I have a nuisance 10054 error occurring between two servers...
basics:
4 (v5.6) servers, each replicating to the other three for the last 5 years.
Suddenly since xmas, one server can't replicate to one of it's destinations.
The other two CAN replicate to the destination in question.

At first I thought it was the destination server at fault.

CommDiag tests show no problem with attach and detach. SQLExplorer can show
the problem though as the connection successfully made (logon) but as soon
as you try to browse a table or query a table, the socket crashes. But the
destination server can happily connect to the source server and replicate
without error. (i.e. the reverse direction is fine).

I have done traceroutes from both directions and found that destination to
source (the one which works fine) is OK, but the source to destination shows
a timeout on one device consistently.

Does anyone have any other clues. I've search the IBPhoenix knowledgebase.
I've tried the recommendation of disconnecting all clients and starting
again - no change. I can only think that the comments on hardware issue is
the one for me. Any other thoughts from anyone on how to track/test this
would be appreciated. Can I make the assumption that a timeout thru a
traceroute in one direction and not another is correlated to the failure of
socket connection in that direction?

thanks
Alan