The first one was one end of a point to point link with two RB435G boards and 2 Atheros 11N cards. This links two OSPF routers together on an OSPF ring. Multicast Helper is enabled. (The OSPF ring routers are all Routeboard1100AH+ units)
The link would show down because no IP traffic was passing through. Reboots would restore service but would shortly go back down.
Then one time I telnet from the other side to the device that was down via the wireless connection and discovered that the interface said the port was up OK and auto neg at the appropriate rate.
Replaced the board with a new one and it stayed up. RMA the board and it was sent back no fault found. (However there is no load on that link at this time since the primary P2P Dragonwave link is operational)

Then there are two PC type systems running Mikrotik software that exhibit the same behavior except this is the router that on one side connects to our OSPF ring via static route and the other side services our internet customers. The link on the ring side would show down in The Dude but The Dude would still show the CPU as being up. Reboot of these two routers will last about week or two before they went back to the down state but link still ok. One of them I now have set to reboot itself once a week in the wee hours of the morning to keep it going. The other of the two had been running for awhile but just went into that state night before last.

Then there are the two other RB435G boards and 2 Atheros 11N cards on another section of the network (This links two OSPF routers together on an OSPF ring. Multicast Helper is enabled) that do the same thing but only when they have a bandwidth load on them. When sitting idle (traffic handled by the other P2P link) they never drop.
Another funny thing about this is I have two cat6 cables connected to one end. One goes to the OSPF router and the other plugs in to our office switch so I can access the routerboard directly from my office PC. When it goes down, both ethernet links show up and OK but neither one passes IP traffic.

I have had to set netwatch on each end of the P2P to ping through its ethernet connection and reboot when it goes down in order to keep them and our customers working. (Does not help that the Dragonwave point to point on the other side has failed also)

I am turning to the forum in hopes some one else has seen this behavior and knows how to fix it.

Thanks for any help!

Day two... What a fun time we had last night. The P2P went down last night at 5pm. Since I set the netwatch too tight it caused the routerboard to reboot over and over. So up to the roof and reset the board and reprogram from scratch. But we could not get the other end to connect. Turns out the other board on the mountain was doing the same thing. So up the mountain - reset the board up there and right in the middle of programming the radio interface I get router disconnected. What?!? Now the indicator light on the power supply is flashing on and off. The tech on the tower said the boards power light did the same thing up there. Tried three other power supplies and still the blinking power light. By this time it is too dark to work on towers.

This morning we are replacing both of the RB435G boards with RB433AH which we have historically had good luck with.

Next... Well we replaced the routerboard up on the mountain this morning and ran into the same problem - routerboard shutting down - then either would not power up or would just blink the power LED. Now that we were up there and could get access to the equipment room it was discovered that the Webpower Controller was complaining about high amp usage on the bank it and the Dragonwave was plugged into. So we moved the Dragonwave off to a another Webpower Controller and then the routerboard booted up and stayed up.
A reboot of the Dragonwave brought that primary connection back on line also. We concluded that there was too much draw on the Webpower Controller to properly power all the devices plugged into it. This may be the issue at the other locations where Mikrotik equipment is going into the funny link but no IP traffic state. Investigating further.

Day three... Woke this morning and the routerboard we replaced up on the mountain was down again. Thankfully the Dragon wave link is still working so I was able to power cycle the routerboard remotely. It is back on line. Still do not know why it went down. We even replaced the power supply for this device. Will be going up to the other mountain on Sat where we have the PC based mikrotiks and re-arrange the power up there also.