Taz WakeSo, I've been playing with ChatGPT today and, as is often the case, it is really confidently wrong.
LLM/AI tools have genuine value but it concerns me that whenever I ask a question I already know the answer to, it gives very misleading responses.
Now a lot of this is because the internet data is inconsistent - but that really does highlight LLM is basically "Let me google that for you".
In this case, it was convinced you can set the backlog size in the auditd.conf file - I assume because someone has implied it somewhere on Stackoverflow. But you can't. This is set in the audit.rules. It gets weirder when it claims the max_log_file_action can be set to a size in KB. This is just not true.
Don't blindly trust AI.