Dead connection detection with SQL Server Clustering

[nshen]

Hi! My application retries the last SQL query when a conncetion went down then came backup again. This worked well with standalone SQL Server 2000.
But in a SQL Sever 2000 clustering, if one of the nodes went out then came back up again, the dead connection attribute SQL_ATTR_CONNECTION_DEAD was never set, therefore the connection was not usable yet it was not "dead".

How can I detect a dead connection in a clustering situation? I was hoping the cluster itself would automatically replace the bad one with a valid connection but I guess this might be either too mcuh to ask for or maybe I missed a cluster setup parameter.
Any hints or tips are appreciated!

Regards, Nancy

 

[zemp]

What does this have to do with ODBC? I think that this is a question for an SQL Server forum. Try the link to the SQL Server Programming forum found in the 'Related Forums' section on the right side of this page.

zemp

 

[zemp]

Double post, see thread183-870016.

nshen, read faq222-2244 for information on why double posting is frowned upon.

zemp

 

[nshen]

I posted it here because after more testing, we found INSERT/SELECT type of SQL queries went through the connections fine after the lost SQL Server came back up again. But all the stored procedure calls were "put on hold" as if the connections were hanged.

This is definitely something that only ODBC client could know.
Thank you.
regards, Nancy