[ 
https://issues.apache.org/jira/browse/QPIDIT-120?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Kim van der Riet reassigned QPIDIT-120:
---------------------------------------

    Assignee: Kim van der Riet

> ProtonCpp LargeContentTest Sender closes connection too soon
> ------------------------------------------------------------
>
>                 Key: QPIDIT-120
>                 URL: https://issues.apache.org/jira/browse/QPIDIT-120
>             Project: Apache QPID Interoperability Test Suite
>          Issue Type: Bug
>          Components: Proton C++ Shim
>    Affects Versions: 0.1.0
>            Reporter: Chuck Rolke
>            Assignee: Kim van der Riet
>            Priority: Major
>
> In researching the bug reported in DISPATCH-939 (router closes connection 
> with error) a problem in the test has emerged.
> The Cpp shim uses on_tracker_accept and then closes the connection when when 
> the number of confirmed messages equals the number of total messages. In 
> theory, then, the test should never close the connection before the router 
> has confirmed and accepted all the messages. From the trace in DISPATCH-939 
> the connection is closed about 2 mS after the fourth Sender message has gone 
> over the wire to Dispatch. Not enough dispositions have been received to 
> cover the number of messages sent so why has the connection been closed?
> Adding some print debugging reveals the issue. In the test as it is today the 
> _totalMessages_ is 2. However, in the send loop the actual number of messages 
> sent is 2 times the number of elements in the incoming list of test values. 
> In today's case the values list has four elements so a total of 8 messages 
> should go over the wire.
> A hack '* 4' is added to the on_tracker_accept to make the test work:
> {{if (_msgsConfirmed >= _totalMsgs * 4) {}}
> print debugging session shows:
> {{InteropTestError: Send shim 'ProtonCpp':}}
> {{on_container_start: _totalMsgs: 2}}
> {{on_sendable: msgsSent: 1}}
> {{on_sendable: msgsSent: 2}}
> {{on_sendable: msgsSent: 3}}
> {{on_sendable: msgsSent: 4}}
> {{on_sendable: msgsSent: 5}}
> {{on_sendable: msgsSent: 6}}
> {{on_sendable: msgsSent: 7}}
> {{on_sendable: msgsSent: 8}}
> {{on_tracker_accept: msgsConfirmed: 1}}
> {{on_tracker_accept: msgsConfirmed: 2}}
> {{on_tracker_accept: msgsConfirmed: 3}}
> {{on_tracker_accept: msgsConfirmed: 4}}
> {{on_tracker_accept: msgsConfirmed: 5}}
> {{on_tracker_accept: msgsConfirmed: 6}}
> {{on_tracker_accept: msgsConfirmed: 7}}
> {{on_tracker_accept: msgsConfirmed: 8}}
>  The test needs to get the test elementsList factor back into the tracker to 
> decide correctly when to close the connection.
> This test has still been valuable pointing out an issue in Dispatch that 
> needs some attention.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@qpid.apache.org
For additional commands, e-mail: dev-h...@qpid.apache.org

Reply via email to