CRC erros on Ubuntu router

Hi, we recently upgraded to a new metrofiber connection that connects 5 buildings together. Each of the buildings has a tellabs 8813-310 access node provided by our ISP. We use port 3 at all the remote stores and they each connect to a different port on the tellabs at the main store. We made an Ubuntu router with a total of 5 NIC's in it, 4 for the remote stores and 1 for our main location. 4 of the NIC's on the Ubuntu router plug directly into the corresponding port on the Tellabs node provided by our ISP. The other NIC plugs directly into a switch at our main store on our main subnet. The other 4 remote stores are all on different subnets.  We set up the routing table and we have connectivity between stores, but we're running into CRC errors being generated by the router. I worked with our ISP for several days on this and they said their Tellabs access node was receiving CRC erros but not transmitting them. This is causing problems with sharing files between computers on different subnets, for example, if I try and copy a file that's much more than a couple Mb's , it'll start to copy, then stop with the windows error message "the specified network name is no longer avaiable". These erros are very random and I can find no consistant pattern in them.  This is causing problems copying files between subnets and running programs hosted on our main server. We have full internet and network connectivity at each of the remote locations, though they are not running as fast as they should be. I ran ifconfig's on each of the NIC's on the router and saw lots of collisions for the NIC's that go to our remote stores through the tellabs access node, but no collisions on the NIC that plugs directly into our switch at the main store. I configured all the NICs to our ISP's requirements (100 mbps, full duplex, no auto negotation), tried swapping out CAT5 cable, even tried putting a small 5 port switch between our router and the tellabs to see if that might help with the CRC erros and nothing worked.  When I run a continuous ping from a remote store to it's corresponding gateway (NIC on the the Ubuntu router), I get random "request timed out" messages, but like I said before, have found no consistancey on it. From what we determined with our ISP, it sounds like it's an issue with the Ubuntu router and I was wondering if anyone has ran into similar problems running a home made linux router or would be able to point me in the right direct as to resolving this issue.  I know this is a lot of info and I hope I didn't leave anythign out, if I did please ask. Thanks!