Hello again.
As I said, we use JDBC persistence, and in a 2 instances
configuration, the two brokers compete to gain the leadership on
the DB lock table. The only special configuration of the broker
is setting the JDBC datasource to reoplace the KahaDB default
configuration.
On the clients side (Camel routes in containrized Karaf
instances and Java application, both as producers and consumers)
the brokers access URL is something like
"failover(tcp:amqservice1:port1, tcp:amqservice2:port1)" where
amqserviceX and portX are defined by our k8s services and
endpoints. Nothing more.
We've tested with batches of more than 1 million messages, and
all seems to be stable. Sometimes the leadership changes, but
clients detect it and the failover do its job.
Ephemeris Lappis
Le 13/03/2023 à 11:24, [email protected] a écrit :
split brain scenarios
--
Cet e-mail a été vérifié par le logiciel antivirus d'Avast.
www.avast.com
