Strange behavior with accelerate mode

Hi,
we are facing a problem using accelerate mode on a PlayBook .
I particular we run the playbook from a Jenkins slave (single client) to two remote server.
At the first run , all is work fine, the daemon is started and task executed. Also the following execution is ok. The problem coming when I run the PB after more of 2-3 hours, in this case the daemon on the remote server are still present and the PB logs are :

TASK: [skel | Create folder] ***********************************

attempting connection to 172.31.10.10 via the accelerated port 5099

172.31.10.10: sending request for validate_user

172.31.10.10: waiting for validate_user response

172.31.10.10: in recv_data(), waiting for the header

172.31.10.10: received nothing, bailing out

Falling back to ssh to startup accelerated mode

attempting to start up the accelerate daemon…

Is seems that is not validate the user . If I kill the daemon all is work fine again .

Looking in the remote server : /var/log/messages :

In this case we start with a clean situation, no daemon present :

Mar 9 11:19:45 md1-ss002-wik01 ansible-accelerate: Invoked with timeout=300 ipv6=False debug=4 password=NOT_LOGGING_PASSWORD minutes=3 port=5099 multi_key=False

Mar 9 11:19:45 md1-ss002-wik01 ansible-accelerate: exiting pid 29716

Mar 9 11:19:45 md1-ss002-wik01 ansible-accelerate: daemon pid 29717, writing /home/prod/.accelerate.pid

Mar 9 11:19:45 md1-ss002-wik01 ansible-accelerate: daemonizing successful

Mar 9 11:19:45 md1-ss002-wik01 ansible-accelerate: pid file written

Mar 9 11:19:45 md1-ss002-wik01 ansible-accelerate: serving!

Mar 9 11:20:15 md1-ss002-wik01 ansible-accelerate: daemon idle for 30 seconds (timeout=180)

Mar 9 11:20:45 md1-ss002-wik01 ansible-accelerate: daemon idle for 60 seconds (timeout=180)

Mar 9 11:21:15 md1-ss002-wik01 ansible-accelerate: daemon idle for 90 seconds (timeout=180)

Mar 9 11:21:45 md1-ss002-wik01 ansible-accelerate: daemon idle for 120 seconds (timeout=180)

Mar 9 11:22:15 md1-ss002-wik01 ansible-accelerate: daemon idle for 150 seconds (timeout=180)

Mar 9 11:22:45 md1-ss002-wik01 ansible-accelerate: server has been idle longer than the timeout, shutting down

If I run now [root@md1-ss002-wik01 ~]# ps -auwx | grep python

 prod     29717  0.0  1.0 296096 11004 ?        Sl   11:19   0:00 /usr/bin/python    /home/prod/.ansible/tmp/ansible-tmp-1425896385.1-120528273122498/accelerate
 root     29926  0.0  0.0 103232   832 pts/1    S+   11:29   0:00 grep python

It seems that the problem is that the daemon not shutdown correctly.

Any suggestions ?

Thanks,

Marco.