Open Stack

On 12/16/2013 08:47 AM, Alvise Dorigo wrote:
> Hi,
> in this
> page http://openstack.redhat.com/Highly_Available_Qpid_for_OpenStack#Clustered_without_pacemaker 
> they explain how to make a glance-api service to connect to a redundant
> qpid broker server on two or more hosts. The parameter to use is
> qpid_hosts, but it seems to be ignored.
> 
> In fact, if I switch qpid off on the first host listed in the
> qpid_hosts, the glance-api will continue to try to connect to it
> receiving "connection refused"... and never tries to connect to the
> other one, in which qpid is still alive.
> 
> Any hint ? 
> 
> In the following the relevant part of glance-api.conf
> ==========================
> notifier_strategy=qpid
> 
> # Configuration options if sending notifications via Qpid (these are
> # the defaults)
> qpid_notification_exchange = glance
> qpid_notification_topic = notifications
> qpid_hosts=192.168.122.41:5672,192.168.122.42:5672
> #qpid_hostname = localhost
> qpid_username =
> qpid_password =
> qpid_sasl_mechanisms =
> qpid_reconnect_timeout = 0
> qpid_reconnect_limit = 0
> qpid_reconnect_interval_min = 0
> qpid_reconnect_interval_max = 0
> qpid_reconnect_interval = 0
> qpid_heartbeat=2
> # Set to 'ssl' to enable SSL
> qpid_protocol = tcp
> qpid_tcp_nodelay = True
> ==========================

I suspect qpid_hosts isn't suppoted by glance.  The qpid and rabbit code
(until very recently, icehouse) is separate code for glance
notifications from the code used by every other service for notifications.

-- 
Russell Bryant