[RESOLVED] Network Connection Problems...Intermittent!!!

I hope somebody can give me a few pointers. We have a network of 2 Nt 4 Servers - One Running as a RAID File server and the other a MDaemon Mail server. There is a third Server - Unix OS running WDW kerridge Network application.

There is an ISDN Router(3-Com) - using a 56K channel for ISP connectivity.
Another SOHO router for dial in access for maintenance of Unix server.

The offices are devided in two - the older cabling in the network connected to a MRI Hub - 10/100 Ethernet, downstairs in one office. Connected to this are approximately 6 workstations, the two ISDN routers and one IO lan Hub for 4 serial printers.

This hub is connected to another hub - same type upstairs, with one uplink running between the two.

Connected to the hub upstairs are the 3 servers and around 14 workstations. All workstations and Servers are running at 100mbs.

**The problem is that at several points in the day, we are either disconnected from the network or if not from the network, the unix application server - affecing productivity. At one stage, I put a ghost image onto a server and while downloading onto a system, the whole network stalled.**

At first, I replaced the two hubs, and the problems still existed. I then got up above the ceiling and was shocked to see the electricians job of cabling - UTP running along with electric and phone cable - twisted and cable tied very ticktly. We even had one electrical cable looped around the network cable. This has kinda been rectified, but the cable still runs along the same conduit as the electric cables. I used a DSP 4000 Fluke tester and found a few problems with cross over. This was resolved by replacing the cat 5 glasses (plugs). Anyway, essentially, all tests passed. The only thing I can do now is to replace all NICs independently. Thats gonna take a while. We have 3com 3c905cs and bs and a few realtex nics too. We are running TCP IP protocol with static IPs, no DHCP server. All pcs access the ISDN router directly for Internet access.

We had a cable specialist who says it's over his head, and we've had a quote from one bloke stating that we need to replace all the cabling with STP and has given us one hell of an expensive quote.

[email protected]

WOW!

Can you pull the event logs at the trouble times and see if the system is getting overloaded by a routine - backup, replication, dial in - out? I assume you have some servers as BDC's - got all the replications going to the right places so the workload (security etc.) is shared?

All I can think of, let us know how you get on!

Good luck!

Were you using the multicast option in Ghost to load the machine? I only ask because our network came crashing down rather quickly when we tried it... I was quite surprised as we use Cisco Catalyst 1600 Switches and I didn't think Broadcast traffic was supposed to travel over switches. The rapidly flashing green lights on them DID look cool though :D

It sounds like a bandwidth problem. You might want to try replacing the hubs with switches, especially the one that the servers connect to. You may also want to check the nic in the pdc.