[OASIS Issue Tracker] Updated: (MQTT-18) Clarifying what MQTT means by storing data.

[OASIS Issues Tracker <workgroup_mailer@lists.oasis-open.org>]

     [ http://tools.oasis-open.org/issues/browse/MQTT-18?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Andrew Banks updated MQTT-18:
-----------------------------

    Proposal: 
I suggest we define the term "store" as follows.

The client and server implement data storage independently and the duration for which data persists can be different in each. The client and server MUST store data for at least as long as the client server connection lasts. Quality of service guarantees are only valid so long as both client and server store data. Durable subscriptions and retained publications are only good for the length of time that the server stores them.

Make a global change to replace "persist" with "store".

  was:
I suggest we define the term "store" as follows.

The client and server implement data storage independently and the duration for which data persists may be different in each. The client and server MUST store data for at least as long 
as the client server connection lasts. Quality of service guarantees are only valid so long as both client and server persist data. Durable subscriptions and retained publications are only 
good for the length of time that the server can persist them.


> Clarifying what MQTT means by storing data.
> -------------------------------------------
>
>                 Key: MQTT-18
>                 URL: http://tools.oasis-open.org/issues/browse/MQTT-18
>             Project: OASIS Message Queuing Telemetry Transport (MQTT) TC
>          Issue Type: Improvement
>          Components: core
>            Reporter: Andrew Banks
>
> The MQTT Specification talks about storing state for example in terms of durable subscriptions, retained [publications and partially completed message transfers for Qos 1 and Qos 2.
> In practice server  implementations interpret storing state in a number of ways ranging from a forced write to a disk to keeping the data in volatile memory which
> is cleared if the server is restarted. This is also the case for the client implementation, though most of these seem to actually force the data to disk.
> The consequence of not storing data by forcing it to disk  in the server are:
> Loss of retained publications, loss of durable subscriptions, loss of Qos =1 publications, loss or duplication of Qos=2 publications
> In the client this would also result in loss or duplication of messages.
> The client and server are only able to detect  indirectly that the other has lost data, for example by failing to receive a retained publication.

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://tools.oasis-open.org/issues/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira