It looks like it is possible to connect via https but no way to load a client certificate. With out the client certificate there is no way to authenticate the client using the certificate. username/password at the database level is just not sufficient. Access should be blocked/granted at the certificate level. There needs to be a way to add the two certificates and private key to the pfsense telegraf client.

@freak4915 said in pfSense, Haproxy, cloudflare cname DDNS letsencrypt certs Timeout:

IPv4 TCP * Source * Port This Firewall Destination 443 (HTTPS) Port * Gateway

No exactly sure how to read that, if you have a gateway filled in in the rule can you remove that? Other than that there shouldn't be any issues with the config you have.. Not regarding a client-connection timeout anyhow..

All the NAT and reflection settings should be irrelevant for the connection through haproxy..

On the haproxy stats, are the frontends counting incoming connections? I think not.. as i would then expect a different error result in the browser or perhaps a proper response..

Can you check on pfSense console that connection attempts are actually arriving on your wan interface?
Login with ssh go to the shell, then run something like

tcpdump -ni WAN-NIC port 443 and host 1.2.3.4

where WAN-NIC must be the name of the actual nic like em1 or igb2 or rt3 vtnet4 or pppoe5 so what your wan nic is actually called ;) and the 1.2.3.4 must be the public ip of the client that tries to connect. Then look for the [S] packet that tries to connect and see if a [S.] is send back for a request made from outside.. if so then haproxy (or at least something) apparently accepted the connection.. but im suspecting that modem or isp might be blocking traffic already.?.

Linked issue:
https://redmine.pfsense.org/issues/10445

It is CLI but you can run it from the GUI through Diagnostics>Command Line.

@sparkman123 said in Pkg upgrade or pkg delete/install:

Thank you for the info.

I think it should be pointed out that I'm not the first to try to use dante- there's this post, this one and this tutorial. If you do a forum search you'll find quite a few hits on other posts.

The utility I'm installing is dante, or sockd, which is a fairly well fleshed out distro (it has its own freebsd manpage). I'm also installing a small number of support packages, so that limits the potential performance degradation vector.

Finally, I'm doing all my testing on a VM for weeks prior to rolling it out on bare metal.

Your initial point is correct though- dante and its supporting libs are not supported by pfsense out of the box and thus could cause problems.

But not all packages are created equal- some pose more risk than others. Case in point, before I settled on dante, I looked at another socks proxy, SS5. While it is in the freebsd packages library it is not nearly as well documented and does not appear to be as well vetted as sockd. It is my judgement that sockd poses less of a risk than SS5.

Others are free to come to their own conclusions, however. Some might not want to take any risk at all. If the need for functionality outweighs the risk my suggestion would be to look through the code of a package, read its documentation and see how many others have posted/played with it before coming up with a final decision.

Just wanted to be sure you were aware of the risks. When folks post it is sometimes hard to determine their "skill level" with pfSense and firewalls in general. So I tend to start with the more elementary things first and move up from there depending on my perception of the poster's skills. We get all kinds here ... 😉. More than once folks have come here asking why they can't get pfSense to install under Windows. They actually think it is just some application you install like Microsoft Office or something ... 😕 .

Sounds like you are aware of the risks and willing to accept them.

An update that is actually not an update.

I have tried to factory settings and reflashed the firmware but same result.
I have searched the internet an only found an old post on this same forum but the solution for them was reinstall the package (thing that did not work for me).

I have tried to wipe the package configuration removing the folders /var/etc/haproxy but configuration is still there so I have no clue how to reset the package... Any clues here?

In the end I have finished installing haproxy on a VM on my hypervisor. I lost the convenience of having a GUI and the auto renew of the certificates.. but I guess I'll have to live without that.

@andrewK said in BIND DNS recursion points to what by default?:

at the very least those 13 IPs stored in system by default...?

On most of my Debian systems I have a file in /etc/bind9/ that lists these servers :

db.root :

..... ; FORMERLY NS.INTERNIC.NET ; . 3600000 NS A.ROOT-SERVERS.NET. A.ROOT-SERVERS.NET. 3600000 A 198.41.0.4 A.ROOT-SERVERS.NET. 3600000 AAAA 2001:503:ba3e::2:30 ; ; FORMERLY NS1.ISI.EDU ; . 3600000 NS B.ROOT-SERVERS.NET. B.ROOT-SERVERS.NET. 3600000 A 199.9.14.201 B.ROOT-SERVERS.NET. 3600000 AAAA 2001:500:200::b ; ; FORMERLY C.PSI.NET ; . 3600000 NS C.ROOT-SERVERS.NET. C.ROOT-SERVERS.NET. 3600000 A 192.33.4.12 C.ROOT-SERVERS.NET. 3600000 AAAA 2001:500:2::cl: etc. ....

I guess changing to an automatically assigned address is still a MAC address change. I couldnt find anything in any manuals about if reporting these is normal or not for Arpwatch and I never noticed before so I'm just looking for confirmation. Not really a big deal and potentially helpful for DHCP issues. I believe this one was due to the computer being in sleep mode when the DHCP address renewed but that's just a guess.

@liboriolibs said in pfSense + Haproxy - internal LAN redirect backend with acme valed certification:

screenshots from frontend and backend please
also screenshots from your acl (host match)
and screenshot from dns resolver

brNP

Ok, so resetting the data has helped a bit. I've waited before coming back, because I'd hoped that the daily data would recover, but there is still no data there under "Data Summary". However, the rest appears to be working meanwhile.

I'll say "thanks!" at this point, because with this, I can get the e-mails to me working again, which is the part I need.

Thanks!

mongodb34 is not longer supported.. EOL Jan 2020..

So yeah it was prob pulled from repo... Run your controller on something else, or using VM to do such things as firewall and controller on the same hardware.

I got my controller using 3.6 running on a VM on my nas.. Just no place to run your controller on your firewall if you ask me.. Get yourself a clould key if you have no where to run it, or run it in the cloud..

@coatmaker618
Indeed just rename the topic. There is no 'solved' button.

Your front end ssl cert does not have to match the backend cert. Where you get the certs is up to you.. If duckdns will create a dns entry for you to grab a cert via acme, that works.

You can also use wildcard certs so *.whatever.tld works..

If page looks broken - normally this points to css not loading.. Have to look to how the site tells the browser to load the css file.

Yeah eap-tls is great, but you don't have to manually edit the eap.conf for it to work - can all be done in the gui.

You pick the CA and certs from the gui.

tls-config tls-common { private_key_file = ${certdir}/server_key.pem certificate_file = ${certdir}/server_cert.pem ca_path = ${confdir}/certs ca_file = ${ca_path}/ca_cert.pem dh_file = ${certdir}/dh random_file = /dev/urandom fragment_size = 1024 include_length = yes check_crl = no check_cert_cn = %{User-Name} cipher_list = "DEFAULT" cipher_server_preference = no ecdh_curve = "prime256v1"

Example that is from my eap.conf - snipped out some stuff.. But you can see the path to certs and ca, and the cipher list, etc.

So was just curious why you were trying set specifics for your list and curve, etc.