Simple Oozie Job Dispatcher

>>> About Me I'm an IT contractor working on a Hadoop project for one of my clients. I'd like some help to accelerate the delivery. >>> The Job Implement the following using Oozie in Cloudera CDH 5.8.3 Oozie must watch for a file arriving in a local directory d with a mask m. The value for d and m should be configurable. Example values are: d: /tmp m: customer_????????.txt where ? means any character. File-watching must use built-in Oozie features. An example would be Oozie's file coordinator. Hand-coded file watching must not be used, for example shell scripts around the linux inotify or watch commands. File-watching should use an event-trigger rather than polling if it is available. A file is considered stable if its size has not changed for n seconds. The figure for n can be a system-wide Oozie setting, rather than job-specific. For every matching file, run a Hive job h. The name of the matching file should be passed to the Hive job. The Hive job itself is not in scope - a dummy job is acceptable for testing purposes. Example: /tmp/[login to view URL] arrives, run Hive job h with /tmp/[login to view URL] passed to the job. /tmp/[login to view URL] arrives, run Hive job h with /tmp/[login to view URL] passed to the job. Multiple files can arrive at the same time. I don't mind if they are processed serially or concurrently. There is no requirement to process them in order. A failed job must not prevent further jobs from starting e.g. in the example above, [login to view URL] should be started even if 0001 did not process successfully. A failed job must be rerunnable, either from the command-line or Cloudera UI. Describe the steps. No formal documentation is required - a text file is acceptable. Please price these separately if it contributes more than 30% of the overall job cost. The file m should not be reprocessed if the timestamp or size changes. The file m should not be reprocessed if the Oozie job is restarted. The file m should not be reprocessed if the host is restarted.