Are you running autosensing hubs or switches on your nrtwork?

Old Advance Man

As posted earlier, I still have several frustrated users that are receiving the "Communication with the host was interrupted..." error. I have several of the following in my server event log:

4/15/02 1:39:56 PM FileMaker Server Information User "Ms. Churack" no longer responding, connection closed. (17)

I suspect that they correspond with the disconnect messages the client is receiving.

I'm searching far and wide for ideas.

I think I know where you are going with that... The server is plugged in to a 10/100 switch. The clients are plugged into the whole spectrum. Most of the complaints are coming from sites with 10mb hubs... others are coming from sites with 10/100 switches.

In the case of all switches, the ports are left to autonegotiate... including the switch the server is plugged into.

As an aside, I have my Windows 2000 workstation logged into the same database... have been for going on 20 hours now and haven't been disconnected yet.

My suspicions are pointing more and more toward the client end, but there are so many different combinations of workstations/OS levels/network scenarios that it is nearly impossible to narrow this thing down.

We have a client who was running 10/100 hubs, FM Server on Mac Cubes (5), and backing up from the FM Servers to a AppleShare IP volume. Using TCP/IP mounting AppleScripts caused backups of large files to fail on a regular basis. Swithing to a mounting script using AppleShare would eliminate the problem. The problem disappeared when we changed out the hubs for 10/100 switches. Our guess is the large packet size in TCP/IP.

-bd

This is nothing even remotely as elaborate as that... We are talking about FM server sitting on a Blue G3 running a bunch of related databases (perhaps the client has 3 open at any given time). Some of the clients are connected to hubs... some are connected to switches and they are scattered in different buildings throughout the district. A heavy load for this server is 8 users at a time. The peak bandwidth on the port the server is connected to is 28kbits!

The client workstations vary from iMacs, G3 towers, to Titanium G4 Laptops.

The Open Transport error these clients are getting leads me to believe that something is timing out... are there any thresholds that can be changed on the client or server with respect to timeout?

Man, this is puzzling.

I dont suppose you have the disconnect guest if idle for x set on the server.

Just a thought. Everyone else is going down the complex route.

I have tried running with it on and off... Right now it is turned on and set to disconnect an idle user after 12 hours...