I’ve restarted the AWS instance from scratch to try and tackle down the problems from my other post. This setup is pretty particular since it’s under a pfSense with 2 private IP addresses in different NICs (ens5, ens6) and LANs (one for management), so when the install cut me off from using ssh to set it up, I rolled it all back using the AMI and started over, this time from the AWS Console.
I found that the script tries to set up FirewallD, but we’re using iptables/nftables to divert what enters/exits to/from where, so here’s when Virtualmin fails to properly install itself.
Any ideas to prevent the Virtualmin installer from activating FirewallD and work with iptables/nftables instead?
I guess the installer fails on trying to install firewalld which, dependant on which order firewalld is installed within virtualmin’s dependencies may leave virtualmin without some dependencies. To be fair I have never had the installer failing at this point and I use your suggestion to disable firewalld. This makes me think that the underlying OS has something installed that drops the installer. However this could be an edge case
If they’re losing connectivity because of Firewalld (though it doesn’t look like that’s happening in the install screenshot above?) shutting it down after won’t be possible.
I redid the VM, repeated install, failed again, but forced it a second time after disabling firewalld and rebooting the VM from console. The installer complained it detected another Virtualmin installation in place, but I proceeded anyway and successfully completed without losing connectivity. Weird edge case, I’ll repeat this process since I’m ditching the RDS approach in favor of a monolithic setup (mariadb running in the server). I’ll report back with new details.
