Master Not Discovered
Severity:
Critical
Elasticsearch Version:
8.5.0
Problem
Cluster fails to elect a master node, leading to a split-brain scenario or no cluster coordination.
Root Cause
Insufficient eligible master-eligible nodes, network partition, or misconfiguration preventing election.
How to Detect
Symptoms
- Cluster health shows 'red' with no master node assigned
- Nodes are in 'disconnected' or 'initializing' state
- Cluster state API indicates 'no master node'
- Logs show repeated master election failures
Commands
curl -X GET 'localhost:9200/_cluster/health?pretty'
curl -X GET 'localhost:9200/_cat/nodes?v&h=ip,master,name'
curl -X GET 'localhost:9200/_cluster/state?pretty'Remediation Steps
- Verify network connectivity between nodes
- Check 'discovery.zen.minimum_master_nodes' setting matches the quorum size
- Ensure all master-eligible nodes are running and properly configured
- Restart nodes one at a time to allow re-election
- Review logs for errors related to master election
- Adjust 'discovery.zen.ping.timeout' and 'discovery.zen.fd.ping_interval' if network issues are suspected
Prevention
- Configure an odd number of master-eligible nodes to avoid split-brain
- Set 'discovery.zen.minimum_master_nodes' to (N/2)+1
- Implement proper network segmentation and monitoring
- Regularly update Elasticsearch to latest stable version
- Use dedicated master nodes with sufficient resources
Production Example
curl -X GET 'localhost:9200/_cluster/health?pretty'