Hi,
Option 2 is afaik the most stable option.
Please also check your "durable"-setting in openstack and "replication"-policies in rabbitmq.
I would suggest to use
* durable-queue without replication
*OR*
* non-durable-queue with replication
*OR*
* do it the big-scale-way (f.e. vexxhost) and run one rabbitmq server per service and HA (failover) is done by some external service (like k8s, pacemaker, ...)
Fabian
Am Mi., 28. Apr. 2021 um 06:45 Uhr schrieb Shalabh Goel shalabhgoel13@gmail.com:
Hi,
I have been trying to configure HA for rabbitmq service for my Openstack installation. I found varying options on the Internet. I am facing some issues with rabbitmq with HAproxy.
Option 1. I want to know if victoria supports the following configuration directive for openstack services.
rabbit_hosts=rabbit1:5672,rabbit2:5672,rabbit3:5672
as told in this guide
https://docs.openstack.org/ha-guide/control-plane-stateful.html
Option 2. Can we configure as follows:
transport_url = rabbit://RABBIT_USER:RABBIT_PASS@rabbit1:5672, RABBIT_USER:RABBIT_PASS@rabbit2:5672,RABBIT_USER:RABBIT_PASS@rabbit3:5672
Option 3. I have been using this configuration in haproxy
listen rabbitmq_cluster_openstack bind controllers:5672 mode tcp balance roundrobin server controller1 controller1:5672 check inter 2000 rise 2 fall 5 server controller2 controller2:5672 backup check inter 2000 rise 2 fall 5
I am getting the following error in my log files for all the services.
2021-04-28 09:59:05.857 129188 INFO oslo.messaging._drivers.impl_rabbit [-] [bbf40d1a-e095-47ad-8629-931af485d4cf] Reconnected to AMQP server on controllers:5672 via [amqp] client with port 44818.
I have set transport_url as follows:
transport_url = rabbit://openstack:rabbit123@controllers:5672The error goes away if I change host to controller1 or controller.
Please suggest which is the recommended one of the 3.
Thanks in advance
Shalabh Goel