cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2134
Views
0
Helpful
21
Replies

Cisco Solution to Microsoft Browser Election Process

amir.safayan
Level 1
Level 1

We have a significant and recurring issue in which PCs will become rogue master browsers. We don't have the human bandwidth to touch every desktop and perform the recommended registry hack to prevent PCs from iniating and winning browsing elections.

How can I prevent the propagation of these broadcast requests to other vlans????? The config of our core 6509 doen't have any ip helper address statements so I don't understand how the damn broadcasts are getting propogated beyond the vlan in which the rogue PC is on anyway????

I have used the no ip forward-protocol udp netbios-ns and no ip forward-protocol udp netbios-dgm statements in global config mode and that didn't help. As I stated earlier, I don't have a helper address configured anyway so why would I try to deny the udp port 137 and 138 broadcasts?????

21 Replies 21

moorj
Level 1
Level 1

It's been while since I've touched this, but doesn't the Master Browsers (per subnet) send directed NBT Datagrams to the Domain Master Browser as opposed to broadcast. Are you saying you have turned off ALL devices on the subnet from participating in Browser election process and broadcasts are still be recieved?

I believe your statement about directed NBT datagrams is correct. Regarding your last question, the scenario is that we have a VLAN dedicated to wireless users. If I fire up my WLAN adapter on my WIN2K Pro notebook, I become the Domain Master Browser for the entire 3000 end-user NT domain. NOT GOOD... Because the first PC to jump on the wireless VLAN doesn't see a segment master browser, it forces an election and wins! What I don't understand is why/how that process gets through the wireless VLAN and into the VLAN in which the PDC/Domain Master Browser lives. I have sniffed the traffic and I see IP directed broadcast to the wireless VLAN with UDP source/destination port 138 ,Netbios source IP (my WLAN adapter) , source name (my WLAN adapter), source port 138 and destination name "our NT Domain".

I am a Cisco engineer and I frankly don't understand how Netbios traverses across a directed broadcast packet and makes it way out of our wireless VLAN and over to another VLAN where the PDC/Domain Master Browser lives...that is what that sniff trace is saying. The source IP is my WLAN adapter and the destination IP is a directed broadcast YET the Netbios destination name is the "NT domain name". Does the router bridge the Netbios traffic by default?

raarons
Level 1
Level 1

A simpler and more effective method might be to run the registry hack as part of a network login script.

Hope this helps.

Do you have any experience with this suggestion?

konigl
Level 7
Level 7

This is a dumb question, but do you have multiple protocols loaded on each PC? I have seen environments where the desktop technicians load everything -- TCP/IP, NWLink IPX/SPX, and NetBEUI -- on each PC. Within each broadcast domain/VLAN, there will be browser elections held for each of the protocols.

Are you routing some protocols, and bridging others? This can be a problem with the non-routed protocols because the browser elections will take place across VLANs, since those protocols are bridged.

I have had two customers who were routing IP and IPX, and bridging all others. Needed to bridge for DEC LAT enterprise-wide, but also got NetBEUI bridged as a side effect. Wasn't a problem until they each went above 250+ PCs across the enterprise with NetBEUI loaded. We put a stop to the NetBEUI browser wars by filtering bridged traffic based on protocol type (passed anything DEC, dropping the rest).

We do have IPX on the PDC, but not on the wireless PCs that are forcing browser elections and winning. So on the wireless VLAN, there would only be the IP browser election and what I don't understand is how that process is making it off the wireless VLAN and over to the VLAN in which the PDC lives. We don't have the IP helper address configured on the wireless VLAN...But we DHCP is configured on our CAT 6509 to hand out IP address to the wireless clients.

We are not bridging anything as far as I know. Is anthing bridged my default on a Cat6509? How can I tell what is being bridged? We have a Cat 6509 with dual soups / MSFCs. So it is a L3 switch.

We have tried to filter the Netbios 137,138 ports and that didn't stop the wireless VLAN clients from forcing and winning the browser election. How did your filter?

matt.alvord
Level 1
Level 1

how about "no ip directed broadcasts"?

Great suggestion. I will check it out. My trace showed a directed broadcast at the IP level and UDP 138 and Netbios source name (rogue PC) and destination name (our domain name). So if we prevent the directed broadcast, that might do it.

I will let you know.

Thanks

j-metcalf
Level 1
Level 1

We are having the same exact issues on our network. We have done the following and it STILL does not fix the master browser issue.

Our vlans are configred as follows:

ip access-group 100 out

ip helper-address 172.19.1.1

ip helper-address 172.19.1.2

ip helper-address 172.19.1.3

no ip redirects

no ip directed-broadcast

And Access-Group 100 is as follows:

access-list 100 deny udp any any eq netbios-ns

access-list 100 deny udp any any eq netbios-dgm

access-list 100 permit ip any any

Any other suggestions?

Not expert on this... You can specify your Netbios node type. Does Netbios not just get encapsulated in and routed over the IP if the client has that particular 'feature' activated in their network client TCP/IP settings? That setting can be manually specified or Microsoft DHCP clients can be mandated to push this node type out by the DHCP scope setting.

why not use "no ip forwarding-protocol xxx"?

edelman
Level 1
Level 1

I know your primary goal is to find out how the broadcast-based NetBIOS packets are traversing VLANs and to stop the disfunction from a network device approach...

However, "fixing" the problem from a Windows application level is another approach. It was touched upon briefly about stopping the computer-browser service with a logon script. It sounds like you are running a mixed NT4 Server/Win2000 Workstation environment since you mentioned PDCs. I have more experience with Win2000 servers than NT4 servers, but I did work in a 2000+ PC environment where a lot of daily changes were accoplished with NT logon scripts. Whenever any domain-based PC logs into its domain, it is forced domain policy, startup + logon scripts that override any local policies. In both NT + 2000 Domain controllers, there is a shared Sysvol folder that all workstations get their policies through. You can either change policies by running scripts that change registry settings, run executables, etc or you can "cleanly" target specific domain policy settings with a simple mouse click.

In Win2000's active directory User's + Computers, you locate the top of the domain hierarchy, which can be either the forest or domain level, choose properties, and view/edit the policy associted with the hierarchy level you desire. In the policy is pretty much every desirable workstation/security/windows/etc feature you could want... specifically, there is the computer browser service. You simply check to "enable" this policy and choose what you want to do with the policy. Disable would disable the service for every single workstation that logs into the domain. Of course if you don't have DNS or WINS services running, then your PCs with disabled computer browser services will not be able to "find" any other PCs int heir Windows Network Neighborhood... You should only disable browsing if you have some other non-broadcast based service running for name resolution. Since I run Win2000 servers with DNS in my home lab with my 15 PCs, I turn of the browser service on all domain controllers and workstations... and rely soly on unicast-based DNS for all name resolution functions.

If you have a Windows "specialist" that works with you (you being the network engineer/net device specialist), the two of you should be able to get the script/domain policy working with minimal time + trouble. To clarify how the scripting/policy propagation works, just checkout Microsoft's Technet web site. You should be able to test this change in a small test lab with a single PDC, a few VLANs, and a few workstations...

Then of course there is the issue with the Catalysts and why they are propagating those pesky NetBIOS packets...

Hey, I got a sugestion - lets look at past posts for these same answers. This has been like the 6th thread on this same topic in two or three weeks. Here is a link to the post with the answer...

-Bo

http://forums.cisco.com/eforum/servlet/NetProf?page=netprof&CommCmd=MB%3Fcmd%3Ddisplay_location%26location%3D.ee91c21

Hey, I got a sugestion - lets look at past posts for these same answers. This has been like the 6th thread on this same topic in two or three weeks. Here is a link to the post with the answer...

-Bo

http://forums.cisco.com/eforum/servlet/NetProf?page=netprof&CommCmd=MB%3Fcmd%3Ddisplay_location%26location%3D.ee91c21

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: