Pod management policy

If you have a fresh deployment of nexeed-infra, you have the opportunity to set the default pod management policy to Parallel before the deployment.

This field becomes immutable after the first deployment, if you would like to set this value differently, you can perform kubectl -n shared delete sts/rabbitmq-statefulset --cascade=orphan and perform a re-deployment of nexeed-infra.

Why to set this: RabbitMQ requires all nodes to be started at the same time to re-form a cluster after a cluster-wide shutdown, however the Kubernetes statefulset default behaviour is to wait for the pods to be up in sequential order. This setting changes the statefulset behaviour to allow pods to be run in parallel.

In your custom-values.yaml, please add the following parameter to the rabbitmq module, Parallel is case-sensitive:

global:
  modules:
    rabbitmq:
      podManagementPolicy: Parallel

This feature requires utility-toolkit version >= 1.18.6

If the utility-toolkit is below 1.18.6, you need to add the following text to custom-values.yaml:

global:
  ...

# On the same level of global
rabbitmq:
  statefulSets:
    rabbitmq:
      podManagementPolicy: "Parallel" # or OrderedReady