I have a current project I am working on. It basically requires me to build out a solution to intelligently take an action whenever scheduled jobs fail in AutoSys. Since some jobs fail due to connectivity issues resolved with re-runs or disk space or memory issues. I have a basic structure for the code below. Current Architecture Gather AutoSys Report with Job error information. Feed it to a python script, scheduled using AutoSys. Feed a action config file to autosys. Have the python script, choose from best course of action based on the error logs from the action list in the config file. Then execute that action. This will all be done before job failure reports are sent out through an email system. This way if an action could not be taken, we could work around by manual intervention. Questions What would be the best method to intelligently determine action to be taken? Natural Language Processing? Execution time isn't the biggest concern for my usecase. A hardcoded algorithmic approach? Like a fuzzy search matching the error output to an action in the config file? Or something else? Are there any resources I could look into for this? Continue reading...
