To answer your question on security of honeypots a little further:
I'm currently writing my theses on honeypots - security-risks of honeypot systems are a big issue. A honeypot should never be run in reach of a production system. Set up in its own VM, cut all outgoing traffic via a firewall. Most open source low interaction honeypot systems, like
Artillery, are considered secure. But as is with every software, there is no such thing as a guarantee for being bug free.
High interaction honeypots tend to be more vulnerable than low interaction systems, which is e.g. why
Cowrie states in its FAQ on the Question "Is Cowrie secure?" a 'maybe'. Unless you know exactly what you are doing, I would really recommend not to run something like cowrie on a production system. Artillery can be used, though, I think. It is not the best practice, of course.
www.nestle.de is a good example for a low-interaction honeypot. Portscan it, you will see a service jungle like never before.
If you have any questions, feel free to ask. I'm happy to have more practical problems as input on for my thesis!