VM Network Link Down on NSX Segment after /tmp Directory Reaches 100% Capacity

Products

• VMware NSX

Issue/Introduction

A Virtual Machine (VM) connected to an NSX Segment (Overlay or VLAN) shows a "Link Down" status within the Guest Operating System (GOS) and no network connectivity. Symptoms include:

• VM reporting LINK STATE DOWN on the network interface.

• Issue persists after vMotion to different ESXi Transport Nodes.

• Issue persists after removing and re-adding the Virtual Network Adapter (vNIC).

• Connectivity does not return even when moved to a standard vSphere Distributed Switch (VDS) portgroup.

Environment

• VMware NSX
• Guest OS: Linux (RHEL, Ubuntu, Photon OS, etc.)

Cause

The Guest OS filesystem (specifically the /tmp directory) is at 100% utilization. Modern Linux networking stacks and daemons (such as NetworkManager, systemd-networkd, or dhclient) require available space in /tmp to create lock files, temporary sockets, or lease information. When space is exhausted, the interface fails to initialize or transition to an "UP" state.

Resolution

1. Open the Web Console or Remote Console for the affected VM via vCenter.

2. Log in as a user with sudo or root privileges.

3. Verify disk space usage using the following command:

   df -h
   

4. If /tmp or the root partition (/) is at 100%, identify and clear unnecessary temporary files:

   ls -lS /tmp | head -n 10
   rm -rf /tmp/<identified_stale_files>
   

5. Manually bring the network interface up:

   # Identify interface name
   ip addr
   # Bring interface up
   sudo ip link set <interface_name> up
   

6. (Optional) Restart networking services to ensure all transient files are recreated correctly:

   sudo systemctl restart NetworkManager  # For RHEL/CentOS/Ubuntu
   # OR
   sudo systemctl restart systemd-networkd # For Photon OS/Debian

Additional Information

To prevent recurrence, it is recommended to:

• Configure systemd-tmpfiles-clean.timer to automate /tmp cleanup.