See this howto: https://poolp.org/posts/2019-09-14/setting-up-a-mail-server-with-opensmtpd-dovecot-and-rspamd/

I have delivery to the inbox of all major providers using this. Email is not that hard…

Been hosting my own email for over 20yrs. Get a vps at some reputable hoster, and make this a dedicated mailserver. Be sure to setup all dns records that are required, and rollout antispam measures.

Thing is, you grew up in the pioneering age of computing, and in that time you needed to do everything yourself. This gave you a bunch of skills for free, that are hard to do today, because most of the hard stuff is automated away and snuck behind a gui and/or containers.

Alma, Talos, OpenBSD

I run unattended-upgrades on all the debian/ubuntu deployments I manage. One of the deployments even has automatic reboots enabled. I still do major upgrades by hand/terraform, but the process itself works flawless in my experience.

OpenLDAP multi-master with a bunch of custom schemas.

Syslog is what you are looking for.

Can you explain what you mean with lightweight?

SCO OpenServer

You expressed yourself just fine and my question is still valid. Do you have the capacity to handle multi Tbit traffic on the edge ips that you use to hide the backend ips? Because if all of those are flooded, not only will the backend app be unreachable, but all your customers will be unreachable as well.

See a search engine and do some research. Maybe try out a bunch of different distros and find out for yourself

We have a bunch of these at the hackerspace. They run vanilla debian and connect to a terminal server using xdmcp. Works like a charm.

Even if you can get the appZTNA stuff to work (which I doubt), how is your infra going to absorb multi Tbit traffic without customer impact?

You will be in for a surprise once you learn about tcpdump ;-) Welcome to internet hosting!

Some script that parses blocklists into unbound local-data statements, combined with cron and unbound-control

http://www.catb.org/~esr/faqs/smart-questions.html

Read the changelogs from 1.5.3 to 1.5.5 and see if you need to take additional steps

ngircd has the concept of a global password. Distribute this password and you’ve got a simple psk based acl mechanism

what error message are you getting?

It all depends on what you do with the box, how the applications you run utilize the cpu (single vs multithreaded/multiprocess but also stuff like cpu cache utilization. if you profile the workload that you want to run (see if its cpu, memory or io bound for instance) will help you in figuring out what works best for you.