A facility for event failure management is provided, which includes providing a failed event database containing failed event information relating to failed events and one or more components associated with each of the failed events. Upon modification to a component associated with a failed event, the failed event is retried. Based on a result of retrying the failed event, failed event information of the failed event database is updated. The failed event database may therefore be dynamically and/or automatically updated so that it contains up-to-date and appropriate information for predicting and/or managing event failures.