Thank you, I tried your approach, but the situation remains the same

I also have started several FastCGI processes in Nginx, so when one process delays waiting for an answer, another one can be used by another user.

As I can see, the problem is not in the error handle, but in the time waiting for a response. Even if it is a Syncronous answer, I do not understand the it stops ALL the application.

Any other suggestion?, or a working example for a similar function?

Miguel