On Wed, Jun 10, 2015 at 8:40 AM, Matt Broadstone <[email protected]> wrote:

>
>
> On Wed, Jun 10, 2015 at 7:09 AM, Gordon Sim <[email protected]> wrote:
>
>> On 06/10/2015 12:19 AM, Matt Broadstone wrote:
>>
>>> On Tue, Jun 9, 2015 at 6:11 PM, Matt Broadstone <[email protected]>
>>> wrote:
>>>
>>>  Hi all,
>>>>
>>>> I seem to be running into a bug with the cpp broker v0.32 (installed
>>>> from
>>>> mcpierce's ubuntu packages) regarding detach frames using a node library
>>>> I'm working on.
>>>>
>>>> I can verify that we are encoding the detach frame appropriately, and
>>>> running wireshark on the box with qpidd on it indicates that a detach
>>>> frame
>>>> has made it over with the expected content. qpidd, however, seems to do
>>>> nothing with the frame whatsoever. I've run with both debug+ and trace+
>>>> log
>>>> levels enabled, and see nothing related to detaching a frame.
>>>>
>>>> Is there an existing bug for this? Or perhaps some sort of divergence
>>>> from
>>>> the spec?
>>>>
>>>> Regards,
>>>>
>>>>  Matt Broadstone
>>>
>>>>
>>>>
>>> False alarm, it turns out the "closed" flag was not actually being sent
>>> per
>>> spec! Operator error as usual, apologies for the noise.
>>>
>>
>> Funnily enough I hit the exact same thing yesterday. It is still a bug
>> somewhere as the broker should respond with a detach regardless of the
>> closed flag.
>>
>>
> Yeah, I thought there should at least be an error returned. I'm actually
> finding that an attempt to attach on the same channel, after a previous
> detach, results in the same behavior (client sends attach frame, nothing
> shows up on the qpid side). Have you found any other issues while
> investigating this that might be causing this behavior?
>
> Matt
>
>
Looks like this is indeed a bug. I have a piece of code that runs these
steps:

  1. connect
  2. attach link to 'amq.topic'
  3. detach link
  4. repeat steps 2 and 3
  5. disconnect

Running against qpidd I never reach step 5, after the first detach
subsequent attach calls hang. This code works just fine against ActiveMQ.

Matt


>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
>
>

Reply via email to