HPE CSI Driver for Kubernetes 2.1.0 introduces a whole new category of functionality to stateful workloads: monitoring and alerting with Prometheus. It is important to mention upfront that the HPE Storage Array Exporter for Prometheus can be used to collect storage array metrics for any solution or workload and is not exclusive to Kubernetes and Cloud Native applications. The exporter and CSI driver is also accompanied by a HPE CSI Info Metrics Provider for Kubernetes which helps build Kubernetes centric queries of the time-series data.
In this blog post, I’ll cover the key concepts of cloud native monitoring and alerting with Prometheus, along with an overview of new features and capabilities of the newly released HPE CSI Driver for Kubernetes.
Monitoring and alerting with Prometheus
Running Kubernetes in production means so much more than just standing up a cluster and deploying your workloads. Essential tasks include central logging, security and hardening, observability, monitoring, and alerting. HPE just released HPE Storage Array Exporter for Prometheus and HPE CSI Info Metrics Provider for Prometheus. The Array Exporter is a standalone metrics exporter that may be used outside the realm of Kubernetes. The CSI Info Metrics Provider allows Kubernetes users to combine metrics collected with the Array Exporter using Prometheus labels pertinent to Kubernetes objects, such as Namespaces, Persistent Volume Claims and Persistent Volumes.
Users are now able to build dashboards with tools such as Grafana to monitor every aspect of running applications on Kubernetes along with rich storage-related metrics. The metrics differ slightly between HPE Alletra 6000 and 9000 and their respective legacy systems, but all the relevant capacity and performance metrics are available to accommodate several different use cases, such as capacity tracking and trending with watermark alerts, latency heatmaps, IOPS and throughput gauges and graphs.
Information about how to install and use the Array Exporter is available on GitHub. The Kubernetes and CSI driver centric components are available on the Storage Container Orchestration Documentation (SCOD) portal.
Check out these assets and resources to get started with the exporter from various perspectives.
A tutorial is available on HPE DEV: Get started with Prometheus and Grafana on Docker with HPE Storage Array. This is suitable for users who want to deploy all the necessary software to get a sense for how Prometheus, Grafana and the Array Exporter works. In the tutorial, the only requirement is Docker, and Kubernetes is not mandatory.
Take a look at a more intricate screencast available on YouTube that goes through all the steps needed to get everything running on Kubernetes, including basic alerting with Grafana: Get started with the HPE Storage Array Exporter for Prometheus on Kubernetes.
Improved Remote Copy Peer Persistence
You’ll see that there’s been an extensive overhaul on how to manage and operate Remote Copy Groups (RCG) using the HPE CSI Driver with HPE Alletra 9000, HPE Primera, and HPE 3PAR. RCGs provisioned with the CSI driver will now automatically have the auto_failover and auto_synchronize policy set by default, which improves resilience and availability for environments deployed with Peer Persistence.
It’s now also possible to add existing Persistent Volume Claims (PVC) into an RCG by annotating the PVC even if the Storage Class didn’t not allow RCG provisioning initially or add non-replicated PVCs to existing RCGs.
Learn more about Peer Persistence and the new parameters on SCOD.
A positive side effect of this improved capability is that volume mutation overall has become more flexible and forgiving as allowMutations no longer need a default value in the Storage Class for the parameter subject to the mutation. Parameters still needs to be present in the allowMutations though. Volume mutations is a HPE CSI Driver exclusive feature that was introduced in version 1.3.0 and an example may be found in this HPE DEV blog post: Introducing Kubernetes CSI Sidecar Containers from HPE.
Expanded ecosystem support
The 2.1.0 release includes numerous updates to the support matrix. The HPE CSI Driver now supports Kubernetes up to version 1.22. Initial formal support for VMware Tanzu Kubernetes Grid using Ubuntu 20.04 as the compute node operating system.
Red Hat OpenShift 4.8 is now fully certified and version 4.7 may be used as an upgrade path only from version 4.6. SUSE Rancher 2.6 will soon be integrated into the “Apps & Marketplace” and you’re encouraged to use the Helm chart if an upgrade to HPE CSI Driver 2.1.0 is needed immediately.
As for all partner and upstream Kubernetes, including HPE Ezmeral Runtime Enterprise with the exception of Red Hat OpenShift, validate the Kubernetes version and host operating system against the support matrix on SCOD.
Next steps
Whether you’re in a complete greenfield situation with Kubernetes or upgrading an existing deployment of the HPE CSI Driver, all the resources are at your fingertips.
- Install the HPE CSI Driver with Helm via Artifact Hub
- Check out the latest requirements for your Kubernetes deployments on SCOD
- Learn how to setup a Prometheus, Grafana and HPE Storage Array Exporter for Prometheus in this HPE DEV blog
- Watch the tutorial on how to get started with Prometheus and Grafana on Kubernetes with the HPE Storage Array Exporter and HPE CSI Info Metrics Provider for Prometheus
- View the example HPE dashboards for Grafana on grafana.com
Watch for more exciting news from HPE storage in 2022. Check back here on the Around The Storage Block for more updates in the Kubernetes and cloud native space to help fuel your business and infrastructure transformation projects.
MichaelMattsson
Data & Storage Nerd, Containers, DevOps, IT Automation
