Posted on 2011-03-17
At work I look after 4 data centres, with a router utilizing 4/4 IPMAN cross connects. Each of these routers has atleast one link to this IPMAN, over which we run EIGRP as the IGP.
Now everything had been working fine till a few months ago, when one of the links, connecting Colo B ---- Colo C went down. I noticed the EIGRP neighbor relationship was no longer forming on Colo C side (nothing in the debugs) and on Colo B side the neighbor as Colo C would show up, RTO would expire, and it would go down/up again. We tried everything (link speeds, duplexes, EIGRP statements, no static routes) - but resigned it as a hardware/link fault which we would need to chase up with the IPMAN provider.
A few months down the line, everything was working fine but yesterday the neighbor relationship between Colo B ---- Colo D went down. The issue was exactly the same, this time no debugs or neighbors (with Colo D) showing up on Colo B , and Colo D keeps trying to form a relationship with Colo B, times out, and starts over again.
This is pretty strange and since Colo B --- Colo D link is critical for business operation, I am keen to figure out the cause of this. Currently i'm routing the traffic through another router with static routes and all is well, but would love some feedback to see what your opinions are. Is it a link fault? Is there anything I could do to confirm this?
There are no CRC/errors on the actual interfaces, the routers are all fine with rest of the links working well. No duplex/speed issues as this was working well just yesterday.
Appreciate all the comments guys.