Managing shared disk write latency in Jira Data Center

Shared disk write latency measures the time taken to write sample data to the shared home directory in Jira Data Center clusters. This metric is crucial for understanding the performance of your shared storage system, which is a critical component in Jira Data Center architecture.

How does high CPU Load affect Jira Data Center performance?

High shared disk write latency can significantly impact your Jira Data Center instance in several ways:

1. Slower attachment and avatar uploads

2. Delayed index snapshot creation and restoration

3. Inconsistent performance across cluster nodes

4. Increased response times for operations involving shared resources

5. Potential data synchronization issues between nodes

Consistently high Shared disk write latency can degrade the overall performance and user experience across your entire Jira Data Center cluster.

What's the recommendation?

You should aim to keep the Shared disk write latencybelow 30ms for optimal performance. If you're consistently seeing latencies above 30ms, consider the following actions:

• Test disk access speed

• Check network connectivity

• Verify and optimize NFS mount configuration

• Review potential interference

• Diagnose runtime issues with Java Flight Recorder

Check network connectivity

Network issues can significantly contribute to high SharedDisk write latency. Performing basic network diagnostics can help isolate these problems.

Basic network diagnostic steps:

1. Use ping to check basic connectivity and latency

2. Employ traceroute to identify any problematic network hops

3. Utilize netstat to examine network connections and identify any issues

See our guide on Interpreting infrastructure metrics for in-product diagnostics.

If you identify network issues:

1. Work with your network team to resolve any identified problems

2. Consider optimizing network routes between Jira nodes and the shared storage

3. Evaluate the need for network hardware upgrades if consistent bottlenecks are found

Verify and optimize NFS mount configuration

Proper NFS configuration is critical for optimal SharedDisk performance in Jira Data Center.

Verifying NFS configuration:

1. Ensure you're using a supported NFS version (4.0 or 4.1)

2. Check that NFS is only used for the shared home directory, not for local home directories

Optimizing NFS mount options: Use the following recommended NFS mount options:

rw,nfsvers=4.1,lookupcache=pos,noatime,intr,rsize=32768,wsize=32768,_netdev

Refer to the Supported platforms for Jira Data Center documentation.

Additional guidance on NFS setup can be found in the Installing Jira Data Center guide.

Diagnose runtime issues with Java Flight Recorder

Diagnose runtime issues with Java Flight Recorder Java Flight Recorder (JFR) can provide valuable insights into JVM behavior that might be contributing to shared disk latency.

Using Java Flight Recorder:

1. Enable JFR on your Jira instances

2. Collect JFR recordings during periods of high SharedDisk write latency

3. Analyze the recordings to identify any JVM-related issues affecting disk performance

More on Diagnosing runtime issues using the Java Flight Recorder.

Interpreting JFR results:

• Identify any threads that are blocking I/O operations

• Check for patterns in disk access that might indicate inefficient data access or caching