I'm dealing with only one big static IP-pool. We have mostly public webservers (and often control panels), so little private IPs and only public static IPs in the vm-environment.
For testing, I have set up LXC containers on vmbr0.
My concern is that some guest inside an LXC might want to hack servers or internal services (or a weakness in an app for that case). We have a dedicated fw before our network of course, but here we would have some clients on LXC (inside it) - and we can't easly vlan it in transparent fw mode from what I have been told. So they (the LXC guests) would in fact be inside our fw and can connect to every port.
Most servers have software fw inside there again. Is there a way to maybe at least deny connections to a defined set of IPs (outoing from the LXC)? I can set it up on the LXC-container through the GUI of course and that works, but this can just be disabled or removed by the user.
I tried to define fw at cluster and datacenter level, but that doesn't impact the LXC from what I have been able to test quickly.
For testing, I have set up LXC containers on vmbr0.
My concern is that some guest inside an LXC might want to hack servers or internal services (or a weakness in an app for that case). We have a dedicated fw before our network of course, but here we would have some clients on LXC (inside it) - and we can't easly vlan it in transparent fw mode from what I have been told. So they (the LXC guests) would in fact be inside our fw and can connect to every port.
Most servers have software fw inside there again. Is there a way to maybe at least deny connections to a defined set of IPs (outoing from the LXC)? I can set it up on the LXC-container through the GUI of course and that works, but this can just be disabled or removed by the user.
I tried to define fw at cluster and datacenter level, but that doesn't impact the LXC from what I have been able to test quickly.