My users get disconnected

Colubra · Post by **Colubra** » 2003-05-09 02:24

I am running a small private hub with nmdc and stimbot 1.1, each time the hub receives a validate nick request from a user that is not registered with the hub my users get disconnected for 2-4 minutes.

NMDC hub runs on a Win2k sp3/AMD T-bird 800/128 MB PC133/WD 4 GB HDD/ADSL router.

All needed tcp/udp ports are opened in router and I believe windows tcp/ip settings are ok. There is no firewall currently installed. I have been able to clear up a lot of this by changing ports but it still gets a little buggy.

Any and all help in resolving or understanding this problem will be accepted and appreciated.

Thank you,
Colubra

ButterflySoul · Post by **ButterflySoul** » 2003-05-09 05:04

That's a really weird one. I don't recall anyone who had that problem yet, so I can't magically pull out a solution for you based on what worked the last time; but here are the steps to narrow down the problem a bit.

1) Make sure you have the NeoModus client installed on the computer running the hub (it's easy to forget, especially after a re-install of windows for example). If it's missing, your hub will really do crazy stuff =)

2) Try it out without without stimbot or any scripts for just a few hours or a day. See if it still does it. That should isolate the problem so you can know for sure if it's comming from the hub software or from stimbot.

Just out of curiosity, do you have other internet related applications running on your computer aside from the hub, your client, and your DNS service ? If changing ports helped the problem a bit, there's a fair chance that both the hub and stimbot work fine, and that your problem comes from a third-party proggie messing things up.

Colubra · Post by **Colubra** » 2003-05-09 11:21

Thank you ButterflySoul.

Yes, I have ran the hub minus all scripts and had the same results.
NMDC client is installed but not configured.

Win2k pro is at fresh default install with ony NMDC hub & client, AdAware 6.0, and RegCleaner installed. AdAware is updated and has removed Alexa.

In windows local area connection properties only Internet Protocol (TCP/IP) is enabled, nothing else. In tcp/ip settings radio button for use the following ip address is checked with internal ip address set to the first available addy after gateway (I failed to mention Lynksys 5 port workgroup switch in earlier post).
Use the following DNS server is checked and settings reverified with ISP. ISP also claims not to monitor, look for, or block any P2P. All advanced tcp/ip settings are at default. File and Print sharing have been removed.

I do use the dyndns.org naming service, but users have same problem using unmasked address. I connect to the hub using the second pc on a 3 machine network.

This is the barebones machine running the hub software,
http://www.bigedtech.com/articles/147.php

I am at a complete loss without a clue, and not to mention way-in-over my wannabe technical head. I have spent hours in this forum reading posts looking for that magic potion, but to no avail. =(

Thank you again for the recommendations ButterflySoul.

Cheers! =)

Colubra · Post by **Colubra** » 2003-05-09 12:52

Today another user and I were able to duplicate what appears to be a $ValidateNick problem. My user changed nick to one not registered with the hub and attempted to connect. This forced the remaining users to somehow freeze as I could not get their files (connection timed out) and they eventually disconnect, for up to 6 minutes.

Another sympton appears like a ripple in the network, when one user logs off the hub the remaining users experience disconnects and get connection reset by server messages.

I believe the NMDC hub is running as designed and problem is somewhere in the switch/router. But, beyond simple port forwarding in the router's NAT settings, I am in uncharted territory.

Thanks again to all who take the time to read this thread and offer help.

Colubra

ButterflySoul · Post by **ButterflySoul** » 2003-05-10 23:02

If I understood it properly, you have a switch handling your LAN traffic, and the switch's uplink goes into your router, which in turn goes into your (cable/dsl/whichever connection you use).

If your router has several plugs and you can add extra connections on it, the next thing I'd try would be to plug the computer hosting the hub (which appears to be a dedicated server) directly into the router, instead of plugin it into the switch -while keeping all your other computers on your LAN pluged into the switch, and the switch pluged into the router as well-. At least for a bit, just to see if the problem still happens.

Your setup is about as good as it gets. The only thing that I could think of is if your switch suffers from some subtile confusion somewhere in its own tables when the same computer shows up a second time, or some other hard to figure out strange rare issue =)

Colubra · Post by **Colubra** » 2003-05-12 03:05

Thanks again ButterflySoul, Unfortunately I don't have but one connection off the router. Your recommendations have certainly given me new ideas for some new hardware.

Cheers!

Colubra · Post by **Colubra** » 2003-06-08 09:53

Just thinking...what if I put two NICs in the server box and ran it right after the dsl router?

ButterflySoul · Post by **ButterflySoul** » 2003-06-08 11:31

Just thinking...what if I put two NICs in the server box and ran it right after the dsl router?

That will work fine, as long as you have a proxy software on the server box (or if you don't want the other computers to have access to the internet as well).