Kubernetes (AWS, Azure, GCP and on-prem)

Deploy Alma’s eBPF sensor on Kubernetes to collect runtime telemetry.

You get visibility into common protocols out of the box:

• HTTP

• PostgreSQL

• Redis

• Kafka

• and more

All traffic from your cluster to the Alma platform is TLS encrypted.

Alma does not persist sensitive data.

Alma eBPF on Kubernetes overview

Requirements

• Any live environment with traffic.

• Sensor resources (per node): up to 0.6 GB RAM + 0.6 CPU.

• Kubernetes control plane resources (total): 10 GB RAM + 10 CPUs.

Prerequisites

• Kubernetes v1.21+.

• Works with EKS (AWS), AKS (Azure), GKE (GCP), and on-prem OpenShift (plus VMware and more).

• Outbound connectivity from the cluster to the Alma endpoint on port 4317.

• Helm installed (Helm 3 recommended).

Port 4317 is commonly used for OTLP/gRPC. Confirm it’s allowed in your egress rules.

How it works

The eBPF sensor runs on your Kubernetes nodes and sends telemetry to the Alma platform.

Install with Helm

The eBPF sensor is provided as a Helm chart.

1

Install the chart

Run:

helm install alma-ebpf alma-helm/ --namespace pl --create-namespace

The example uses namespace pl. Use your internal standard if needed.

Validate the deployment

1. Confirm the namespace exists: kubectl get ns pl.

2. Confirm workloads are running: kubectl get pods -n pl.

3. Confirm the cluster can reach the Alma endpoint on port 4317.

If pods aren’t ready, start by checking: kubectl describe pod -n pl <pod>.