Troubleshooting “No buffer space available” Errors

On occasion traffic on a NIC may have trouble getting out with an error similar to:

ping: sendto: No buffer space available

Or:

dpinger WANGW x.x.x.x: sendto error: 55

The most common causes of this are:

  • No route to the target network (or no default route)

  • Missing link route for a local target

  • Stale state in pf sending the connection out an invalid path (reset states)

  • Network memory buffer exhaustion - See Tuning and Troubleshooting Network Cards

  • Faulty NIC and/or driver issue

    • Sometimes resetting the NIC can bring it back again:

      # ifconfig em3 down; ifconfig em3 up
      
  • Faulty cable

  • Traffic shaping (ALTQ or Limiters) dropping the packet

  • Virtual NIC being throttled by the hypervisor or host, such as an AWS instance using more throughput than an instance size can support

    • In this case, change the throttling in the host (not guest) or upgrade to a larger instance/higher tier on a hosted platform such as AWS.

  • Virtual NIC being disconnected/disabled in certain hypervisors

  • An otherwise overloaded NIC exhausting its send/recv buffers

  • Other various switch/buffer/connectivity issues

Trying to bounce the NIC with ifconfig is the easiest thing to try first. After that, save/apply the interface settings on each interface (or at least WANs and the LAN in question). Check/(re)set the default route if it has been lost. Reset States. Replacing the cable may also help. Removing traffic shaping if it is enabled is also a good test.

Otherwise investigate the traffic on the NIC and look for other buffer-related causes. Seek help from Netgate Global Support for assistance in diagnosing the issue, or post on the forum/mailing list.