Menu

Show posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Show posts Menu

Messages - nashmeira

#1
General Discussion / IPsec VPN Lan Isolation
May 18, 2019, 12:05:07 AM
I really kind of new to IPsec VPN tunnels, as-in I've only setup one. I have the tunnel setup and Phase 1 & 2 show all good and connected.

The question I now have is can I isolate all traffic going over this VPN to one Lan that I created just for it. I'll refer to it as the VPN-Lan. I have one VM computer on the VPN-Lan and I can on it. The VPN-Lan has its own network all by itself.

The remote subnet I'm using for Phase-2 is 10.0.0.8/8, and the local subnet is 192.168.41.0/24 (VPN-Lan)

I'm just not sure how or what the rules would need to be. I did notice that under Rules there is now an IPsec connection. So I don't know if I should edit it or edit the rules for the VPN-Lan.

Other notes we do have an SSL VPN setup as well for a remote users and it seems to work fine.
#2
General Discussion / NAT Reflection for Web Servers
April 02, 2019, 11:31:15 PM
I have a few webserver on a DMZ that I have setup with OPNsense. Everything works find from outside the LAN-1 however internally we are not able to access these servers. I am using Port Forwarding not 1-to-1.

Looking around online it seems a simple enough task to turn on NAT reflection: FIREWALL > SETTINGS > ADVANCED > Network Address Translation enabling the following settings.

ENABLED - Reflection for port forwards
ENABLED - Reflection for 1:1  (Even though I'm not using 1:1)
ENABLED - Automatic outbound NAT for Reflection

Each of the sites have a Static IP with ports 80, 443 forwarded to the internal servers IP on the DMZ-1.

NAT Forward Rules look like this:
^Source
IF: WAN1
PROTO: TCP
ADDRESS: *
PORTS: *

^Destination
ADDRESS: PUBLIC-WAN IP: 99.88.77.66
PORTS: 80,443 (Created as an Aliases)

^NAT
IP: INTERNAL SERVER IP: 192.168.10.10 (Created as an Aliases)
PORTS: 80,443 (Created as an Aliases)

All other settings are default. NAT reflection uses System Default, Filter rule association uses Rule NAT: Site-1 (The info from the rules description).

Even though I have NAT reflection enabled nothing seems to help if I'm on the internal LAN-1 network. I tried enabling NAT reflection in the individual rule but still nothing. So, I'm kinda stumped. I can see everything from outside but nothing from inside.

One thing I have had happen is if I try to go to one of the sites using the url (From LAN-1) the port 8443 is added to the end, which then loads to be the logon page for OPNsense. So, I am wondering if there is a firewall rule I need to edit?

My setup also uses two different internet sources. So, there is a multi-gateway setup for fail over. But this does not seem to affect the external to internal DMZ-1 traffic, nor can I reach the OPNsense interface from outside when testing.

I'm really liking OPNsense and I know its just lack knowledge that is the issue. I have tried referencing some pfsense articles but even still I've had no luck figuring this one out.
#3
Quote from: bartjsmit on March 16, 2019, 10:55:01 AM
- Create a management network to access your firewall, hypervisors, network devices, etc. Separate switches if possible, separate VLAN's if not.

There is a kind of, not really, mgnt network in place. But its on the same subnet as the LAN1. What the person before setup was EXAMPLE: 192.168.110.0/23 with all the users and Voip on 192.168.110.X and all the servers, and hardware (Printers, Displays etc) on 192.168.111.X so uhh yeah... I put that on my list but have it as a low priority at the moment. Biggest fear was the DMZ setup and what was there. I'll get to the really scary thing in just a moment.

Quote from: bartjsmit on March 16, 2019, 10:55:01 AM
- Don't allow any traffic from/to your firewall on the production network. Only allow traffic *through* it. The firewall should make for a hole in the network. I've deployed (not OPNsense) firewalls that increment the TTL so that they don't even show up in traceroute. Management web interfaces, SSH access, etc. all listen on your management network only.

I think this is part of the above in the regards I need to see about getting all of my services requiring control on another network and off LAN1.


Quote from: bartjsmit on March 16, 2019, 10:55:01 AM
- Have as little database/directory/business information in the DMZ as possible. Database authentication for your websites, RADIUS for AD logins with RODC as a last resort. The full data assets stay on the internal network.

All of the DB's on the DMZ are part of the individual web servers. There all built on Ubuntu from what I have found to there just LAMP stacks, am I saying that right, and are currently 16.05. only a few ports are open like 80, 443 and 22. SSH is setup with a pem for what that's worth, I would think more than password.

Quote from: bartjsmit on March 16, 2019, 10:55:01 AM
- DMZ servers should be stateless, ideally able to be deleted at the slightest whiff of suspicion, and quickly rebuilt through orchestration. Servers are cattle, not pets. Load balancers make this low impact to your clients.

The VMs are all backed up daily at night so should/if/when the unthinkable happen they can be restored pretty fast. Seven days' worth are kept before retention removed the oldest.

Quote from: bartjsmit on March 16, 2019, 10:55:01 AM
- VLAN's are cheap. If webservers have less than a few dataflows between them (ideally zero) put them on separate DMZ's. Trunk your VLAN's through resilient physical interfaces to ESXi and OPNsense to reduce your cabling. Consider making OPNsense a virtual server to benefit from vShpere HA and vNIC's instead of VLAN's.

They my diagram doesn't show it OPnSense and all of the systems other then those in the LAN1 bubble at the top of the diagram are VMs. The host: QUARTA actually hosts OPNsense as well as a DC and RODC. The DC her is not an AD DC that runs LAN1 but was a separate DC used just for the people that needed access to the websites and the nextcloud setup. So, all the user accounts for those users was on this DC. LAN1's AD DC is completely separate and ny the two shall meet.

Here is the scary part. The RODC was not always there, before the DC1, and its backup DC2 were part of the DMZ. Again, I'm no network engineer/architect what have you, but OMG that scared the crap out of me. So, the first thing I did was read up on what to do and RODC was my quickest solution. The DC1 and DC2 are running a directory controller called Univention, and I kind of like it. I wish the community was larger so if I had an issue I could google fu it and find answers like I can for Active Directory, but we work with what we have. Again, DC1 and DC2 are no longer on the DMZ, and LAN1's ADDC1 and ADDC2 are not touching.

Quote from: bartjsmit on March 16, 2019, 10:55:01 AM
- IPS is a must. Mistrust your DMZ hosts with the greatest of paranoia. Even if you have restricted the traffic by firewall rules to what is allowed, you still need to make sure it follows normal patterns.

Way a head of you on paranoia, now I just need to get better at what I'm looking for as far as trouble goes. X3

Quote from: bartjsmit on March 16, 2019, 10:55:01 AM
- Use a distributed vSwitch to centralize your port group management if you can afford the vSphere Enterprise+ licence, or if you need that for other features.

I have not really read up on the Distribution vSwitch functionality or what/how it differs from a normal vSwitch. I'm still learning a lot about VMware. Our vCenter is just Standard 😐 but better than no vCenter at all.
The only saving grace is at home I use Xen a little for my NAS, VPN and Dev server for my Automation projects. So I have some experience with virtualization, granted Xen and VMware are different but it helps, but I am by no means a master of ether.

I guess that's all I can think of, thanks for the insight and help.
#4
Quote from: rungekutta on March 16, 2019, 08:49:44 AM
I'm not a professional network engineer either but here's me 5c and take them for what it's worth. Some considerations are necessary here which may help drive the decision; do the servers in the DMZ need to talk to each other or are they all completely stand-alone? What are they actually running in terms of OS and applications and how secure......

Only one server has to talk to another, that being a file share server, (nextcloud) that talks with onlyoffice, which is internal only and does not accept public traffic. Its now setup to only communicate with the nextcloud server on the internal DMZ network. The other servers have no reason to talk with each other at all.

Kinds of server in the DMZ are 3x Website servers, including Nexctcloud, 1x FTP server, 1x proxy server.

Thanks for your insight.
#5
Hello o/ first post here. I'm looking for some insight on a network design. Let me first say that I am not a network, hardware, system engineer or any of that. So please forgive an ill use of terminology. While this is not my forte I do have the internet and I can read so we work with what we have.

There is however a lot of conjecture on the internet. Since I have recently been somewhat thrown into a position that requires me to assume the role of a network administrator and engineer I am looking for is some input on how the network we have is setup and if there is a better way to deal with it then what I am currently planning.

As of right now we have 2 separate networks, with 2 different firewalls. One for the DMZ and one for the office LAN. The DMZ plays host to about 6 different servers, some web servers, some other. All hosted in VMware ESX across 3 deferent hosts. Networked to the DMZ switch (LAN Switch is separate) then through the firewall. All the servers on the DMZ port forwarded through the firewall by way of their external static IP.

The part that is worrisome to me and why I am attempting to change and rock the boat is there all on the same DMZ LAN. IE: Server1 = 192.168.20.11, Server2 =192.168.20.12, Server2 =192.168.20.13 etc. All are on the same vlan and can see each other ping etc. Nothing is stopping them from communicating amongst themselves, they are not but could.

So as of now here is a rough idea of how things go:
wtf1.shft.com(Static Public IP: 123.1.1.1) > [FIREWALL:Port forward 443] > Server1_192.168.20.11
wtf2.shft.com(Static Public IP: 123.1.1.2) > [FIREWALL:Port forward 443] > Server1_192.168.20.12
wtf3.shft.com(Static Public IP: 123.1.1.3) > [FIREWALL:Port forward 443] > Server1_192.168.20.13

I guess the real question I have is would I be better to have two networks each with there own firewall or two networks with one firewall? I tried to do up a simple version of whats in my head and what I planned on implementing to try and get the server own there on and not all on the same vlan and subnet.

While separate I planned on using something like DMZ-1=192.168.10.XXX, DMZ-2=192.168.20.XXX, I know there separate but logically in my head it helps not to confuse them if there different.

I would greatly a appreciate and input or suggestions. If you have any question from me, I'm sure some clarification will be needed, and I will answer to the best of my abilities.

Thanks, in advanced.  :D

https://pasteboard.co/I5yC2al.png