This guide describes how the different storage volumes are used by Dynatrace Operator.
The following table shows the storage requirements of Dynatrace Operator deployment modes.
Deployment type
CSI driver disabled
CSI driver enabled
classicFullStack
2 GB for OneAgent configuration and binaries on the node root filesystem
N/A
hostMonitoring
2 GB for OneAgent configuration and binaries on the node root filesystem
2 GB for OneAgent configuration and binaries in the kubelet root directory
applicationMonitoring
1 GB per monitored pod from local ephemeral storage (for example, kubelet root directory)
1 GB per tenant and running OneAgent version in the kubelet root directory
0.1 GB per injected pod from local ephemeral storage (for example, kubelet root directory)
cloudNativeFullStack
Combination of hostMonitoring and applicationMonitoring
Combination of hostMonitoring and applicationMonitoring
The kubelet root directory is the central storage location for all directories required by the CSI driver. Allocate approximately 30 GB for the kubelet directory to cover all Dynatrace-related storage requirements.
These numbers should be increased for volatile environments and can be decreased for environments with disk space monitoring.
Depending on your setup, the kubelet root directory might not be located in /var/lib/kubelet.
Volume name
Path on the host
Permissions
Purpose
Sizing recommendation
mountpoint-dir
/var/lib/kubelet/pods
Read-write necessary
Stores node information
N/A
registration-dir
/var/lib/kubelet/plugins_registry
Read-write necessary
Contains the CSI driver "registration" socket. This socket allows the registrar to use it to register the CSI driver.
10 MB
plugin-dir, data-dir
/var/lib/kubelet/plugins/csi.oneagent.dynatrace.com
Read-write necessary
Stores all files required for the CSI driver's operation.
See plugin directory directory disk usage for all factors that influence storage consumption.
Contains the CSI driver components and enables kubelet to interact with the driver for operations like health checks.
Storage usage depends on your environment:
If OneAgent is monitoring the host additional storage is consumed.
Disk usage per version is depending on the architecture. Without CSI driver the listed amount of storage is consumed per pod.
Allows the OneAgent to persist its configuration.
Volume name
Volume Type
Host path
Purpose
volume-storage
hostPath
/var/opt/dynatrace (default) 1
Without CSI driver configuration is written directly to the host filesystem.
volume-storage
csi
/var/lib/kubelet/plugins/plugins/csi.oneagent.dynatrace.com
With CSI driver no configuration is written to an ephemeral volume, instead plugin directory on the host is used to persist OneAgent configuration.
Configurable using storageHostPath field in the Dynakube. See Dynakube parameters for more information.
These directories are specific to Dynatrace. They are injected into workloads by the Webhook and don't require user management or configuration.
Volume name
Volume type
Purpose
Sizing recommendation
oneagent-bin
emptyDir
Stores the OneAgent binary and logs.
An emptyDir is only used if CSI driver is not installed. 1
Note: you can configure emptyDir volume size limit by annotating your workloads using
volume.dynatrace.com/oneagent-bin: 1500Mi.
dynatrace-config
emptyDir
Stores configuration data and logs for the OneAgent, including custom CAs.
Stores configuration data for metadata enrichment (dt_metadata).
Note: you can configure emptyDir volume size limit by annotating your workloads using
volume.dynatrace.com/dynatrace-config: <size> 2.
The Operator puts ~5Mb (mainly affected by how many certs the user has) of config files into the dynatrace-config,
the rest is managed by the CodeModule according to file aging mechanism.
All the certs provided will be replicated for each container, so they will be present in the dynatrace-config X times.
(X is the number of injected containers in the pod).
Example: if we have a Pod with 3 containers <size> would be 3Gi, because 3GB (~2.8Gi) according to the file aging docs and 3x5Mi=15Mi for certs/config and rounding to a nice number.
Same disk usage as described in plugin directory disk usage. You can save storage by configuring the node image pull feature.
More details about the format of <size> can be found in the Kubernetes documentation.
Overview of the volume size limits for ephemeral volumes of the Dynatrace Operator components and which Helm switches can be used to configure them.
/tmp/k8s-webhook-server/serving-certswebhook.certsDir.sizeLimitThere are no default values for ephemeral storage resource configuration. They can be configured using the Helm switches shown in the following table:
Component
Container(s)
Helm value
dynatrace-operator
operator
operator.requests.ephemeral-storage
operator.limits.ephemeral-storage
dynatrace-webhook
webhook
webhook.requests.ephemeral-storage
webhook.limits.ephemeral-storage
dynatrace-csi-driver
init
csidriver.csiInit.resources.requests.ephemeral-storage
csidriver.csiInit.resources.limits.ephemeral-storage
dynatrace-csi-driver
provisioner
csidriver.provisioner.resources.requests.ephemeral-storage
csidriver.provisioner.resources.limits.ephemeral-storage
dynatrace-csi-driver
server
csidriver.server.resources.requests.ephemeral-storage
csidriver.server.resources.limits.ephemeral-storage
dynatrace-csi-driver
registrar
csidriver.registrar.resources.requests.ephemeral-storage
csidriver.registrar.resources.limits.ephemeral-storage
dynatrace-csi-driver
livenessprobe
csidriver.livenessprobe.resources.requests.ephemeral-storage
csidriver.livenessprobe.resources.limits.ephemeral-storage
codemodule-download-<hash>
codemodule-download
csidriver.job.resources.requests.ephemeral-storage
csidriver.job.resources.limits.ephemeral-storage
The Helm switches can be used in a custom values.yaml file to control the limits during Operator install with the Helm chart.