NSX Installation on an ESXi Host stalls at “6% Uploading >> Uploading NSX" and fails with Error: “Failed to install software on host. Failed to download NSX components on host. Check the host connectivity and if /tmp folder has enough space.”

Products

• VMware NSX

Issue/Introduction

When attempting to install NSX on an ESXi host, the installation may stall at “6% Uploading >> Uploading NSX” and eventually fail with the following error:

"Failed to install software on host. Failed to download NSX components on host. Check the host connectivity and if "/tmp" folder has enough space" 

This error generally points to one of the following:

• Insufficient free space in the /tmp directory of the ESXi host.

• Network connectivity issues between the ESXi host and the NSX Manager.

Verification Steps

1. Verify /tmp Folder Space on the Host

Run the following command on the affected ESXi host: "vdf -h"

The nsx-lcp-bundle for an ESXi host can consume up to 175 MB of space. Ensure there is sufficient free space in the /tmp directory.

 

2. Verify Connectivity Between Host and NSX Manager

1. Check logs on NSX Manager
   
   Review /var/log/proton/nsxapi.log for errors. Example:
   
   

   ERROR ActivityWorkerPool-1-11 #####Hostname##### 266533 FABRIC [nsx@6876 comp="nsx-manager" errorCode="MP26168"
 level="ERROR" subcomp="manager"] Install/Upgrade on ESX - Uploading LCP bundle to host ########-####-4462-####-############ failed
com.vmware.nsx.management.ovfops.exception.CommandExecutionFailed:    % Received % Xferd  Average Speed   Time    Time     Time  Current

   105M      0   359k  0:07:31  0:04:59  0:02:32  327k
curl: (28) Operation timed out after 300000 milliseconds with 0 bytes received

2. Test port connectivity from the ESXi host to each NSX Manager node using nc (netcat):
   
   nc -zv <Manager-IP> 443
   nc -zv <Manager-IP> 1234
   nc -zv <Manager-IP> 1235
   
   

3. Capture traffic to confirm connectivity:

   • On ESXi host: capture traffic from the management VMkernel interface.

   • On NSX Manager: run from CLI (admin user): "start capture interface eth0 expression host <problematic-hostname>"
     
     Reference: Packet Captures in ESXi using tcpdump-uw

Environment

VMware NSX 

Cause

This issue can occur due to:

• Network connectivity issues causing the nsx-lcp-bundle transfer to exceed 5 minutes.

• Insufficient free space in the ESXi host /tmp filesystem to store the nsx-lcp-bundle file.

Resolution

 

1. If /tmp Space Is Insufficient

    Free up space by deleting unnecessary files under /tmp on the ESXi host. Ensure at least 175 MB is available.

2. If Connectivity Is Successful (Ports 1234 & 1235 Reachable)

Follow the resolution steps in the KB:
NSX Installation Fails While Uploading LCP Bundle to ESXi Host (KB 379291)

3. If Connectivity Fails (Ports Blocked / Timeout)

• This indicates an underlying network issue.

• When attempting manual upload of the LCP bundle from NSX Manager (/repository/) to the ESXi host /tmp/, the process may fail and request a host reboot, discarding the incomplete update.
  
  

Action:

• Ensure the required NSX communication ports are open: NSX Ports Reference

• Allow NSX Manager traffic (both individual NSX Manager node IPs and VIP) through any firewall in the dataplane path between the host and NSX.

Additional Information

In some cases, the nsx-lcp-bundle may be partially copied to the host /tmp directory. Verify bundle size on both NSX and host sides:

• On NSX Manager:

   

  ls -lrth /repository/<NSX-version>/HostComponents/<esx-version>/nsx-lcp-<nsx-version>-<esx-version>.zip

• On ESXi Host:

   

  ls -lrth /tmp/nsx-lcp-<nsx-version>-<esx-version>.zip