Before the final v0.11 of Nomad hits the streets, HashiCorp customers have been treated to a beta version of the distributed cluster and application scheduler, featuring autoscaling, CSI plugins, and more.
The next Nomad release will for example feature lifecycle hooks such as Prestart and Poststart. Those can be used to run initialisation tasks like mounting volumes and getting configurations, or run long-lived applications within batch jobs via a new sidecar parameter.
Those looking for a more bleeding edge addition should try the autoscaling preview included in the beta. The add-on component retrieves policies from the scheduler, looks into Prometheus metrics (though support for other monitoring tools is planned) and takes those as a basis for scaling operations. For now, the tool is restricted to horizontal application autoscaling, though cluster scaling is on the Nomad team’s todo list as well.
Another beta might be interesting to admins who tend to forget to update their Nomad client configurations when changing their cluster’s storage configuration. Version 0.11 comes with a csi plugin stanza which contains Controller Plugins and Node Plugins. While the former facilitates communication with each storage provider’s API, the latter allows the execution of tasks on each client node.
As soon as those are up and running, storage volumes can be registered with a nomad volume register command. This lets users deploy stateful applications using storage volumes from any provider using CSI plugins.
Nomad users with an enterprise subscription can also look forward to getting a “complete set of records for all user-issued actions in Nomad” via audit logging. Events are apparently captured with HTTP request and response information in a JSON format familiar to operators. The feature is meant to help them when looking for access anomalies or exploring cluster behaviour among other things.