No new insight into the conditions involved In the error/crash/restart with our ETL and FCoE data protocol.
But a new path of investigation: Storport trace. Will update later if a breakthrough.
http://blogs.msdn.com/b/ntdebugging/archive/2010/04/22/etw-storport.aspx
Filter data:
15ms
F
0F 00 00 00 00 00 00 00
I have played around with this a while back, not to investigate but to see how it works. Did you test your investigation by creating a (fake) case? --@dataartisan
ReplyDeleteIn this case, the FCoE errors being generated are not on a production system, but a performance test system. We are putting FCoE through the paces, evaluating for production service. So far, we haven't been successful. Same error conditions with Emulex, Cisco, and QLogic converged network adapters and FCoE. Present the same LUNs to the same SQL Server via fibre channel and route through the HBAs... no errors. Very similar read and write disk latencies whether FC or FCoE. Until the error condition starts. :(
ReplyDelete