Slave failing to connect about activemq-artemis-helm

[tiagokos]

Odd is that slave works but second node is never started when replication factor = 2

[DanSalt]

From what I understand, most of the above is expected behaviour. The slave runs in 'passive' mode, meaning that it won't start it's connector (on port 61616) until it becomes 'active' (when the master fails). So the readiness probe (which is inspecting the live connection) will fail (which is ok, as you don't want clients trying to connect to it via the LoadBalancer).

However, we had to add the following line to the spec in the slave StatefulSet to get all slaves to deploy when using more than one replica:

podManagementPolicy: "Parallel"

This allows the deployment of Slave pods to continue even if one isn't ready. We made fixes to this and a few other things in our local version of the chart. I'm working on creating a branch/PR for @vromero to review.

Hope this helps,
Cheers,
Dan

[vromero]

@DanSalt is completely spot on. I never figured out the balance between a parallel policy (that adds risk on cluster reduction) or serial that has problems with the replicas. I'm looking forward to see Dan's PR.