Thanks Pete.
What I do now is I have my own library of calls for opening a connection (which
is really my own internal system of checking to make sure the parameters are in
order and setting a global if successful) then for reconnecting, which is
really making a connection in the LC sense, the
I'm assuming this is mySQL?
There is a server setting that determines the inactivity timeout period.
It's either wait_timeout for a non-interactive client or
interactive_timeout for an interactive client. I assume that LC opens
mySQL dbs as an interactive client. According to the mySQL document
I was getting disconnected from On-Rev within an hour of inactivity, and all my
database calls failed when the connection was closed by the host. Also, certain
network outages or problems can cause managed switches to reset, notably
excessive broadcasts, and when this happens, the SQL connection
well "COMMIT/ROLLBACK" should handle the possibility of data loss.
As far as "connections" I've found that there is little time difference
between being 'always connected' and making a connection open and close per
transaction, unless one is hitting it repeatedly for a single result (as I
had to t
