Configure persistent storage for the ActiveGate

1-min read

The log_analytics_collector ActiveGate module utilizes disk buffers to temporarily store data. To avoid data loss across ActiveGate restarts, we recommend attaching a PersistentVolumeClaim (PVC) to the ActiveGate.

Adding a PersistentVolumeClaim

The following snippet shows how you can attach the PersistentVolumeClaim to the ActiveGate in the DynaKube.

apiVersion: dynatrace.com/v1beta5
kind: DynaKube
metadata:
  name: dynakube
  namespace: dynatrace
spec:
  activeGate:
    volumeClaimTemplate:
      accessModes:
      - ReadWriteOnce
      resources:
        requests:
          storage: 1Gi

Kubernetes environments without a default StorageClass require you to set storageClassName field. Without it, the ActiveGate pod may fail to schedule with the error: pod has unbound immediate PersistentVolumeClaims. This requirement also applies to EEC (Extensions Execution Controller) deployments.

Ensure that the storageClassName value corresponds to a valid StorageClass in your cluster:

persistentVolumeClaim:
  storageClassName: fast-ssd

Adjusting ActiveGate shutdown grace period

When the ActiveGate performs a graceful shutdown (for example, in a scale-in scenario), it needs to flush buffers to avoid data loss. In large environments, this can take longer than the default grace period of Kubernetes, which is 30s. To avoid this, setting a longer terminationGracePeriodSeconds on the ActiveGate pods can be helpful. You can change it as shown in the following snippet.

apiVersion: dynatrace.com/v1beta5
kind: DynaKube
metadata:
  name: dynakube
  namespace: dynatrace
spec:
  activeGate:
    terminationGracePeriodSeconds: 120s