"Agent can't send heartbeats: No buffer space available" error in the VPXA log

Products

• VMware vCenter Server

Issue/Introduction

Symptoms:
You see the following error in the VPXA.log file:

Agent can't send heartbeats: No buffer space available

Cause

When the TCP/IP stack determines that a networking adapter is utilized at 80% or more, VPXA will stop sending management traffic for 70 milliseconds to help lower congestion. If a VPXA heartbeat is to be sent out during that 70 milliseconds, VPXA stops the packet from being sent and triggers the "VPXA agent can't send heartbeats: No buffer space available" error which is logs in vpxa.log on the ESXi host.

Resolution

This error is expected when networking adapter utilization exceeds 80% and can be safely ignored as long as your host is not dropping out of vCenter. By default, ESXi sends a heartbeat every 10 seconds and will not disconnect from vCenter until 120 seconds worth of heartbeats are missed. If your host is dropping out of vCenter, you can adjust any of the following parameters:

Notes:

• You can modify these parameters by running this command:
  
  esxcfg-advcfg -s <value> <Parameter>
  
  For example:
  
  esxcfg-advcfg -s 90 /Net/TcpipTxqMaxUsageThreshold
  Value of TcpipTxqMaxUsageThreshold is 90
  
  
• Alternatively, you can also use the vSphere Client or the vSphere Web Client to view and adjust these parameters. For more information, see ESXi host disconnects intermittently from vCenter Server (1005757).

To increase the amount of time vCenter will waiting before marking a host as disconnected (default 120 seconds), modify the following parameter in vCenter:

config.vpxd.heartbeat.notRespondingTimeout

To increase the network adapter utalization percentage before VPXA will cancel management traffic (default 80%), modify the following parameter on the ESXi host:

/config/Net/intOpts/TcpipTxqMaxUsageThreshold

To decrease the amount of time that VPXA will cancel management traffic after the utalization limit it hit (default 70 milliseconds), modify the following parameter on the ESXi host:

/config/Net/intOpts/TcpipTxqBackoffTimeoutMs

Alternatively, you can keep this issue from occurring by decreasing the amount of traffic being passed over the networking adapter used for management traffic.