>AFAIK the only way to check a connection is to do a query and see if
>it raises an exception.
I have thinked a possible solution. I readed this info:

http://www.sqlalchemy.org/docs/dbengine.myt#dbengine_options
http://www.sqlalchemy.org/docs/pooling.myt#pooling

I think its possible to write a class for get a connection from the pool and
execute a query. The pool will do the 'hard work'. With
     'pool_recycle=3600' the pool recycle connections after one hour, so MySQL
connections always 'lives'.  But,  what happens if the pool recycle a connection
while an user execute a query?
I haven't found info about it.

On the other hand, I would like to use only the needed classes of SQLAlchemy for
pooling, but I think it is not possible because 'create_engine' returns a
SQLAlchemy's connection object.

In my view I have three options:
a) Use one connection for each process and write a cron job for restart my
application each 6 hours.
b) Use a connection pool (SQLAlchemy) with 'pool_recycle'.
c) Write a class with a method where I ask to pool for a connection database,
execute a query and return the connection to pool. Each query must use this
method..

What do you think about?

----------------------------
Arturo Fernandez (artufm at yahoo dot es)





______________________________________________
LLama Gratis a cualquier PC del Mundo.
Llamadas a fijos y móviles desde 1 céntimo por minuto.
http://es.voice.yahoo.com