Super User is a question and answer site for computer enthusiasts and power users. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Some time ago nearly all workstations in our team (Windows XP SP2) exhibited intermittend but frequent delays when accessing shares on the network. Typically the first access to a share which hadn't been accessed for some time resulted in a nearly frozen workstation for up to 30 seconds. Then everything started working fine again.

Using TCPView from Sysinternals I saw that during this delays there was a connection to the netbios-ssn port on the file server which was in state SYN_SENT.

First try:

Disable NetBIOS over TCP/IP for the intranet network adapter.

Problem solved, but I didn't like to manipulate our centrally managed network configuration for the intranet.

Second try:

Disable NetBIOS over TCP/IP only for the VMWare network adapter (VMNet1 used for host only communications).

Problem solved again!

My questions:

  • Why does NetBIOS over TCP/IP on one network adapter disturb NetBIOS over TCP/IP on another network adapter?
  • Is this problem specific to VMWare network adapters?
  • Has anybody else seen this phenomen?

Additional information:

  • VMWare Workstation version 6.0.3
  • At the time I started seriously analysing the problem it was no more possible to find out what had been changed to our systems at the time the problems started.
share|improve this question

migrated from Jul 21 '09 at 20:53

This question came from our site for professional and enthusiast programmers.

From Microsoft's NetBIOS Over TCP/IP:

NetBIOS Name Registration and Resolution for Multihomed Computers

NetBT binds to only one IP address per physical network interface. From the NetBT viewpoint, a computer is multihomed only if it has more than one network adapter installed. When a name registration packet is sent from a multihomed computer, it is flagged as a multihomed name registration so that it does not conflict with the same name being registered by another interface in the same computer.

If a multihomed computer receives a broadcast NetBIOS Name Query, all NetBT and interface bindings that receive the query respond with their addresses, and by default the client chooses the first response and connects to the address supplied by the responder.

It looks as if the problem was caused by the fact that all interfaces were being queried. By reducing the number of eligible interfaces, you apparently escaped being considered as a multihomed computer, therefore speeding up the process.

share|improve this answer

You must log in to answer this question.