Deckhouse Platform supports integration with Kaspersky Unified Monitoring and Analysis Platform (KUMA), a unified monitoring and analysis system by Kaspersky Lab. As part of the integration, security events and audit logs from the cluster are sent to KUMA for further analysis.

Sending logs to KUMA

To send logs to KUMA, configure log collection and delivery in Deckhouse Platform using the following resources:

On the KUMA side, configure the appropriate resources for receiving events.

Example ClusterLogDestination and ClusterLoggingConfig configurations

In the following examples, you can find ClusterLogDestination and ClusterLoggingConfig configuration options for sending logs in various formats.

  • In JSON format over UDP
  • In JSON format over TCP
  • In CEF format over TCP
  • In Syslog format over TCP
  • Apache Kafka
apiVersion: deckhouse.io/v1alpha1
kind: ClusterLogDestination
metadata:
  name: kuma-udp-json
spec:
  type: Socket
  socket:
    address: IP_ADDRESS:PORT # Replace as needed.
    mode: UDP
    encoding:
      codec: "JSON"
---
apiVersion: deckhouse.io/v1alpha2
kind: ClusterLoggingConfig
metadata:
  name: kubelet-audit-logs
spec:
  type: File
  file:
    include:
    - /var/log/kube-audit/audit.log
  destinationRefs:
    - kuma-udp-json
apiVersion: deckhouse.io/v1alpha1
kind: ClusterLogDestination
metadata:
  name: kuma-tcp-json
spec:
  type: Socket
  socket:
    address: IP_ADDRESS:PORT # Replace as needed.
    mode: TCP
    tcp:
      verifyCertificate: false
      verifyHostname: false
    encoding:
      codec: "JSON"
---
apiVersion: deckhouse.io/v1alpha2
kind: ClusterLoggingConfig
metadata:
  name: kubelet-audit-logs
spec:
  type: File
  file:
    include:
    - /var/log/kube-audit/audit.log
  destinationRefs:
    - kuma-tcp-json
apiVersion: deckhouse.io/v1alpha1
kind: ClusterLogDestination
metadata:
  name: kuma-tcp-cef
spec:
  type: Socket
  socket:
    extraLabels:
      cef.name: d8
      cef.severity: "1"
    address: IP_ADDRESS:PORT # Replace as needed.
    mode: TCP
    tcp:
      verifyCertificate: false
      verifyHostname: false
    encoding:
      codec: "CEF"
---
apiVersion: deckhouse.io/v1alpha2
kind: ClusterLoggingConfig
metadata:
  name: kubelet-audit-logs
spec:
  type: File
  file:
    include:
    - /var/log/kube-audit/audit.log
  logFilter:
    - field: userAgent
      operator: Regex
      values: [ "kubelet.*" ]
  destinationRefs:
    - kuma-tcp-cef
apiVersion: deckhouse.io/v1alpha1
kind: ClusterLogDestination
metadata:
  name: kuma-tcp-syslog
spec:
  type: Socket
  socket:
    address: IP_ADDRESS:PORT # Replace as needed.
    mode: TCP
    tcp:
      verifyCertificate: false
      verifyHostname: false
    encoding:
      codec: "Syslog"
---
apiVersion: deckhouse.io/v1alpha2
kind: ClusterLoggingConfig
metadata:
  name: kubelet-audit-logs
spec:
  type: File
  file:
    include:
    - /var/log/kube-audit/audit.log
  logFilter:
    - field: userAgent
      operator: Regex
      values: [ "kubelet.*" ]
  destinationRefs:
    - kuma-tcp-syslog
apiVersion: deckhouse.io/v1alpha1
kind: ClusterLogDestination
metadata:
  name: kuma-kafka
spec:
  type: Kafka
  kafka:
    bootstrapServers:
      - kafka-address:9092 # Replace as needed.
    topic: k8s-logs
---
apiVersion: deckhouse.io/v1alpha2
kind: ClusterLoggingConfig
metadata:
  name: kubelet-audit-logs
spec:
  destinationRefs:
    - kuma-kafka
  file:
    include:
      - /var/log/kube-audit/audit.log
  logFilter:
    - field: userAgent
      operator: Regex
      values:
        - kubelet.*
  type: File

KCS configuration features

Kaspersky Container Security (KCS) requires extended access rights to the OS kernel in order to monitor the execution environment. Specifically, the kcs-ih (Image Hub) component must run in privileged mode. To do this, certain process privilege management mechanisms at the Linux OS kernel level must be disabled. For more details, see the Kaspersky Container Security documentation.

To enable KCS runtime monitoring in a Deckhouse Platform cluster, assign the label security.deckhouse.io/pod-policy=privileged to the namespace where KCS components are located.

Example command for assigning a label:

d8 k label namespace kcs security.deckhouse.io/pod-policy=privileged

Antivirus scanning exclusions for nodes

If antivirus software is installed on Deckhouse Platform cluster nodes (for example, Kaspersky Endpoint Security for Linux, KESL), you may need to exclude Deckhouse service directories from scanning to avoid false positives.

List of Deckhouse service directories (also available CSV format):

Directory Purpose
/etc/cni/ (any node) CNI module configuration files
/etc/kubernetes (any node) Static pod manifests, PKI certificate files
/mnt/kubernetes-data (master node) Only present in clusters deployed in the cloud when a separate disk is used for the etcd database
/mnt/vector-data (any node, log-shipper module) Service data for log delivery status
/opt/cni/bin/ (any node) CNI module executables
/opt/deckhouse/bin/ (any node) Executables required for Deckhouse Platform operation
/var/lib/bashible (any node) Node configuration files
/var/lib/containerd (any node) Stores CRI-related data (for example, containerd). Contains container image layers, filesystem snapshots, metadata, logs, and other container information
/var/lib/deckhouse/ (master node) Deckhouse Platform module files dynamically loaded from the image registry
/var/lib/etcd (master node) etcd database
/var/lib/kubelet/ (any node) kubelet configuration files
/var/lib/upmeter (master node, upmeter module) upmeter module database
/var/log/containers (any node) Container logs (when using containerd)
/var/log/pods/ (any node) Logs of all pod containers running on the node

KESL configuration recommendations

To ensure Deckhouse Platform functions correctly with KESL installed, follow these steps:

  1. Disable the following KESL tasks:

    • Firewall Management (ID: 12)
    • Web Threat Protection (ID: 14)
    • Network Threat Protection (ID: 17)
    • Web Control (ID: 26)
    • Mail Monitoring ( ID: 35 )

    The list of tasks may differ in future KESL versions.

  2. Make sure node resources meet the requirements of:

  3. For performance optimization, follow the official Kaspersky recommendations.

Additional resources