users
>
_______________________________________________
jetty-users mailing list
[email protected]
To change your delivery options, retrieve your password, or unsubscribe from 
this list, visit
https://dev.eclipse.org/mailman/listinfo/jetty-users

Reply via email to