ActiveMQ Artemis failover versus clustering

1
votes

I am working on an application that has a cluster of Artemis servers. Each live server is paired with a backup server for failover. I happened on an article by Bilgin Ibryam, and it has me wondering.

If the clustered live servers are running as containers, and the orchestrator restarts any containers that die, is the failover configuration necessary? The article says "no".

So if I understand correctly, running a cluster of Artemis brokers in an environment that detects and restarts failed brokers will provide the same semantics (and similar availability) as running a cluster where each live server is paired with a backup. Is that right?

1
activemq-artemis

1 Answers

0
votes

Yes. Running a cluster of ActiveMQ Artemis brokers in an environment that detects and restarts failed brokers will provide the same semantics (and similar availability) as running a cluster where each live server is paired with a backup.

The caveat here is that the data for each broker (i.e. everything in the the broker's data directory) needs to be "stateful" so that in the event a broker dies it is restarted with the same data. As the article mentions:

For example, in the case of a node failure, Kubernetes would start a broker pod on a different node and make the same PV and data available.