pg_timetable: PostgreSQL scheduling on steroids

pg_timetable has been written by us and implements a sophisticated PostgreSQL based scheduler for database tasks. It has been written in Go (codes and Github can be provided). What is the purpose of the thing? There are many reasons we wrote it: a.) cron is not powerful enough. It does not prevent that the very same job is executed twice at the same time b.) cron can execute at a certain point in time but it cannot execute "as soon as possible". c.) cron does not take parameters and all it does it launching a script pg_timetable is different. It stores all its configuration in a PostgreSQL database. It reads this stuff once a minute or so and launches a job. pg_timetable does not only launch a job - that would be too simple. We got a bit different semantics. We got: base tasks, builtin tasks and chains. What we actually schedule is chains. We also preserve a transactional semantic. So what is a base task? Base tasks could be ... - send an email given a certain mail template - run some SQL. e.g.: SELECT func1($1), func2($2) FROM tab WHERE f = $3; COPY t_table FROM '$1' CSV HEADER; -- $1 = filename - run a shell script ./[login to view URL] url ... This allows us to form chains which might do the following: BEGIN; ./[login to view URL] url ... SELECT aggregate_data($1); INSERT INTO log ... COMMIT; ./delete_downloaded_files ... INSERT INTO log ... builtin_send_mail(...) As you can see the idea is to mix elements of SQL and shell and builtin stuff while maintaining PostgreSQL transactions. Currently pg_timetable is pretty good already. It works but it is not heavily tested, poorly documented, lacks documentation and most importantly examples. There are tests already but we want to expand them and run things for a while to see, if it reliable. It will be the backbone of one of our products and it will be released as Open Source once it is stable.