I am trying to configure a policy in fail2ban - the detection and identity of bad behaviour is rather complex in my case, so I was planning to handle this in a dedicated program. Rather than having that program writing to the log, and fail2ban polling the log to see if it should act, I thought it would be simpler to invoke
fail2ban-client banip directly from the detector. however fail2ban is refusing to recognise my jail config because it doesn't have a filter defined.
Must I provide a log file and filter? Is there a way to convince fail2ban to work without this? If I need to tell fail2ban about a log files, does it have to exist on the filesystem?