Packet discards at VMXNET3 adapter on ESXi host

netstat -e

Strange Packet discards

In the last time I encountered to a strange problem.
The following components have been involved:

Win2008 R2 servers with VMXNET3 Adapters.

Used VMware configuration

  • vCenter Server 5.1.0a
  • ESXi 5.1 Patch 1

Looking at “netstat -e” shows the following strange output.
The discard counters for for send and receive side were growing and had also the same amount.
The output got my special attention, because the RX and TX counters shows always the same value.
Example “netsat -e” output collected with the tool CRWinNetDiag:

======================= netstat -e ===========================
Interface Statistics
                            received            send
Bytes                     1445355112      3181346202
Unicastpakets                1704730         2459144
Non-Unicastpakets             260454            1471
Discarded                        657             657
Error                              0               0
Unknown Protocol                   0

So the follwing investigation at the ESXi host shows a misconfigured Parameter.
ethernet0.emuRxMode = “0

The parameter should be enabled in environments with a high multicast rate.

After the parameter was set to:
ethernet0.emuRxMode = “1
The “netstat -e” output looks like:

======================= netstat -e ===========================
Interface Statistics
                             received            send
Bytes                      1345356213      3191356110
Unicastpakets                 1804730         2559144
Non-Unicastpakets              250453            1672
Discarded                           0               0
Error                               0               0
Unknown Protocol                    0

So this problem that in burst situations packets has been discarded could be fixed.

Have you seen this behaviour, too? Please tell me!

One comment

  • I saw this at a Win 7 virtual machine on VMvisor 5.5.0 update 1, too.

    The parameter ethernetx.emuRxMode=1 enables the SplitRx mode, which allows a host to use multiple physical CPUs to process network packets received in a single network queue

    But in my environment there is no way to fix this issue, because my machine has only 1 physical processor.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s