Troubleshooting Checklist for Nutanix Cloud Platform with Everpure FlashArray

Nutanix
Audience
Public
Product
FlashArray
FlashArray > Purity//FA
Technology Integrations
Source Type
Documentation

To validate NVMe/TCP connectivity and ensure stable, high-performance operation, we recommend that you follow the verification steps in this section.

These checks help confirm correct session establishment, path health, MTU consistency, and end-to-end network reliability. Where indicated, customers should work with Nutanix Support to perform or validate these actions.

  • Investigate operational issues using the Nutanix Prism: Nutanix Prism provides dashboards for cluster health, performance trends, alerts, and VM/storage/network metrics, which can quickly highlight anomalies and bottlenecks.

    • Be sure to leverage Prism Element for local cluster status and Prism Central for aggregated, multi-cluster insights and analytics.

    • Use Prism's built-in health checks (such as NCC health checks) to surface underlying system issues, review alerts and event logs for error patterns, and correlate performance graphs across compute, storage, and network layers to isolate root causes.

    • When Prism Central is used, validate that it is properly registered to the managed clusters and check PC VM resource utilization and health as part of your troubleshooting process.

  • Gather logs and performance data: These tasks can provide insight into deeper workflows and unresolved conditions.

    • Gather relevant logs and performance data from Prism and engage Nutanix Support with comprehensive context and findings for accelerated resolution.

    • Verify active NVMe sessions. From the initiator side, confirm that all expected NVMe subsystems are connected by running nvme list-subsys. Engage Nutanix Support to execute and interpret these commands as needed.

  • Check NVMe multipath health: Validate that all paths are active and operating as expected using Prism or by inspecting /sys/class/nvme-fabrics/.

  • Inspect /var/log/nvme.log on the hosts: Review alerts and events in the Everpure FlashArray GUI to identify errors or anomalies.

  • Validate end-to-end MTU consistency: Ensure that MTU settings are consistent across the entire path—typically 9216 bytes within the network fabric and 9000 bytes on the nodes and storage arrays.

  • Test end-to-end connectivity: Verify jumbo-frame support by running ping -D -s 8972 <target-IP> from the Nutanix nodes. The -D option sets the Don't Fragment bit, and an ICMP payload size of 8972 bytes validates the MTU of 9000 across the entire path. For 1500 bytes, ping with a payload size of 1472 bytes.