Jira Data Center Clustering: Architecture, Benefits, and a Real-World POC Implementation

[Pandya Jay]

Introduction

Jira Data Center is designed for organizations that need high availability, fault tolerance, scalability, and enterprise-grade performance. Unlike single-node Jira Server deployments, Jira Data Center operates in an active-active clustered architecture, where multiple Jira nodes run simultaneously behind a load balancer.

Recently, I performed a Proof of Concept (POC) for Jira Data Center clustering by deploying two Jira application nodes and successfully configuring them into a working cluster. This article explains:

• Jira Data Center clustering fundamentals
• Core architecture and components
• Benefits of clustering
• Step-by-step POC approach based on my implementation
• Key learnings and best practices

---

Key Components of a Jira Data Center Cluster

A standard Jira Data Center clustered setup consists of the following components:

Multiple Application Nodes

• Identical Jira installations running on separate servers
• Handle user requests and application logic
• In my POC, I configured two Jira nodes with identical versions and configurations

Shared Database

• A single external database used by all nodes
• Ensures consistent issue, project, and configuration data
• Supported databases include PostgreSQL, MySQL, and Oracle

Shared Home Directory

• Network File System (NFS) mounted on all nodes.
• Attachments
• Avatars
• Index snapshots
• Cluster-wide configuration files

Load Balancer

• Sits in front of Jira nodes and distributes traffic
• Performs health checks
• Must support sticky sessions (session affinity)
• Examples: HAProxy, AWS ELB/ALB, F5, NGINX (software LB)

---

Benefits of Jira Data Center Clustering

High Availability : If one node fails, traffic is automatically redirected to the remaining nodes

Performance at Scale : User load is distributed across nodes, improving response time

Disaster Recovery : Enables multi-data-center or geographically distributed deployments

Zero-Downtime Upgrades : Nodes can be upgraded one by one without impacting users

---

Jira Data Center Clustering: My POC Implementation (2-Node Setup)

This section highlights the actual steps I followed during my POC.

1. Preparation and Prerequisites

Before starting the setup, I ensured the following:

• Active Jira Data Center license
• Two Linux servers for Jira application nodes
• One shared database server
• One shared NFS location accessible by both nodes
• Proper network connectivity between all components

2. Jira Installation on Both Nodes

• Installed the same Jira Data Center version on both nodes
• Verified Java version compatibility
• Confirmed both nodes could independently start Jira

Each node had:

• Its own local home directory
• Access to the shared home directory

3. Cluster Configuration (cluster.properties)

For clustering, I configured the cluster.properties file on each node.

Key parameters included:

• jira.node.id (unique per node)
• jira.shared.home (same NFS path for both nodes)
• jira.cluster.name

This allowed Jira to recognize both nodes as part of the same cluster.

4. Shared Home Directory Setup

• Mounted the NFS location on both nodes
• Copied required files from the first node into the shared home
• Verified file permissions and ownership

Once configured, Jira automatically synchronized shared data across nodes.

5. Database Configuration

• Configured both nodes to point to the same external database
• Verified database connectivity
• Ensured database connection pooling was properly configured

6. Load Balancer Integration

For the final step of the POC:

• Configured the load balancer in front of both Jira nodes
• Enabled sticky sessions, which is mandatory for Jira Data Center
• Set up health check URLs to monitor node availability

After this step, users could access Jira via a single load balancer URL, while traffic was distributed across both nodes.

---

POC Validation and Testing

To validate the success of the clustering POC, I performed:

• Node shutdown testing to verify failover
• Session persistence checks
• Simultaneous user access from multiple browsers
• Log verification to confirm cluster heartbeat communication

All tests confirmed that:

• Nodes were active simultaneously
• Failover worked as expected
• No data inconsistency was observed

---

Key Learnings from the POC

• Sticky sessions are critical — without them, users face login and session issues
• Proper NFS performance and permissions are essential
• All nodes must be version-aligned and identically configured
• Monitoring cluster health is crucial in production environments

---

Conclusion

Jira Data Center clustering is a robust solution for enterprises that require continuous availability, scalability, and resilience. Through this 2-node POC implementation, I was able to validate the core clustering concepts and gain hands-on experience with real-world challenges such as shared storage, load balancing, and failover testing.