[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [paho-dev] Retrying message sending in the Java implementation

Thank Ian,

I lost this change from MQTT 3.1 to 3.1.1 and my .Net implementation has this feature still there.
Working on the native Vert.x MQTT client I was re-thinking about that and all the related problems but then I checked the Paho Java. You reply fixed the problem at the beginning :)
No need for such implementation.

Can you point me to the MQTT 3.1 paragraph spec where the retried are described for on going sessions and the where MQTT 3.1.1 remove that please ?

Thanks
Paolo

From: paho-dev-bounces@xxxxxxxxxxx <paho-dev-bounces@xxxxxxxxxxx> on behalf of Ian Craggs <icraggs@xxxxxxxxxxxxxxxxxxxxxxx>
Sent: Tuesday, August 8, 2017 5:45:12 PM
To: paho-dev@xxxxxxxxxxx
Subject: Re: [paho-dev] Retrying message sending in the Java implementation
 

The retry while a session is active behaviour was removed from MQTT 3.1.1 (versus MQTT 3.1).  In 3.1.1  the only time retries are attempted are at reconnection time. 


In practice, I found that this behaviour caused more problems than it solved.  The usual reason for acks taking a long time would be an overloaded system, which retrying only exacerbated.


Ian


On 08/08/2017 16:38, De Alti, Cristiano wrote:

Ciao Paolo,

  Last time I've checked, message delivery retry was not implemented by Paho and there is no consensus on wheter it should be.

If you have time you may want to read this long thread [1].


Ciao,

  Cristiano


[1] https://dev.eclipse.org/mhonarc/lists/paho-dev/msg03431.html



Da: paho-dev-bounces@xxxxxxxxxxx <paho-dev-bounces@xxxxxxxxxxx> per conto di Paolo Patierno <ppatierno@xxxxxxxx>
Inviato: martedì 8 agosto 2017 17:20:09
A: General development discussions for paho project
Oggetto: [paho-dev] Retrying message sending in the Java implementation
 

Hi guys,

am I right saying that the Eclipse Paho Java implementation doesn't have a built in feature for retrying.

I mean, for example, if a QoS 1 packet is sent (and it's alive in the outbound qos1 queue) but the related PUBACK isn't received after a certain "timeout", the client doesn't resend the same PUBLISH packet with duplicate = true but it just continues to wait for the ack.

The re-send is active "only" in case the client goes down and the state is recovered from a persistent storage so the PUBLISH is re-sent.

Is that right ?


Thanks,


Paolo Patierno
Senior Software Engineer (IoT) @ Red Hat
Microsoft MVP on Windows Embedded & IoT
Microsoft Azure Advisor 

Twitter : @ppatierno
Linkedin : paolopatierno
Blog : DevExperience


_______________________________________________
paho-dev mailing list
paho-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/paho-dev

-- 
Ian Craggs
icraggs@xxxxxxxxxx                 IBM United Kingdom
Paho Project Lead; Committer on Mosquitto