MQTT for Pharo
Locked
 
|
Hi,
MQTT is a light-weight publish/subscribe messaging protocol, originally created around 1998. It is now an official open industry ISO standard. It is perfect for large-scale Internet of Things applications and high performance mobile messaging. The publish/subscribe messaging pattern requires a message broker. The broker is responsible for distributing messages to interested clients based on the topic of a message. Parties communicating with each other over MQTT would all be clients in different roles, like producers and consumers, using the broker as middleware. Many client libraries for different programming languages and multiple brokers/servers are available. Facebook Messenger and Amazon AWS IOT are two users, among many others. A good client library for Pharo was not yet available. I started a new MQTT project and I am looking for collaborators to help me finish it. The official specification is quite readable and there is a lot of information available (see the References/Links section at the end). Right now, a first version of the following features is available: - reading & writing of all 14 binary packet types - an experimental client with support for connection open/close, ping, subscribe/unsubscribe, QoS levels 0 (at most once), 1 (at least once) and 2 (exactly once) for application (publish) messages in both directions, message/package IDs and keep alive (heartbeat) - unit tests, for packet reading/writing and for clients against 3 publicly available sandbox/test brokers Basically, the code works but needs polishing and maturing. Also, it might be useful to experiment with alternative client API design. Not all features are yet implemented. It would also be nice to implement an actual server/broker, not to replace production quality servers, but as a proof of concept and tools during development. Code http://smalltalkhub.com/#!/~SvenVanCaekenberghe/MQTT/ Right now, documentation is limited, but there are class comments and the most important public API methods are commented too. There is no Metacello configuration yet, just load the 3 packages. References/Links - http://mqtt.org - https://en.wikipedia.org/wiki/MQTT - http://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html - https://github.com/mqtt/mqtt.github.io/wiki/software - http://mosquitto.org - https://github.com/emqtt/emqttd - http://kamilfb.github.io/mqtt-spy/ - https://github.com/eclipse/paho.mqtt-spy/wiki - https://eclipse.org/paho/ - https://eclipse.org/paho/clients/c/embedded/ - http://www.rabbitmq.com/mqtt.html - https://vernemq.com - https://www.ibm.com/developerworks/community/blogs/c565c720-fe84-4f63-873f-607d87787327/entry/mqtt_security Sandbox/test Servers/Brokers iot.eclipse.org:1883 test.mosquitto.org:1883 broker.mqtt-dashboard.com:1883 Sven -- Sven Van Caekenberghe http://stfx.eu Smalltalk is the Red Pill |
|
|
Hi Sven!
One thing I noticed when testing the RabbitMQ client with keepalive > 0, was connection being closed and all subscriptions lost when receiving large payloads, due to timeout before the keepalive packet could be sent before waiting to receive next object. I see the default in ExperimentalClient is to disregard ConnectionTimedOut when that happens, and resend a ping. Are MQTT servers generally more lenient wrt dropping subscriptions once the timeout occurs, or will that lead to no new data being received from the broker? (Also, doesn't there need to be a guarantee that timeout < keepalive when keepalive is > 0, for this logic to work as intended?) Cheers, Henry > On 21 Dec 2016, at 12:15 , Sven Van Caekenberghe <[hidden email]> wrote: > > Hi, > > MQTT is a light-weight publish/subscribe messaging protocol, originally created around 1998. It is now an official open industry ISO standard. It is perfect for large-scale Internet of Things applications and high performance mobile messaging. > > The publish/subscribe messaging pattern requires a message broker. The broker is responsible for distributing messages to interested clients based on the topic of a message. Parties communicating with each other over MQTT would all be clients in different roles, like producers and consumers, using the broker as middleware. > > Many client libraries for different programming languages and multiple brokers/servers are available. Facebook Messenger and Amazon AWS IOT are two users, among many others. > > > A good client library for Pharo was not yet available. I started a new MQTT project and I am looking for collaborators to help me finish it. The official specification is quite readable and there is a lot of information available (see the References/Links section at the end). > > > Right now, a first version of the following features is available: > > - reading & writing of all 14 binary packet types > > - an experimental client with support for connection open/close, ping, subscribe/unsubscribe, QoS levels 0 (at most once), 1 (at least once) and 2 (exactly once) for application (publish) messages in both directions, message/package IDs and keep alive (heartbeat) > > - unit tests, for packet reading/writing and for clients against 3 publicly available sandbox/test brokers > > Basically, the code works but needs polishing and maturing. Also, it might be useful to experiment with alternative client API design. Not all features are yet implemented. It would also be nice to implement an actual server/broker, not to replace production quality servers, but as a proof of concept and tools during development. > > > Code > > http://smalltalkhub.com/#!/~SvenVanCaekenberghe/MQTT/ > > Right now, documentation is limited, but there are class comments and the most important public API methods are commented too. There is no Metacello configuration yet, just load the 3 packages. > > > References/Links > > - http://mqtt.org > - https://en.wikipedia.org/wiki/MQTT > - http://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html > - https://github.com/mqtt/mqtt.github.io/wiki/software > - http://mosquitto.org > - https://github.com/emqtt/emqttd > - http://kamilfb.github.io/mqtt-spy/ > - https://github.com/eclipse/paho.mqtt-spy/wiki > - https://eclipse.org/paho/ > - https://eclipse.org/paho/clients/c/embedded/ > - http://www.rabbitmq.com/mqtt.html > - https://vernemq.com > - https://www.ibm.com/developerworks/community/blogs/c565c720-fe84-4f63-873f-607d87787327/entry/mqtt_security > > > Sandbox/test Servers/Brokers > > iot.eclipse.org:1883 > test.mosquitto.org:1883 > broker.mqtt-dashboard.com:1883 > > > Sven > > -- > Sven Van Caekenberghe > http://stfx.eu > Smalltalk is the Red Pill > > |
|
|
In reply to this post by Sven Van Caekenberghe-2
Sven, I am currently working with IoT and MQTT is necessary. At this point I use mosquitto as the broker and your Pharo client is nice to have and I will test it and give you feedback. I've got a working microPython codebase so can compare what is in there (this for the LoPy device): Thx for the module! Phil On Wed, Dec 21, 2016 at 12:15 PM, Sven Van Caekenberghe <[hidden email]> wrote: Hi, |
|
|
In reply to this post by Henrik Sperre Johansen
Hendrik,
Thank you for this detailed feedback. > On 21 Dec 2016, at 12:44, Henrik Johansen <[hidden email]> wrote: > > Hi Sven! > One thing I noticed when testing the RabbitMQ client with keepalive > 0, was connection being closed and all subscriptions lost when receiving large payloads, due to timeout before the keepalive packet could be sent before waiting to receive next object. Indeed, I used my Stamp (STOMP) Rabbit MQ client as a model for the MQTT one - there is a lot of similarity, especially in concept. Keep alive processing is not that easy. I tried to do it by using read timeouts as a source of regular opportunities to check for the need to process keep alive logic. But of course, if you have no outstanding read (in a loop), that won't work. The fact that receiving a large payload would trigger an actual keep alive time out is not something that I have seen myself. It seems weird that the reading/transferring of incoming data would not count as activity against keep alive, no ? > I see the default in ExperimentalClient is to disregard ConnectionTimedOut when that happens, and resend a ping. I use a similar technique, yes, > Are MQTT servers generally more lenient wrt dropping subscriptions once the timeout occurs, or will that lead to no new data being received from the broker? I don't know. I think this needs to be sorted out during actual usage. Note that MQTT has to the concept of resuming an existing session that keeps subscriptions and outstanding undelivered data. I have not explored that area. > (Also, doesn't there need to be a guarantee that timeout < keepalive when keepalive is > 0, for this logic to work as intended?) Yes, that is something that should be enforced. BTW, the MQTT spec says: << If the Keep Alive value is non-zero and the Server does not receive a Control Packet from the Client within one and a half times the Keep Alive time period, it MUST disconnect the Network Connection to the Client as if the network had failed [MQTT-3.1.2-24]. >> So I guess there is some leniency. Sven > Cheers, > Henry > >> On 21 Dec 2016, at 12:15 , Sven Van Caekenberghe <[hidden email]> wrote: >> >> Hi, >> >> MQTT is a light-weight publish/subscribe messaging protocol, originally created around 1998. It is now an official open industry ISO standard. It is perfect for large-scale Internet of Things applications and high performance mobile messaging. >> >> The publish/subscribe messaging pattern requires a message broker. The broker is responsible for distributing messages to interested clients based on the topic of a message. Parties communicating with each other over MQTT would all be clients in different roles, like producers and consumers, using the broker as middleware. >> >> Many client libraries for different programming languages and multiple brokers/servers are available. Facebook Messenger and Amazon AWS IOT are two users, among many others. >> >> >> A good client library for Pharo was not yet available. I started a new MQTT project and I am looking for collaborators to help me finish it. The official specification is quite readable and there is a lot of information available (see the References/Links section at the end). >> >> >> Right now, a first version of the following features is available: >> >> - reading & writing of all 14 binary packet types >> >> - an experimental client with support for connection open/close, ping, subscribe/unsubscribe, QoS levels 0 (at most once), 1 (at least once) and 2 (exactly once) for application (publish) messages in both directions, message/package IDs and keep alive (heartbeat) >> >> - unit tests, for packet reading/writing and for clients against 3 publicly available sandbox/test brokers >> >> Basically, the code works but needs polishing and maturing. Also, it might be useful to experiment with alternative client API design. Not all features are yet implemented. It would also be nice to implement an actual server/broker, not to replace production quality servers, but as a proof of concept and tools during development. >> >> >> Code >> >> http://smalltalkhub.com/#!/~SvenVanCaekenberghe/MQTT/ >> >> Right now, documentation is limited, but there are class comments and the most important public API methods are commented too. There is no Metacello configuration yet, just load the 3 packages. >> >> >> References/Links >> >> - http://mqtt.org >> - https://en.wikipedia.org/wiki/MQTT >> - http://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html >> - https://github.com/mqtt/mqtt.github.io/wiki/software >> - http://mosquitto.org >> - https://github.com/emqtt/emqttd >> - http://kamilfb.github.io/mqtt-spy/ >> - https://github.com/eclipse/paho.mqtt-spy/wiki >> - https://eclipse.org/paho/ >> - https://eclipse.org/paho/clients/c/embedded/ >> - http://www.rabbitmq.com/mqtt.html >> - https://vernemq.com >> - https://www.ibm.com/developerworks/community/blogs/c565c720-fe84-4f63-873f-607d87787327/entry/mqtt_security >> >> >> Sandbox/test Servers/Brokers >> >> iot.eclipse.org:1883 >> test.mosquitto.org:1883 >> broker.mqtt-dashboard.com:1883 >> >> >> Sven >> >> -- >> Sven Van Caekenberghe >> http://stfx.eu >> Smalltalk is the Red Pill >> >> > > |
|
|
In reply to this post by philippeback
> On 21 Dec 2016, at 12:50, [hidden email] wrote: > > Sven, > > I am currently working with IoT and MQTT is necessary. At this point I use mosquitto as the broker and your Pharo client is nice to have and I will test it and give you feedback. That is good to here, Phil. Yes, such real world testing is much needed. This is actually quite fun stuff to experiment with. Let me know how it goes. Sven > I've got a working microPython codebase so can compare what is in there (this for the LoPy device): > > Thx for the module! > > Phil > > On Wed, Dec 21, 2016 at 12:15 PM, Sven Van Caekenberghe <[hidden email]> wrote: > Hi, > > MQTT is a light-weight publish/subscribe messaging protocol, originally created around 1998. It is now an official open industry ISO standard. It is perfect for large-scale Internet of Things applications and high performance mobile messaging. > > The publish/subscribe messaging pattern requires a message broker. The broker is responsible for distributing messages to interested clients based on the topic of a message. Parties communicating with each other over MQTT would all be clients in different roles, like producers and consumers, using the broker as middleware. > > Many client libraries for different programming languages and multiple brokers/servers are available. Facebook Messenger and Amazon AWS IOT are two users, among many others. > > > A good client library for Pharo was not yet available. I started a new MQTT project and I am looking for collaborators to help me finish it. The official specification is quite readable and there is a lot of information available (see the References/Links section at the end). > > > Right now, a first version of the following features is available: > > - reading & writing of all 14 binary packet types > > - an experimental client with support for connection open/close, ping, subscribe/unsubscribe, QoS levels 0 (at most once), 1 (at least once) and 2 (exactly once) for application (publish) messages in both directions, message/package IDs and keep alive (heartbeat) > > - unit tests, for packet reading/writing and for clients against 3 publicly available sandbox/test brokers > > Basically, the code works but needs polishing and maturing. Also, it might be useful to experiment with alternative client API design. Not all features are yet implemented. It would also be nice to implement an actual server/broker, not to replace production quality servers, but as a proof of concept and tools during development. > > > Code > > http://smalltalkhub.com/#!/~SvenVanCaekenberghe/MQTT/ > > Right now, documentation is limited, but there are class comments and the most important public API methods are commented too. There is no Metacello configuration yet, just load the 3 packages. > > > References/Links > > - http://mqtt.org > - https://en.wikipedia.org/wiki/MQTT > - http://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html > - https://github.com/mqtt/mqtt.github.io/wiki/software > - http://mosquitto.org > - https://github.com/emqtt/emqttd > - http://kamilfb.github.io/mqtt-spy/ > - https://github.com/eclipse/paho.mqtt-spy/wiki > - https://eclipse.org/paho/ > - https://eclipse.org/paho/clients/c/embedded/ > - http://www.rabbitmq.com/mqtt.html > - https://vernemq.com > - https://www.ibm.com/developerworks/community/blogs/c565c720-fe84-4f63-873f-607d87787327/entry/mqtt_security > > > Sandbox/test Servers/Brokers > > iot.eclipse.org:1883 > test.mosquitto.org:1883 > broker.mqtt-dashboard.com:1883 > > > Sven > > -- > Sven Van Caekenberghe > http://stfx.eu > Smalltalk is the Red Pill > > > |
|
|
FWIW get yourself an account at http://enco.io The devportal has mqtt endpoints so that we can integrate with that and possibly have some market place solutions offered and running with Pharo. I have a small EnCoClient class in Pharo that works. Now making the API support more extensive. Phil On Wed, Dec 21, 2016 at 2:59 PM, Sven Van Caekenberghe <[hidden email]> wrote:
|
|
|
> On 21 Dec 2016, at 15:11, [hidden email] wrote: > > FWIW get yourself an account at http://enco.io OK, I'll put it on my list of things to look at. > The devportal has mqtt endpoints so that we can integrate with that and possibly have some market place solutions offered and running with Pharo. > > I have a small EnCoClient class in Pharo that works. Now making the API support more extensive. > > Docs: http://docs.enco.io/docs/device-sensor-apis > > Phil > > On Wed, Dec 21, 2016 at 2:59 PM, Sven Van Caekenberghe <[hidden email]> wrote: > > > On 21 Dec 2016, at 12:50, [hidden email] wrote: > > > > Sven, > > > > I am currently working with IoT and MQTT is necessary. At this point I use mosquitto as the broker and your Pharo client is nice to have and I will test it and give you feedback. > > That is good to here, Phil. > > Yes, such real world testing is much needed. This is actually quite fun stuff to experiment with. Let me know how it goes. > > Sven > > > I've got a working microPython codebase so can compare what is in there (this for the LoPy device): > > > > Thx for the module! > > > > Phil > > > > On Wed, Dec 21, 2016 at 12:15 PM, Sven Van Caekenberghe <[hidden email]> wrote: > > Hi, > > > > MQTT is a light-weight publish/subscribe messaging protocol, originally created around 1998. It is now an official open industry ISO standard. It is perfect for large-scale Internet of Things applications and high performance mobile messaging. > > > > The publish/subscribe messaging pattern requires a message broker. The broker is responsible for distributing messages to interested clients based on the topic of a message. Parties communicating with each other over MQTT would all be clients in different roles, like producers and consumers, using the broker as middleware. > > > > Many client libraries for different programming languages and multiple brokers/servers are available. Facebook Messenger and Amazon AWS IOT are two users, among many others. > > > > > > A good client library for Pharo was not yet available. I started a new MQTT project and I am looking for collaborators to help me finish it. The official specification is quite readable and there is a lot of information available (see the References/Links section at the end). > > > > > > Right now, a first version of the following features is available: > > > > - reading & writing of all 14 binary packet types > > > > - an experimental client with support for connection open/close, ping, subscribe/unsubscribe, QoS levels 0 (at most once), 1 (at least once) and 2 (exactly once) for application (publish) messages in both directions, message/package IDs and keep alive (heartbeat) > > > > - unit tests, for packet reading/writing and for clients against 3 publicly available sandbox/test brokers > > > > Basically, the code works but needs polishing and maturing. Also, it might be useful to experiment with alternative client API design. Not all features are yet implemented. It would also be nice to implement an actual server/broker, not to replace production quality servers, but as a proof of concept and tools during development. > > > > > > Code > > > > http://smalltalkhub.com/#!/~SvenVanCaekenberghe/MQTT/ > > > > Right now, documentation is limited, but there are class comments and the most important public API methods are commented too. There is no Metacello configuration yet, just load the 3 packages. > > > > > > References/Links > > > > - http://mqtt.org > > - https://en.wikipedia.org/wiki/MQTT > > - http://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html > > - https://github.com/mqtt/mqtt.github.io/wiki/software > > - http://mosquitto.org > > - https://github.com/emqtt/emqttd > > - http://kamilfb.github.io/mqtt-spy/ > > - https://github.com/eclipse/paho.mqtt-spy/wiki > > - https://eclipse.org/paho/ > > - https://eclipse.org/paho/clients/c/embedded/ > > - http://www.rabbitmq.com/mqtt.html > > - https://vernemq.com > > - https://www.ibm.com/developerworks/community/blogs/c565c720-fe84-4f63-873f-607d87787327/entry/mqtt_security > > > > > > Sandbox/test Servers/Brokers > > > > iot.eclipse.org:1883 > > test.mosquitto.org:1883 > > broker.mqtt-dashboard.com:1883 > > > > > > Sven > > > > -- > > Sven Van Caekenberghe > > http://stfx.eu > > Smalltalk is the Red Pill > > > > > > > > > |
|
|
You might like to collaborate with Tim and Craig.
http://forum.world.st/Graphing-weather-data-td4927701.html cheers -ben On Wed, Dec 21, 2016 at 10:17 PM, Sven Van Caekenberghe <[hidden email]> wrote: > >> On 21 Dec 2016, at 15:11, [hidden email] wrote: >> >> FWIW get yourself an account at http://enco.io > > OK, I'll put it on my list of things to look at. > >> The devportal has mqtt endpoints so that we can integrate with that and possibly have some market place solutions offered and running with Pharo. >> >> I have a small EnCoClient class in Pharo that works. Now making the API support more extensive. >> >> Docs: http://docs.enco.io/docs/device-sensor-apis >> >> Phil >> >> On Wed, Dec 21, 2016 at 2:59 PM, Sven Van Caekenberghe <[hidden email]> wrote: >> >> > On 21 Dec 2016, at 12:50, [hidden email] wrote: >> > >> > Sven, >> > >> > I am currently working with IoT and MQTT is necessary. At this point I use mosquitto as the broker and your Pharo client is nice to have and I will test it and give you feedback. >> >> That is good to here, Phil. >> >> Yes, such real world testing is much needed. This is actually quite fun stuff to experiment with. Let me know how it goes. >> >> Sven >> >> > I've got a working microPython codebase so can compare what is in there (this for the LoPy device): >> > >> > Thx for the module! >> > >> > Phil >> > >> > On Wed, Dec 21, 2016 at 12:15 PM, Sven Van Caekenberghe <[hidden email]> wrote: >> > Hi, >> > >> > MQTT is a light-weight publish/subscribe messaging protocol, originally created around 1998. It is now an official open industry ISO standard. It is perfect for large-scale Internet of Things applications and high performance mobile messaging. >> > >> > The publish/subscribe messaging pattern requires a message broker. The broker is responsible for distributing messages to interested clients based on the topic of a message. Parties communicating with each other over MQTT would all be clients in different roles, like producers and consumers, using the broker as middleware. >> > >> > Many client libraries for different programming languages and multiple brokers/servers are available. Facebook Messenger and Amazon AWS IOT are two users, among many others. >> > >> > >> > A good client library for Pharo was not yet available. I started a new MQTT project and I am looking for collaborators to help me finish it. The official specification is quite readable and there is a lot of information available (see the References/Links section at the end). >> > >> > >> > Right now, a first version of the following features is available: >> > >> > - reading & writing of all 14 binary packet types >> > >> > - an experimental client with support for connection open/close, ping, subscribe/unsubscribe, QoS levels 0 (at most once), 1 (at least once) and 2 (exactly once) for application (publish) messages in both directions, message/package IDs and keep alive (heartbeat) >> > >> > - unit tests, for packet reading/writing and for clients against 3 publicly available sandbox/test brokers >> > >> > Basically, the code works but needs polishing and maturing. Also, it might be useful to experiment with alternative client API design. Not all features are yet implemented. It would also be nice to implement an actual server/broker, not to replace production quality servers, but as a proof of concept and tools during development. >> > >> > >> > Code >> > >> > http://smalltalkhub.com/#!/~SvenVanCaekenberghe/MQTT/ >> > >> > Right now, documentation is limited, but there are class comments and the most important public API methods are commented too. There is no Metacello configuration yet, just load the 3 packages. >> > >> > >> > References/Links >> > >> > - http://mqtt.org >> > - https://en.wikipedia.org/wiki/MQTT >> > - http://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html >> > - https://github.com/mqtt/mqtt.github.io/wiki/software >> > - http://mosquitto.org >> > - https://github.com/emqtt/emqttd >> > - http://kamilfb.github.io/mqtt-spy/ >> > - https://github.com/eclipse/paho.mqtt-spy/wiki >> > - https://eclipse.org/paho/ >> > - https://eclipse.org/paho/clients/c/embedded/ >> > - http://www.rabbitmq.com/mqtt.html >> > - https://vernemq.com >> > - https://www.ibm.com/developerworks/community/blogs/c565c720-fe84-4f63-873f-607d87787327/entry/mqtt_security >> > >> > >> > Sandbox/test Servers/Brokers >> > >> > iot.eclipse.org:1883 >> > test.mosquitto.org:1883 >> > broker.mqtt-dashboard.com:1883 >> > >> > >> > Sven >> > >> > -- >> > Sven Van Caekenberghe >> > http://stfx.eu >> > Smalltalk is the Red Pill >> > >> > >> > >> >> >> > > |
|
|
In reply to this post by Sven Van Caekenberghe-2
> On 21 Dec 2016, at 14:57 , Sven Van Caekenberghe <[hidden email]> wrote: > > Hendrik, > > Thank you for this detailed feedback. > >> On 21 Dec 2016, at 12:44, Henrik Johansen <[hidden email]> wrote: >> >> Hi Sven! >> One thing I noticed when testing the RabbitMQ client with keepalive > 0, was connection being closed and all subscriptions lost when receiving large payloads, due to timeout before the keepalive packet could be sent before waiting to receive next object. > > Indeed, I used my Stamp (STOMP) Rabbit MQ client as a model for the MQTT one - there is a lot of similarity, especially in concept. > > Keep alive processing is not that easy. I tried to do it by using read timeouts as a source of regular opportunities to check for the need to process keep alive logic. But of course, if you have no outstanding read (in a loop), that won't work. > > The fact that receiving a large payload would trigger an actual keep alive time out is not something that I have seen myself. It seems weird that the reading/transferring of incoming data would not count as activity against keep alive, no ? I never had a chance to investigate fully, but I distinctly remember having the same reaction! It's quite awhile ago now, so my memory might be hazy, take the following with an appropriate amount of grains of salt. The first times I encountered it, it seemed quite random, occuring after extended periods of client inactivity after receiving only small payloads... Setting a much shorter keepalive timeout than the default was/is very useful in reproducing/verifying if it is an issue. The timeouts then occurred relatively shortly after I'd received a single payload with no other activity, and disappeared once I removed the resetting of lastActivity timestamp on reads, indicating that for at least rabbitmq (3.5 was the version at the time, I believe), receiving data was *not* being counted as keep-alive activity. The issue of receiving large payloads blocking writing in time was still unresolved, consistently cut off in the middle of (its own!) multi-MB payload transfers due to keepalive packet not being sent. I couldn't see a solution other than abandoning the elegant single-threaded approach and do keepalive as a separate high-priority process, but the architectural choice for the app changed at this point to not include a MQ in first delivery, so the more involved rewrite for doing so before deploying in production, kinda got stranded :/ Cheers, Henry |
|
|
> On 21 Dec 2016, at 16:04, Henrik Johansen <[hidden email]> wrote: > >> >> On 21 Dec 2016, at 14:57 , Sven Van Caekenberghe <[hidden email]> wrote: >> >> Hendrik, >> >> Thank you for this detailed feedback. >> >>> On 21 Dec 2016, at 12:44, Henrik Johansen <[hidden email]> wrote: >>> >>> Hi Sven! >>> One thing I noticed when testing the RabbitMQ client with keepalive > 0, was connection being closed and all subscriptions lost when receiving large payloads, due to timeout before the keepalive packet could be sent before waiting to receive next object. >> >> Indeed, I used my Stamp (STOMP) Rabbit MQ client as a model for the MQTT one - there is a lot of similarity, especially in concept. >> >> Keep alive processing is not that easy. I tried to do it by using read timeouts as a source of regular opportunities to check for the need to process keep alive logic. But of course, if you have no outstanding read (in a loop), that won't work. >> >> The fact that receiving a large payload would trigger an actual keep alive time out is not something that I have seen myself. It seems weird that the reading/transferring of incoming data would not count as activity against keep alive, no ? > > I never had a chance to investigate fully, but I distinctly remember having the same reaction! > It's quite awhile ago now, so my memory might be hazy, take the following with an appropriate amount of grains of salt. > > The first times I encountered it, it seemed quite random, occuring after extended periods of client inactivity after receiving only small payloads... > Setting a much shorter keepalive timeout than the default was/is very useful in reproducing/verifying if it is an issue. > The timeouts then occurred relatively shortly after I'd received a single payload with no other activity, and disappeared once I removed the resetting of lastActivity timestamp on reads, indicating that for at least rabbitmq (3.5 was the version at the time, I believe), receiving data was *not* being counted as keep-alive activity. > > The issue of receiving large payloads blocking writing in time was still unresolved, consistently cut off in the middle of (its own!) multi-MB payload transfers due to keepalive packet not being sent. > I couldn't see a solution other than abandoning the elegant single-threaded approach and do keepalive as a separate high-priority process, but the architectural choice for the app changed at this point to not include a MQ in first delivery, so the more involved rewrite for doing so before deploying in production, kinda got stranded :/ Instinctively it feels like messaging and multi-MB payloads would not be a good fit, at least I would suspect that they are an edge case. But maybe I am wrong. I think that the code in StampMedium>>#readBodyBytes: and StampMedium>>#readBodyString: could be refactored to use a loop with chunk buffers and at the same time check the elapsed time to fire back keep alive pings if necessary. A ping too much would not hurt, better safe than sorry. But is is hard to catch any/all network slowdowns, any IO operation could hang/timeout. But all this would assume that the problematic situation can be recreated. > Cheers, > Henry |
|
|
In reply to this post by Ben Coman
> On 21 Dec 2016, at 15:47, Ben Coman <[hidden email]> wrote: > > You might like to collaborate with Tim and Craig. > http://forum.world.st/Graphing-weather-data-td4927701.html > cheers -ben Hmm, that code seems to be quite incomplete (from what I can see publicly): not all packet types and their IO is fully implemented, there are no unit tests, there is no client, and none of the more complex quality of service levels 0, 1 & 2 logic/flow, and there are no functional unit tests against public sandbox/test servers. > On Wed, Dec 21, 2016 at 10:17 PM, Sven Van Caekenberghe <[hidden email]> wrote: >> >>> On 21 Dec 2016, at 15:11, [hidden email] wrote: >>> >>> FWIW get yourself an account at http://enco.io >> >> OK, I'll put it on my list of things to look at. >> >>> The devportal has mqtt endpoints so that we can integrate with that and possibly have some market place solutions offered and running with Pharo. >>> >>> I have a small EnCoClient class in Pharo that works. Now making the API support more extensive. >>> >>> Docs: http://docs.enco.io/docs/device-sensor-apis >>> >>> Phil >>> >>> On Wed, Dec 21, 2016 at 2:59 PM, Sven Van Caekenberghe <[hidden email]> wrote: >>> >>>> On 21 Dec 2016, at 12:50, [hidden email] wrote: >>>> >>>> Sven, >>>> >>>> I am currently working with IoT and MQTT is necessary. At this point I use mosquitto as the broker and your Pharo client is nice to have and I will test it and give you feedback. >>> >>> That is good to here, Phil. >>> >>> Yes, such real world testing is much needed. This is actually quite fun stuff to experiment with. Let me know how it goes. >>> >>> Sven >>> >>>> I've got a working microPython codebase so can compare what is in there (this for the LoPy device): >>>> >>>> Thx for the module! >>>> >>>> Phil >>>> >>>> On Wed, Dec 21, 2016 at 12:15 PM, Sven Van Caekenberghe <[hidden email]> wrote: >>>> Hi, >>>> >>>> MQTT is a light-weight publish/subscribe messaging protocol, originally created around 1998. It is now an official open industry ISO standard. It is perfect for large-scale Internet of Things applications and high performance mobile messaging. >>>> >>>> The publish/subscribe messaging pattern requires a message broker. The broker is responsible for distributing messages to interested clients based on the topic of a message. Parties communicating with each other over MQTT would all be clients in different roles, like producers and consumers, using the broker as middleware. >>>> >>>> Many client libraries for different programming languages and multiple brokers/servers are available. Facebook Messenger and Amazon AWS IOT are two users, among many others. >>>> >>>> >>>> A good client library for Pharo was not yet available. I started a new MQTT project and I am looking for collaborators to help me finish it. The official specification is quite readable and there is a lot of information available (see the References/Links section at the end). >>>> >>>> >>>> Right now, a first version of the following features is available: >>>> >>>> - reading & writing of all 14 binary packet types >>>> >>>> - an experimental client with support for connection open/close, ping, subscribe/unsubscribe, QoS levels 0 (at most once), 1 (at least once) and 2 (exactly once) for application (publish) messages in both directions, message/package IDs and keep alive (heartbeat) >>>> >>>> - unit tests, for packet reading/writing and for clients against 3 publicly available sandbox/test brokers >>>> >>>> Basically, the code works but needs polishing and maturing. Also, it might be useful to experiment with alternative client API design. Not all features are yet implemented. It would also be nice to implement an actual server/broker, not to replace production quality servers, but as a proof of concept and tools during development. >>>> >>>> >>>> Code >>>> >>>> http://smalltalkhub.com/#!/~SvenVanCaekenberghe/MQTT/ >>>> >>>> Right now, documentation is limited, but there are class comments and the most important public API methods are commented too. There is no Metacello configuration yet, just load the 3 packages. >>>> >>>> >>>> References/Links >>>> >>>> - http://mqtt.org >>>> - https://en.wikipedia.org/wiki/MQTT >>>> - http://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html >>>> - https://github.com/mqtt/mqtt.github.io/wiki/software >>>> - http://mosquitto.org >>>> - https://github.com/emqtt/emqttd >>>> - http://kamilfb.github.io/mqtt-spy/ >>>> - https://github.com/eclipse/paho.mqtt-spy/wiki >>>> - https://eclipse.org/paho/ >>>> - https://eclipse.org/paho/clients/c/embedded/ >>>> - http://www.rabbitmq.com/mqtt.html >>>> - https://vernemq.com >>>> - https://www.ibm.com/developerworks/community/blogs/c565c720-fe84-4f63-873f-607d87787327/entry/mqtt_security >>>> >>>> >>>> Sandbox/test Servers/Brokers >>>> >>>> iot.eclipse.org:1883 >>>> test.mosquitto.org:1883 >>>> broker.mqtt-dashboard.com:1883 >>>> >>>> >>>> Sven >>>> >>>> -- >>>> Sven Van Caekenberghe >>>> http://stfx.eu >>>> Smalltalk is the Red Pill >>>> >>>> >>>> >>> >>> >>> >> >> > |
|
|
In reply to this post by Sven Van Caekenberghe-2
Absolutely, far from certain the MQTT server implementations work the same. This is the test code I used for rabbit: consumer1 := StampClient new timeout:1; heartbeat: 5000; login: 'guest'; passcode: 'guest'; open; subscribeTo: '/exchange/testall'; yourself. [[ consumer1 runWith:[:message | Transcript crShow: '1 read: ', message body]] ensure: [consumer1 close]] forkAt: Processor userInterruptPriority . producer := StampClient new login: 'guest'; passcode: 'guest'; open; yourself. [1 to: 10000 do: [:i | producer sendText: i asString to: '/exchange/testall']] forkAt: Processor systemBackgroundPriority. Sometimes, but not always, this would fail, either during processing, or some time after. |
|
|
> On 21 Dec 2016, at 18:19, Henrik Johansen <[hidden email]> wrote: > >> >> On 21 Dec 2016, at 16:52 , Sven Van Caekenberghe <[hidden email]> wrote: >> >>> >>> On 21 Dec 2016, at 16:04, Henrik Johansen <[hidden email]> wrote: >>> >>>> >>>> On 21 Dec 2016, at 14:57 , Sven Van Caekenberghe <[hidden email]> wrote: >>>> >>>> Hendrik, >>>> >>>> Thank you for this detailed feedback. >>>> >>>>> On 21 Dec 2016, at 12:44, Henrik Johansen <[hidden email]> wrote: >>>>> >>>>> Hi Sven! >>>>> One thing I noticed when testing the RabbitMQ client with keepalive > 0, was connection being closed and all subscriptions lost when receiving large payloads, due to timeout before the keepalive packet could be sent before waiting to receive next object. >>>> >>>> Indeed, I used my Stamp (STOMP) Rabbit MQ client as a model for the MQTT one - there is a lot of similarity, especially in concept. >>>> >>>> Keep alive processing is not that easy. I tried to do it by using read timeouts as a source of regular opportunities to check for the need to process keep alive logic. But of course, if you have no outstanding read (in a loop), that won't work. >>>> >>>> The fact that receiving a large payload would trigger an actual keep alive time out is not something that I have seen myself. It seems weird that the reading/transferring of incoming data would not count as activity against keep alive, no ? >>> >>> I never had a chance to investigate fully, but I distinctly remember having the same reaction! >>> It's quite awhile ago now, so my memory might be hazy, take the following with an appropriate amount of grains of salt. >>> >>> The first times I encountered it, it seemed quite random, occuring after extended periods of client inactivity after receiving only small payloads... >>> Setting a much shorter keepalive timeout than the default was/is very useful in reproducing/verifying if it is an issue. >>> The timeouts then occurred relatively shortly after I'd received a single payload with no other activity, and disappeared once I removed the resetting of lastActivity timestamp on reads, indicating that for at least rabbitmq (3.5 was the version at the time, I believe), receiving data was *not* being counted as keep-alive activity. >>> >>> The issue of receiving large payloads blocking writing in time was still unresolved, consistently cut off in the middle of (its own!) multi-MB payload transfers due to keepalive packet not being sent. >>> I couldn't see a solution other than abandoning the elegant single-threaded approach and do keepalive as a separate high-priority process, but the architectural choice for the app changed at this point to not include a MQ in first delivery, so the more involved rewrite for doing so before deploying in production, kinda got stranded :/ >> >> Instinctively it feels like messaging and multi-MB payloads would not be a good fit, at least I would suspect that they are an edge case. But maybe I am wrong. >> >> I think that the code in StampMedium>>#readBodyBytes: and StampMedium>>#readBodyString: could be refactored to use a loop with chunk buffers and at the same time check the elapsed time to fire back keep alive pings if necessary. A ping too much would not hurt, better safe than sorry. But is is hard to catch any/all network slowdowns, any IO operation could hang/timeout. >> >> But all this would assume that the problematic situation can be recreated. > > Absolutely, far from certain the MQTT server implementations work the same. > > This is the test code I used for rabbit: > > consumer1 := StampClient new > timeout:1; > heartbeat: 5000; > login: 'guest'; > passcode: 'guest'; > open; > subscribeTo: '/exchange/testall'; > yourself. > [[ consumer1 runWith:[:message | Transcript crShow: '1 read: ', message body]] ensure: [consumer1 close]] forkAt: Processor userInterruptPriority . > > producer := StampClient new > login: 'guest'; > passcode: 'guest'; > open; yourself. > > [1 to: 10000 do: [:i | producer sendText: i asString to: '/exchange/testall']] forkAt: Processor systemBackgroundPriority. > > Sometimes, but not always, this would fail, either during processing, or some time after. > > A screenshot of how it looks when a disconnect happens during processing due to heartbeat failing can be found at: > https://tresor.it/s#HN4RG9PEX1SYyKy4NcM31w I installed RabbitMQ 3.6.4 on my Mac and tried your example and it succeeded ;-) But I modified it slightly so that both threads have a stop condition (I also used /queue instead of /exchange but I don't know if that makes a difference). limit := 10000. consumer1 := StampClient new timeout:1; heartbeat: 5000; login: 'guest'; passcode: 'guest'; open; subscribeTo: '/queue/testall'; yourself. producer := StampClient new login: 'guest'; passcode: 'guest'; open; yourself. [ [ consumer1 runWith: [ :message | Transcript crShow: '1 read: ', message body. message body = limit asString ifTrue: [ ConnectionClosed signal ] ] ] ensure: [ consumer1 close ] ] forkAt: Processor userInterruptPriority. [ 1 to: limit do: [ :i | producer sendText: i asString to: '/queue/testall' ]. producer close ] forkAt: Processor systemBackgroundPriority. You did see the #testSimpleWorkQueue unit test ? And the StampBenchmarkTests ? You could also try running StampBenchmark manually with higher loads. What I mean is that I normally use the #writeWithReceipt: and #clientIndividualAck subscriptions with explicit ack/nack responses. The reason is that the way you do it (although allowed and the fastest way), there is no traffic flow regulation: the producer basically sends everything in one big go (because it does not wait for receipts from the server), the consumer gets all messages in one big batch from the broker - socket stream buffers at different levels will just fill up. All this does not yet explain your problem, I know. I also did a checkin because I still had this modification in my image: === Name: Stamp-SvenVanCaekenberghe.24 Author: SvenVanCaekenberghe Time: 21 December 2016, 9:54:27.217571 pm UUID: ee3a7276-6e00-4143-9c3e-03a320c14ea6 Ancestors: Stamp-HolgerHansPeterFreyther.23 StampClient - minor refactoring in #writeHeartbeatIfNeeded (#wasInactive ForTooLong) - add #touch call to #writeNoFlush: === Just to make sure we're on the same version (I am still using Pharo 4 for this, BTW). > Cheers, > Henry |
|
|
In reply to this post by Sven Van Caekenberghe-2
Hi Sven,
yes, the code in SqueakSource/MQTT is very early stuff. It's improving quickly though. I'd be happy to get together to work on an as common as possible versions for both Squeak and Pharo. |
|
|
In reply to this post by Sven Van Caekenberghe-2
Back on topic, MQTT, I added more/better functional units tests and more comments, as well as variants against a local broker.
Here is a more complete example: "Fork a process that reads 10 temperature sensor values" [ (Array streamContents: [ :stream | | count | count := 1. MQTTExperimentalClient new clientId: 'listener'; open; subscribeToTopic: '/device/42/outside-temperature'; runWith: [ :message | stream nextPut: message contents asInteger. (count := count + 1) > 10 ifTrue: [ ConnectionClosed signal ] ] ]) inspect ] fork. "Post 10 temperature sensor value readings, using a new client/connection each time" 1 to: 10 do: [ :i | MQTTExperimentalClient new atLeastOnce; open; sendMessage: i asByteArray toTopic: '/device/42/outside-temperature'; close ] Sven > On 21 Dec 2016, at 21:57, Sven Van Caekenberghe <[hidden email]> wrote: > >> >> On 21 Dec 2016, at 18:19, Henrik Johansen <[hidden email]> wrote: >> >>> >>> On 21 Dec 2016, at 16:52 , Sven Van Caekenberghe <[hidden email]> wrote: >>> >>>> >>>> On 21 Dec 2016, at 16:04, Henrik Johansen <[hidden email]> wrote: >>>> >>>>> >>>>> On 21 Dec 2016, at 14:57 , Sven Van Caekenberghe <[hidden email]> wrote: >>>>> >>>>> Hendrik, >>>>> >>>>> Thank you for this detailed feedback. >>>>> >>>>>> On 21 Dec 2016, at 12:44, Henrik Johansen <[hidden email]> wrote: >>>>>> >>>>>> Hi Sven! >>>>>> One thing I noticed when testing the RabbitMQ client with keepalive > 0, was connection being closed and all subscriptions lost when receiving large payloads, due to timeout before the keepalive packet could be sent before waiting to receive next object. >>>>> >>>>> Indeed, I used my Stamp (STOMP) Rabbit MQ client as a model for the MQTT one - there is a lot of similarity, especially in concept. >>>>> >>>>> Keep alive processing is not that easy. I tried to do it by using read timeouts as a source of regular opportunities to check for the need to process keep alive logic. But of course, if you have no outstanding read (in a loop), that won't work. >>>>> >>>>> The fact that receiving a large payload would trigger an actual keep alive time out is not something that I have seen myself. It seems weird that the reading/transferring of incoming data would not count as activity against keep alive, no ? >>>> >>>> I never had a chance to investigate fully, but I distinctly remember having the same reaction! >>>> It's quite awhile ago now, so my memory might be hazy, take the following with an appropriate amount of grains of salt. >>>> >>>> The first times I encountered it, it seemed quite random, occuring after extended periods of client inactivity after receiving only small payloads... >>>> Setting a much shorter keepalive timeout than the default was/is very useful in reproducing/verifying if it is an issue. >>>> The timeouts then occurred relatively shortly after I'd received a single payload with no other activity, and disappeared once I removed the resetting of lastActivity timestamp on reads, indicating that for at least rabbitmq (3.5 was the version at the time, I believe), receiving data was *not* being counted as keep-alive activity. >>>> >>>> The issue of receiving large payloads blocking writing in time was still unresolved, consistently cut off in the middle of (its own!) multi-MB payload transfers due to keepalive packet not being sent. >>>> I couldn't see a solution other than abandoning the elegant single-threaded approach and do keepalive as a separate high-priority process, but the architectural choice for the app changed at this point to not include a MQ in first delivery, so the more involved rewrite for doing so before deploying in production, kinda got stranded :/ >>> >>> Instinctively it feels like messaging and multi-MB payloads would not be a good fit, at least I would suspect that they are an edge case. But maybe I am wrong. >>> >>> I think that the code in StampMedium>>#readBodyBytes: and StampMedium>>#readBodyString: could be refactored to use a loop with chunk buffers and at the same time check the elapsed time to fire back keep alive pings if necessary. A ping too much would not hurt, better safe than sorry. But is is hard to catch any/all network slowdowns, any IO operation could hang/timeout. >>> >>> But all this would assume that the problematic situation can be recreated. >> >> Absolutely, far from certain the MQTT server implementations work the same. >> >> This is the test code I used for rabbit: >> >> consumer1 := StampClient new >> timeout:1; >> heartbeat: 5000; >> login: 'guest'; >> passcode: 'guest'; >> open; >> subscribeTo: '/exchange/testall'; >> yourself. >> [[ consumer1 runWith:[:message | Transcript crShow: '1 read: ', message body]] ensure: [consumer1 close]] forkAt: Processor userInterruptPriority . >> >> producer := StampClient new >> login: 'guest'; >> passcode: 'guest'; >> open; yourself. >> >> [1 to: 10000 do: [:i | producer sendText: i asString to: '/exchange/testall']] forkAt: Processor systemBackgroundPriority. >> >> Sometimes, but not always, this would fail, either during processing, or some time after. >> >> A screenshot of how it looks when a disconnect happens during processing due to heartbeat failing can be found at: >> https://tresor.it/s#HN4RG9PEX1SYyKy4NcM31w > > I installed RabbitMQ 3.6.4 on my Mac and tried your example and it succeeded ;-) But I modified it slightly so that both threads have a stop condition (I also used /queue instead of /exchange but I don't know if that makes a difference). > > limit := 10000. > > consumer1 := StampClient new > timeout:1; > heartbeat: 5000; > login: 'guest'; > passcode: 'guest'; > open; > subscribeTo: '/queue/testall'; > yourself. > > producer := StampClient new > login: 'guest'; > passcode: 'guest'; > open; > yourself. > > [ [ consumer1 runWith: [ :message | Transcript crShow: '1 read: ', message body. message body = limit asString ifTrue: [ ConnectionClosed signal ] ] ] ensure: [ consumer1 close ] ] forkAt: Processor userInterruptPriority. > > [ 1 to: limit do: [ :i | producer sendText: i asString to: '/queue/testall' ]. producer close ] forkAt: Processor systemBackgroundPriority. > > > You did see the #testSimpleWorkQueue unit test ? > And the StampBenchmarkTests ? > You could also try running StampBenchmark manually with higher loads. > > > What I mean is that I normally use the #writeWithReceipt: and #clientIndividualAck subscriptions with explicit ack/nack responses. The reason is that the way you do it (although allowed and the fastest way), there is no traffic flow regulation: the producer basically sends everything in one big go (because it does not wait for receipts from the server), the consumer gets all messages in one big batch from the broker - socket stream buffers at different levels will just fill up. > > All this does not yet explain your problem, I know. > > > I also did a checkin because I still had this modification in my image: > > === > Name: Stamp-SvenVanCaekenberghe.24 > Author: SvenVanCaekenberghe > Time: 21 December 2016, 9:54:27.217571 pm > UUID: ee3a7276-6e00-4143-9c3e-03a320c14ea6 > Ancestors: Stamp-HolgerHansPeterFreyther.23 > > StampClient > - minor refactoring in #writeHeartbeatIfNeeded (#wasInactive ForTooLong) > - add #touch call to #writeNoFlush: > === > > Just to make sure we're on the same version (I am still using Pharo 4 for this, BTW). > >> Cheers, >> Henry |
|
Administrator
|
In reply to this post by timrowledge
Cheers,
Sean |
«
Return to Pharo Smalltalk Developers
|
1 view|%1 views
| Free forum by Nabble | Edit this page |