Hi Luca,

The log message you were seeing is due to Artemis not supporting the
websocket subprotocol ID that is getting passed in by the client.  In my
tests this didn't make a difference, the message was logged and the
protocol carried on as normal.  I see "CONNECTED" being logged in the
console and the broker handling MQTT PING packets sent from your client.
Which tells me it's all working OK.

Perhaps this websocket protocol ID handshake issue is causing problems for
your browser.  I've sent a fix to handle it properly, I'll let you know
once it's merged so you can try it out.

Regards
Martyn




On Wed, Aug 10, 2016 at 8:52 AM, Luca Capra <luca.ca...@gmail.com> wrote:

> Just to close the loop,
> the subprotocol name used by the library is *mqtt *for v3.1.1
>
> Regards
>
> 2016-08-10 9:35 GMT+02:00 Luca Capra <luca.ca...@gmail.com>:
>
> > Hi Martyn,
> > thank you, indeed I corrected the client code (I exported the wrong code
> > with some paramters experiments)
> >
> > I prepared a running example here (output is all in the browser console)
> >
> > - This run the 1.3.0 version
> > http://46.101.120.189/?url=mqtt://api.raptorbox.eu:1883
> >
> > - This runs 1.4 snapshot from my pc
> > http://46.101.120.189/?url=mqtt://localhost:1883
> >
> >
> > On the client side I get such error
> > Error during WebSocket handshake: Sent non-empty 'Sec-WebSocket-Protocol'
> > header but no response was received
> >
> > and both of them refuse to connect with the same notice on artemis side
> > Requested subprotocol(s) not supported: mqttv3.1
> >
> > I noticed the library actually set the subprotocol as a constant here for
> > broser connections:
> > https://github.com/mqttjs/MQTT.js/blob/master/lib/connect/ws.js#L78
> >
> > so it may be a library issue / decision.
> >
> > What I cannot get is to use MQTT as subprotocol modifying the above
> > constant to something meaningful for artemis.
> >
> > Best
> > Luca
> >
> > 2016-08-09 16:30 GMT+02:00 Martyn Taylor <mtay...@redhat.com>:
> >
> >> Hi Luca,
> >>
> >> The feature is supported.  The issue was how you were configuring your
> >> client.
> >>
> >> There are two versions of MQTT that are supported in Artemis: 3.1 and
> >> 3.1.1.  In 3.1.1 the protocol handshake (or ID), the first couple of
> bytes
> >> sent by the client changed.  So, you must specify the corresponding
> >> version
> >> and protocol ID in your client.
> >>
> >> // For MQTT 3.1.1 this is
> >> protocolId: "MQTT"
> >> protocolVersion: 4
> >>
> >> or
> >>
> >> // For MQTT 3.1 - old version
> >> protocolId: MQIsdp
> >> protocolVersion: 3
> >>
> >> The issue was that your client was specifying:
> >>
> >> protocolId: MQTT
> >> protocolVersion: 3
> >>
> >> This does not match any protocol, hence the error you are seeing.
> >>
> >> I modified your javascript to configure your client like below and it
> >> worked for me.
> >>
> >>       var client = mqtt.connect('ws://localhost:1883', {
> >>           protocolId: 'MQTT'
> >>         , protocolVersion: 4
> >>       });
> >>
> >> Can you update accordingly and let me know if it works.
> >>
> >> Thanks
> >> Martyn
> >>
> >> On Tue, Aug 9, 2016 at 3:11 PM, Luca Capra <luca.ca...@gmail.com>
> wrote:
> >>
> >> > Hi Martyn, all
> >> >
> >> > sorry for taking on from such an old thread, but as of today with
> >> > 1.4-SNAPSHOT I've not been able to have MQTT over websocket to work
> >> >
> >> > I continue to receive a debug notice like this
> >> >
> >> > - Requested subprotocol(s) not supported: MQTT
> >> > - Requested subprotocol(s) not supported: mqttv3.1.1
> >> > - Requested subprotocol(s) not supported: mqtt
> >> > - ... many other tentatives
> >> >
> >> > Is it the subprotcol name that is not matching or is this feature
> >> actually
> >> > not supported?
> >> >
> >> > I'm trying to have the broker to talk from a web browser to the broker
> >> and
> >> > still mqtt over ws seems the best way.
> >> > Do you have any suggestion ?
> >> >
> >> > Thank you!
> >> >
> >> >
> >> >
> >> > 2016-05-31 18:50 GMT+02:00 Martyn Taylor <mtay...@redhat.com>:
> >> >
> >> > > Hi Luca,
> >> > >
> >> > > You are seeing the error because you are not setting the correct
> >> version
> >> > > and protocol header in your client.  Your client should set:
> >> > >
> >> > > ProtocolID="MQTT" and ProtocolVersion=4   // 3.1.1
> >> > > or
> >> > > ProtocolID="MQlsdp" and ProtocolVersion=3 // 3,1
> >> > >
> >> > > Artemis will accept both protocol versions (which are practically
> >> > > identical) but both point to the 3.1.1 implementation.  Your mileage
> >> may
> >> > > vary with 3.1.  I suggest always using 3.1.1.
> >> > >
> >> > > line 15853 of your js should be:
> >> > >
> >> > >       var client = mqtt.connect('ws://localhost:1883', {
> >> > >         protocolId: 'MQTT
> >> > >         , protocolVersion: 4
> >> > >       });
> >> > >
> >> > > Instead of:
> >> > >
> >> > >       var client = mqtt.connect('ws://localhost:1883', {
> >> > >         protocolId: 'MQTT' // or 'MQIsdp' in MQTT 3.1.1
> >> > >         , protocolVersion: 4 // or 3 in MQTT 3.1
> >> > >       }); // you add a ws:// url here
> >> > >
> >> > > Note the protocol ID in 3.1.1 is "MQTT".
> >> > >
> >> > > I made these changes locally and it works fine for me.
> >> > >
> >> > > Thanks
> >> > > Martyn
> >> > >
> >> > > On Tue, May 31, 2016 at 2:04 PM, Luca Capra <luca.ca...@gmail.com>
> >> > wrote:
> >> > >
> >> > > > Hi,
> >> > > >
> >> > > > I tried to setup the mqttjs library and run it in a browser.
> >> > > > Unfortunately, it tries to connect over websocket and artemis
> cannot
> >> > > > recognize the subprotocol
> >> > > >
> >> > > > The source is here: http://pastebin.com/Jgeijbz9
> >> > > >
> >> > > > [Thread-10 (activemq-netty-threads-2076099910)] [DEBUG]
> >> > > > io.netty.handler.codec.http.websocketx.WebSocketServerHandshaker
> -
> >> > [id:
> >> > > > 0x8958b0d5, /127.0.0.1:60138 => /127.0.0.1:1883] WebSocket
> version
> >> V13
> >> > > > server handshake
> >> > > > [Thread-10 (activemq-netty-threads-2076099910)] [DEBUG]
> >> > > > io.netty.handler.codec.http.websocketx.WebSocketServerHandshaker
> -
> >> > > > WebSocket version 13 server handshake key:
> pJTCBulMMJvyCoIpj2wEPg==,
> >> > > > response: 6vRlkIXEqz7ihckJ2TFuB6V9uhY=
> >> > > > [Thread-10 (activemq-netty-threads-2076099910)] [DEBUG]
> >> > > > io.netty.handler.codec.http.websocketx.WebSocketServerHandshaker
> -
> >> > > > Requested subprotocol(s) not supported: mqttv3.1
> >> > > >
> >> > > >
> >> > > > In the latest master the issue with the client disconnecting is
> >> still
> >> > > > there, but the error message is slightly different. In 1.2 it used
> >> to
> >> > > work
> >> > > >
> >> > > > [Thread-3 (activemq-netty-threads-2076099910)] [DEBUG]
> >> > > > io.netty.util.internal.JavassistTypeParameterMatcherGenerator -
> >> > > Generated:
> >> > > >
> >> > > > io.netty.util.internal.__matchers__.io.netty.handler.
> >> > > codec.mqtt.MqttMessageMatcher
> >> > > > [Thread-3 (activemq-netty-threads-2076099910)] [DEBUG]
> >> > > > org.apache.activemq.artemis.core.protocol.mqtt - SESSION CREATED:
> >> > > > d2cd94e9-3d23-4a85-bb53-53528cd9fcb1
> >> > > > [Thread-3 (activemq-netty-threads-2076099910)] [DEBUG]
> >> > > > org.apache.activemq.artemis.core.protocol.mqtt - Bad Message
> >> > > Disconnecting
> >> > > > Client
> >> > > >
> >> > > >
> >> > > > Thank you
> >> > > >
> >> > > > Regards
> >> > > > Luca
> >> > > >
> >> > > >
> >> > > > 2016-05-27 13:25 GMT+02:00 Luca Capra <luca.ca...@gmail.com>:
> >> > > >
> >> > > > > Hi Martyn,
> >> > > > > I will bundle the js all in a page for testing.
> >> > > > >
> >> > > > > I think the js library will try to interact with "mqtt over
> >> > websockets"
> >> > > > > from the browser.
> >> > > > > (See just one line above here,
> >> > > > > https://github.com/mqttjs/MQTT.js#contributing)
> >> > > > >
> >> > > > > Is it supposed to work in Artemis? (.. Nice feature btw!)
> >> > > > >
> >> > > > > I can either provide you the client wrapped in docker or vagrant
> >> vm
> >> > if
> >> > > it
> >> > > > > works for you.
> >> > > > >
> >> > > > > Thanks
> >> > > > > Luca
> >> > > > >
> >> > > > >
> >> > > > > 2016-05-27 11:56 GMT+02:00 Martyn Taylor <mtay...@redhat.com>:
> >> > > > >
> >> > > > >> Hi Luca,
> >> > > > >>
> >> > > > >> I had a play around with the JS you supplied but didn't get it
> >> > > working.
> >> > > > >> Any chance you could browsify the test so I can run it without
> >> > having
> >> > > a
> >> > > > >> NodeJS environment set up, (A bash script would do if that's
> >> > > > possible).  I
> >> > > > >> just need to see the issue in the broker.
> >> > > > >>
> >> > > > >> Thanks
> >> > > > >>
> >> > > > >> On Thu, May 26, 2016 at 10:18 AM, Luca Capra <
> >> luca.ca...@gmail.com>
> >> > > > >> wrote:
> >> > > > >>
> >> > > > >> > Ok, I will double check on the client side.
> >> > > > >> >
> >> > > > >> >
> >> > > > >> > If you have the time, here the test case
> >> > > > >> >
> >> > > > >> > The nodejs client code
> >> > > > >> > https://gist.github.com/muka/b6374610618978948f39d83e5248ff
> 7e
> >> > > > >> >
> >> > > > >> > The broker implementation (raptor-broker) is here (branch
> >> artemis)
> >> > > > >> >
> >> > > > >> > https://github.com/muka/raptor/tree/artemis
> >> > > > >> >
> >> > > > >> > To run it should just require to mvn install and run the
> >> > mainClass.
> >> > > It
> >> > > > >> > requires artemis modules of 1.3-SNAPSHOT in the local
> >> repository..
> >> > > I'm
> >> > > > >> > pretty sure you have them already :)
> >> > > > >> >
> >> > > > >> > It tries to load configurations (avail in repository
> ./config)
> >> > from
> >> > > > >> > /etc/raptor or from system property -DconfigDir=<path>
> >> > > > >> >
> >> > > > >> > Thank you for the precious help!
> >> > > > >> >
> >> > > > >> > Regards
> >> > > > >> > Luca
> >> > > > >> >
> >> > > > >> >
> >> > > > >> > 2016-05-26 10:42 GMT+02:00 Martyn Taylor <mtay...@redhat.com
> >:
> >> > > > >> >
> >> > > > >> > > Hi Luca,
> >> > > > >> > >
> >> > > > >> > > Only 3.1.1 is supported in Artemis.  We've not tested
> against
> >> > > older
> >> > > > >> > > versions.  Could you ensure you are setting the version
> >> properly
> >> > > in
> >> > > > >> your
> >> > > > >> > > client.
> >> > > > >> > >
> >> > > > >> > > Error processing control packet usually means the client is
> >> > doing
> >> > > > >> > something
> >> > > > >> > > wrong.  If you send me your test case / reproducer, I will
> >> take
> >> > a
> >> > > > >> look.
> >> > > > >> > >
> >> > > > >> > > Thanks
> >> > > > >> > > Martyn
> >> > > > >> > >
> >> > > > >> > > On Thu, May 26, 2016 at 9:06 AM, Luca Capra <
> >> > luca.ca...@gmail.com
> >> > > >
> >> > > > >> > wrote:
> >> > > > >> > >
> >> > > > >> > > > Hi Martyn,
> >> > > > >> > > > thank you for your feedback!
> >> > > > >> > > >
> >> > > > >> > > > I had in fact two different acceptor on my broker.xml.
> I'm
> >> now
> >> > > > using
> >> > > > >> > just
> >> > > > >> > > > one now. I've also moved to latest master 1.3-SNAPSHOT
> from
> >> > git
> >> > > > >> > > >
> >> > > > >> > > > I noticed I get this exception now on connection via mqtt
> >> > > > >> > > >
> >> > > > >> > > > WARN: Error processing Control Packet, Disconnecting
> >> Client2
> >> > > > >> > > >
> >> > > > >> > > > From the debugger I can see it is an
> >> ArrayOutOfBoundException
> >> > > > which
> >> > > > >> > > happens
> >> > > > >> > > > somewhere around ctx.flush() in MqttProtocolHandler.
> >> > sendConnack
> >> > > > >> > > >
> >> > > > >> > > > The client is mqttjs a nodejs based module implementing 3
> >> ->
> >> > > 3.1.1
> >> > > > >> mqtt
> >> > > > >> > > > version
> >> > > > >> > > >
> >> > > > >> > > > I'm using a custom SecurityManager, but should not be at
> >> the
> >> > > cause
> >> > > > >> of
> >> > > > >> > > such
> >> > > > >> > > > issue
> >> > > > >> > > >
> >> > > > >> > > > Need to investigate better
> >> > > > >> > > >
> >> > > > >> > > > Best
> >> > > > >> > > > Luca
> >> > > > >> > > >
> >> > > > >> > > > 2016-05-25 21:40 GMT+02:00 Martyn Taylor <
> >> mtay...@redhat.com
> >> > >:
> >> > > > >> > > >
> >> > > > >> > > > > Hi Luca,
> >> > > > >> > > > >
> >> > > > >> > > > > You don't need any specific configuration to do this.
> >> You
> >> > can
> >> > > > >> just
> >> > > > >> > set
> >> > > > >> > > > up
> >> > > > >> > > > > a MQTT Client and a STOMP client and start
> >> sending/receiving
> >> > > > from
> >> > > > >> the
> >> > > > >> > > > same
> >> > > > >> > > > > address.
> >> > > > >> > > > >
> >> > > > >> > > > > There are a couple of issues with Artemis versions
> >> <=1.2.0.
> >> > > > I've
> >> > > > >> > fixed
> >> > > > >> > > > > these issues on master and will be doing a 1.3.0
> release
> >> > very
> >> > > > >> soon,
> >> > > > >> > it
> >> > > > >> > > > > should be available in the next week or so, your best
> >> bet is
> >> > > to
> >> > > > >> wait
> >> > > > >> > > for
> >> > > > >> > > > > this.
> >> > > > >> > > > >
> >> > > > >> > > > > I've also added a simple test to that shows this
> working.
> >> > > See:
> >> > > > >> > > > >
> >> > > > >> > > > >
> >> > > > >> > > >
> >> > > > >> > >
> >> > > > >> >
> >> > > > >>
> >> > > > https://github.com/apache/activemq-artemis/pull/543/commits/
> >> > > 62d284d7765218ced2d0f441cf26ffaaaf3c26f2
> >> > > > >> > > > >
> >> > > > >> > > > > Thanks
> >> > > > >> > > > > Martyn
> >> > > > >> > > > >
> >> > > > >> > > > >
> >> > > > >> > > > > On Wed, May 25, 2016 at 3:58 PM, Luca Capra <
> >> > > > luca.ca...@gmail.com
> >> > > > >> >
> >> > > > >> > > > wrote:
> >> > > > >> > > > >
> >> > > > >> > > > > > Hi all,
> >> > > > >> > > > > > on a project I'm working on I've switched from Apollo
> >> to
> >> > > > >> Artemis.
> >> > > > >> > > > > >
> >> > > > >> > > > > > I'm running Artemis as an embed server and would like
> >> to
> >> > > have
> >> > > > >> > > > > > inter-protocol communication between stomp and mqtt.
> >> > > > >> > > > > >
> >> > > > >> > > > > > What I'm missing is if this can be done from
> >> configuration
> >> > > or
> >> > > > >> > > creating
> >> > > > >> > > > an
> >> > > > >> > > > > > mqtt and stomp client each one to deliver messages
> the
> >> > other
> >> > > > >> one.
> >> > > > >> > > > > >
> >> > > > >> > > > > > Thank you
> >> > > > >> > > > > > Luca
> >> > > > >> > > > > >
> >> > > > >> > > > >
> >> > > > >> > > >
> >> > > > >> > >
> >> > > > >> >
> >> > > > >>
> >> > > > >
> >> > > > >
> >> > > >
> >> > >
> >> >
> >>
> >
> >
>

Reply via email to