Configure Multiple OpenELB Replicas

This document describes how to configure multiple openelb-speaker instances to ensure high availability in a production environment.

The openelb-speaker is deployed as a DaemonSet, which means an instance of openelb-speaker will be started on each node in the Kubernetes cluster. If the number of nodes is large, the number of openelb-speaker instances will also be large.

In BGP mode, all openelb-speaker instances will respond to the BgpPeer configuration and attempt to establish a BGP connection with the peer BGP router by default. If the router is configured with BGP peer information for all nodes and establishes BGP connections with all of them, it can lead to significant load on the router.To mitigate this, you can use methods such as nodeSelector, Node Affinity, or Taints and Tolerations to schedule openelb-speaker only on specific nodes. This reduces the number of BGP connections and alleviates the load on the router. It is important to ensure that at least two instances of openelb-speaker are running under each router to maintain high availability.

In Layer2 or VIP mode, if you want only certain nodes to handle traffic, you can also use nodeSelector, Node Affinity, or Taints and Tolerations to schedule openelb-speaker on specific nodes.

• If all Kubernetes cluster nodes are deployed under the same router, it is recommended to configure at least two openelb-speaker instances. These instances should be installed on two different Kubernetes cluster nodes to ensure redundancy and high availability.

• If the Kubernetes cluster nodes are deployed under different leaf routers, it is recommended to configure at least two openelb-speaker instances under each leaf router. This means one instance per node under each leaf router to ensure that each router has redundancy.

  NOTE

  • If the Kubernetes cluster nodes are deployed under different routers, you need to perform further configuration so that the openelb-speaker instances establish BGP connections with the correct BGP routers. For details, see Configure OpenELB for Multi-router Clusters. In such a network topology, only BGP mode can be used. Layer2 mode is not suitable for clusters with nodes spread across different leaf routers.

Example Configuration

Using Node Selector

First, label the target nodes:

kubectl label nodes <node-name> <label-key>=<label-value>

Example:

kubectl label nodes node1 openelb-speaker=true

Then, update the DaemonSet configuration:

spec:
  template:
    spec:
      nodeSelector:
        openelb-speaker: "true"
      ... ...

Using Node Affinity

First, label the target nodes:

kubectl label nodes <node-name> <label-key>=<label-value>

Example:

kubectl label nodes node1 openelb-speaker=true

Then, update the DaemonSet configuration:

spec:
  template:
    spec:
      affinity:
        nodeAffinity:
          requiredDuringSchedulingIgnoredDuringExecution:
            nodeSelectorTerms:
            - matchExpressions:
              - key: openelb-speaker
                operator: In
                values:
                - "true"
      ... ...

Using Taints and Tolerations

First, taint the target nodes:

kubectl taint nodes <node-name> key=value:NoSchedule

Example:

kubectl taint nodes node1 openelb-speaker=true:NoSchedule

Then, update the DaemonSet configuration:

spec:
  template:
    spec:
      tolerations:
      - key: "openelb-speaker"
        operator: "Equal"
        value: "true"
        effect: "NoSchedule"
      ... ...

Feedback

Was this page helpful?

Glad to hear it! Please tell us how we can improve.

Sorry to hear that. Please tell us how we can improve.