Web Services Unresponsive (RHEL) on NSX Overlay network

Products

• VMware NSX

Issue/Introduction

• HTTP web services on Red Hat Enterprise Linux (RHEL) virtual machines were unresponsive to traffic originating from different network segments.
• This issue was specific to RHEL VMs, Windows VMs on the same network segment continued to function correctly.
• Able to ping RHEL VM, and no NSX firewall policies were blocking the traffic.
• RHEL web service were reachable within its own network segment but consistently failing to access it from any external segment.

Environment

VMware NSX

Cause

The root cause was identified as an MTU (Maximum Transmission Unit) mismatch between the NSX overlay network and the underlying physical network fabric, specifically on the physical NICs of the NSX Edge cluster hosts.

Resolution

To resolve this issue, set MTU on the physical NICs of the NSX Edge cluster hosts to 9000 or higher. This enables support for jumbo frames on the physical network.

Note:- Setting the physical network's MTU to 9000 ensures that the underlay network is aligned with the requirements of the NSX overlay. This configuration allows the larger, encapsulated packets to pass through the physical infrastructure without being fragmented or dropped, thereby re-establishing reliable end-to-end communication for all traffic types. Following this change.