EVA Data replication

[Richard Jackson_6]

I have three pairs of EVA5000s, replicating over two different telco networks via DWDM.
Fabric1 is logging out of frame errors at either end of the ISL.

Just so you get the picture
EVA1 Replicates to EVA1DR
EVA2 Replicates to EVA2DR
EVA3 Replicates to EVA3DR

EVA1 and EVA3 pairs have stopped replicating over Fabric1, EVA2 pair are unaffected, and is replicating over both fabrics.

EVA1&3 are running VCS3025
EVA2 is running VCS3020

This problem arose three weeks ago, the solution (after replacing all ISLs, DWDM cards, SFPs, etc) was to move the ISL to a different swictch blade in the Brocade directors, this 'kicked' replication back into action.

Please note that all host ports of the EVA are ok, logged in and being used, the error is seen in the data replication folder, each DR Group that is using the bad ISL is logging, with the connection failed.

Questions:
1. From an EVA point of view, if the replication is going down the good ISL, I have no indication that there is a problem, it is only when a LUN is moved via securepath to the other controller that the 'dodgy' ISL is used, and the EVA flags the DR connection as bad in the DR folder. Surely in the depths of the EVA, there must be of seeing that this DR path is bad, there must also be a path reset command somewhere to 'kick the DR path back into action????

2. I am assuming that VCS3025 has tighter error thresholds than VCS3020, is this resolved with VCS3028?

3. if servers are presenting LUNs (using securepath) to controller A, does replication go from port 1 of controller A?

4. I am assuming that this problem is due to errors over the ISL, I also assume that the EVA is keeping track of these errors and counting them up, can I get to see the counters, see the threshold where the EVA says 'enough is enough', and can I reset these counters?

Cheers all,

Rich

[Richard Jackson_6]

I know this is probably more aimed at the internal workings of the EVA, so would anyone from HP like to comment?