I'm going to assume you just meant to say producers instead of
consumers everywhere in the previous mail? Not that it perhaps matters
here.
It isn't invalid from a client or protocol perspective to create a
session and then not create a producer or consumer on it, though it
wouldn't be particularl
Of course, producers are needed, not consumers.
On Tue, Jul 5, 2016 at 11:13 AM, Dale Green
wrote:
> Hi Robbie,
>
> It seems that I made multiple mistakes in evaluating the problems, but I
> finally managed to reproduce it with some very simple code.
>
> It looks that this timeout after 5 minute
Hi Robbie,
It seems that I made multiple mistakes in evaluating the problems, but I
finally managed to reproduce it with some very simple code.
It looks that this timeout after 5 minutes happens always when there is no
consumer using the session. That is, we have only Open, Begin and Empty
frames
Hi Dale,
There doesnt seem to be much more to try on the client side itself.
The only other thing you could do would be to use a modified local
copy of Proton to ignore the servers advertised 4 minute timeout value
and use something else for remoteIdleTimeout in its 'tick'
calculations to try and
Hi Robbie,
Thank you for your efforts!
I hardcoded this scheduling to be 30 sec. Then the behavior is as follows:
- the first empty packet is sent after 2.5 min;
- after that, the heartbeats are sent every 2 min.
However, this doesn't help, as the broker closes the connections (all but
the first
Hi Dale,
I tried to reproduce this on Friday using the existing idle-timeout
tests for the client and wasn't able to, the first heartbeat was sent
after 2 minutes, but upon a further look today I see why and tweaking
the test it now goes out after 4 and I can see what is going on.
The way the cli
Hi Robbie,
I was debugging a bit and saw that the broker behaves differently if there
is a single connection or multiple connections. That is, when I have only
one connection, it is not closed even after 5 minutes without heartbeats.
So, I'll try to observe the behavior with 2 or more connections
Hi Dale,
It looks like you might have snipped some of the logs. If you removed
any frame logging we will need to see that too (well, at least what
frames and when) to better reason about what the client is (or is not)
doing, as heartbeat frames are only sent in the absence of other
frames. It woul
Hi people,
I have another problem with Service Bus and Qpid JMS 0.9.0.
The broker uses 4 minutes idle timeout. With the sample code in the
examples, I verified that the client sends the first empty frame exactly 4
minutes after the consumer has connected. After that, Qpid sends an empty
frame eve
