This is a three site network which was previously *very* stable (but s.l.o.w.) using vpn over cablemodem. We recently changed to an embarq "fast ether" connection, it is 50mb sync. fiber. Nothing else was changed (beyond routing & firewall rules). Two remote sites connect to the central site which hosts storage and the database server.
I'm not certain any of this is relevent, but here it is anyway:
All three sites are using soekris net4801 running: m0n0wall 1.22 (central), pfSense 1.2 (remotes)
old_remote_site has all stations as part of an independant domain (old_domain_2)
new_remote_site has all stations as part of the central_site domain (central_domain)
all sites are in their own /24 address spaces.
all three (fiber) gateways are in their own /24 address space (192.168.88.0).
I can ping any address at old_remote_site from database_server (at central_site).
I can ping by name database_server from any station at old_remote_site.
tracert from either end shows the expected output.
everything I've looked at says MTU 1500. Embarq says I can use jumbo packets if I want.
The issue is this: when users at old_remote_site scan documents or take xrays (dental office), they get message to the effect that the file cannot be created, usually accompanied by the "delayed write failure" message in the tray. If I attempt a file copy, I get at a random point in the copy "network name no longer available". "net use" at the command prompt tells me the drive is disconnected. but it will reconnect if I change to that drive. If I then change back to drive c:, the status *immediately* changes to "disconnected".
It may be relevent to note that the database client program (which uses sql over tcp/ip) is completely unaffected by all this...only my mapped drives/file shares die.
What I really don't understand is that this all worked perfectly over the VPNs, and I didn't change any domains, users, login scripts, dns servers, client gateways, or anything else except for the routing on the m0n0wall/pfSense boxes.
Start Free Trial
