Why elasticsearch makes two master nodes in cluster – Split brain problem

I have faced major issue in elasticsearch, in my cluster after some time elasticsearch automatically enables more than one node as master nodes. Due to that, it was showing two set of nodes in one cluster. It affects following …Watch movie online The Transporter Refueled (2015)

  • Cluster health becomes yellow.
  • Nodes causing CUP load issue.
  • High usage of system memory.
  • Affected performance of search and indexing.

You might face same issue and its normal, may be it is due to …

  • Problem in inter-communication between nodes.
  • Network issue in cluster.
  • Big index size and high search/index bandwidth.

Its very easy to fix this issue. Master node maintain a cluster, and requests indexing or search to data nodes and Data node stores data. When it receives a request from a client, it searches data from shards or creates an index. If we have asked a node to do both job its become difficult to manage it. So master node has to maintain cluster as well as do search and index data. It cause performance issue. Best solution is keep them separate. In your cluster you should keep one master node only and configure all nodes to look same master for cluster state. For failover you might keep one extra master node as disaster recovery.

Following are the setting to do that.

Master node

node.master: true

node.data: false

transport.tcp.compress: true

discovery.zen.minimum_master_nodes: 1

discovery.zen.ping.timeout: 15s

discovery.zen.ping.multicast.enabled: false

discovery.zen.ping.unicast.hosts: ["master node"]

Data Node

node.master: false
node.data: true

transport.tcp.compress: true

discovery.zen.minimum_master_nodes: 1

discovery.zen.ping.timeout: 15s

discovery.zen.ping.multicast.enabled: false

discovery.zen.ping.unicast.hosts: ["master node"]

Change all nodes config likewise and restart all nodes.

 

 

Categories: Elasticsearch, Website Peformance