amqp1.0 driver - memory increase in RPC client

[Matt Simonin]

Using ombt[1] I observed a memory increase on the client side during different benchmarks.

I reproduced this observation with a small setup of :

- 1 RPC client sending messages at full rate 10^6 messages
- 5 RPC servers
- qpid-dispatch-router build from source (sha-1 22400df)
  * oslo.messaging==5.35.0 using amqp1.0 driver
  * pyngus==2.2.2
  * python-qpid-proton==0.19.0

It appears that the memory increases in both rpc-calls and rpc-casts tests.
I attached the rpc-cast case, showing the RSS memory of the RPC client in the above configuration.

Possible steps to reproduce locally using ombt-orchestrator[2]:

$) ./cli.py deploy --driver=router vagrant
$) ./cli.py test_case_1 --nbr_clients 1 --nbr_servers 5 --nbr_calls 1000000 --timeout 8000

[1]: https://github.com/kgiusti/ombt
[2]: https://github.com/msimonin/ombt-orchestrator/

[Matt Simonin (matthieu-simonin)]

Just a notice that we packaged the orchestrator so the command line interface changed a bit:

For ombt-orchestrator 1.0.1:

$) oo deploy --driver=router vagrant
$) oo test_case_1 --nbr_clients 1 --nbr_servers 5 --nbr_calls 1000000 --timeout 8000

[Launchpad Janitor (janitor)]

[Expired for oslo.messaging because there has been no activity for 60 days.]