Skip to main content

MarkLogic Server on Kubernetes

Configuration

This section includes information on configurations.

ConfigMap

The HAProxy configuration is dynamically generated in ConfigMap with the name "marklogic-haproxy". A custom configuration can be used by creating a new ConfigMap and setting the existingConfigmap parameter in the values.yaml file to the name of the new ConfigMap.

Modify port configuration

By default, ports 8000, 8001, and 8002 are configured to handle HTTP traffic. The default ports can be modified in the values.yaml file. For each entry, name, type, and port must be specified. Type can be one of these values:

  • HTTP - The backend is configured as an HTTP proxy that handles HTTPC traffic. It can also be configured to handle cookie-based session affinity and multi-statement transactions from the MarkLogic client.

  • TCP - The backend is configured as a TCP proxy that handles TCP traffic.

This is the default configuration:

haproxy:
  ports:
    - name: app-service
      type: HTTP     
      port: 8000     
      targetPort: 8000
    - name: admin
      type: HTTP     
      port: 8001     
      targetPort: 8001
    - name: manage
      type: HTTP     
      port: 8002
      targetPort: 8002

To add port 8010 for HTTP load balancing, add this configuration to the values.yaml file:

    - name: my-app-1     
      type: HTTP
      port: 8010
      targetPort: 8010

To add port 5432 for TCP load balancing, add this configuration to the values.yaml file:

    - name: odbc    
      type: TCP
      port: 5432
      targetPort: 5432
Automatic reload configuration

When a change occurs in the current deployment (such as a change to the backend ports, or the number of running MarkLogic nodes), the HAProxy is restarted by default. During the restart process, the new configuration is loaded. However, the default behavior can be changed with this setting:

haproxy:
    restartWhenUpgrade:
        enabled: false

Note

if you update the number of MarkLogic nodes and restartWhenUpgrade.enabled is set to false, you must manually delete the HAProxy deployment to get the latest configuration.

Access HA Proxy

In Kubernetes, each pod is assigned its own IP address and namespace. Containers within a pod share the same namespace, IP, and MAC address as the pod. Therefore, nodes and pods can communicate with one another without using Network Address Translation (NAT). However, if a pod is restarted for any reason, it will be assigned a new IP address. Because a pod does not have a fixed IP or DNS, a service is used to connect to an application.

The HAProxy can be accessed from a service with the name of <RELEASE_NAME>-haproxy. For example, if the release name is marklogic, then the name of the service is marklogic-haproxy.

External access

By default, HAProxy is configured to provide access within the Kubernetes cluster. However, HAProxy can provide external access by setting the service type in the values.yaml file:

haproxy:  
  service:    
    type: LoadBalancer

Warning

By setting the haproxy service type to LoadBalancer, the MarkLogic endpoint is exposed to the public internet. Because of this, networkPolicy should be set to limit the sources that can visit MarkLogic.

In this section: