Hi all, First - if this is the wrong spot for such a question, I apologize ahead of time! I'm hoping someone who knows better can tell me if I'm on the right track or not. I have written a desktop application that is used throughout my company. This application is a wxPython GUI that controls several 3270 terminal emulators in the background as the user interfaces with the GUI. The terminal emulators are used for both screen scraping as well as entering data automatically into several systems, and due to company politics I cannot obtain the needed data any other way. My company wants to take this client program and convert it over from a desktop application that runs on end user PCs to a web based interface. This is where I'm hoping someone with more experience may be able to offer some suggestions. As I stated before - even when I convert this over to a web based system, I am forced to enter and scrape data using 3270 terminal emulators (you can laugh if you want to). The basic idea I have to make this work is as follows: 1 - Create web front end where the end user can enter jobs, view existing jobs, and cancel any jobs that have not been processed. 2 - When user enters a job, the job would be inserted into a job queue table on our postgresql db. 3 - All of the code using the terminal emulators would then be moved into a deamon program. 4 - We would run multiple copies of the deamon program, and each one would monitor the job queue table for new jobs to process. The idea is that we would run as many copies of this deamon program as necessary to make the system scale as needed. We could assign each job in the job queue a status flag that would have a meaning such as "job requested by user", "job assigned to deamon for processing", "job completed", or "job cancelled". Each deamon would periodically check for newly requested jobs, and for each new job found it would mark the status flag "job assigned for processing". Then as each job is processed the daemon would insert the job results into the db and then mark the status flag as "job completed". This is where I hope you guys can help. As this would be my first postgresql app, I'm not very familiar with how locking works. Should I anticipate any sort of locking issues with this design given that I would have several daemons attempting to access/change several tables simultaneously or will this be mostly transparent so long as I use transactions correctly? I would also like to hear about it if you think this is a very bad approach! Thanks for your time! Jeff Peck
