@aniodon

Apologies for the necro posting.

How would you apply such a patch ?

https://github.com/pfsense/FreeBSD-ports/commit/476a7d0e3dca704b236839970f1d215912184f73

I've created a system patch via pfsense GUI with the commit mentioned in this thread, however, it does not apply to my squid config,

It seems that it is not recognizing the file to update (?)
Here's what I'm getting in the debug log (not having changed default patch settings):

/usr/bin/patch --directory='/' -t --strip '2' -i '/var/patches/682f24bdbc39f.patch' --check --forward --ignore-whitespace Hmm... Looks like a unified diff to me... The text leading up to this was: -------------------------- |From 476a7d0e3dca704b236839970f1d215912184f73 Mon Sep 17 00:00:00 2001 |From: Marcos Mendoza <mmendoza@netgate.com> |Date: Tue, 26 Nov 2024 18:36:53 -0600 |Subject: [PATCH] www/pfSense-pkg-squid: remove duplicate option | |--- | www/pfSense-pkg-squid/files/usr/local/pkg/squid.inc | 1 - | 1 file changed, 1 deletion(-) | |diff --git a/www/pfSense-pkg-squid/files/usr/local/pkg/squid.inc b/www/pfSense-pkg-squid/files/usr/local/pkg/squid.inc |index 719cda2fb3cf..129b8b05335c 100644 |--- a/www/pfSense-pkg-squid/files/usr/local/pkg/squid.inc |+++ b/www/pfSense-pkg-squid/files/usr/local/pkg/squid.inc -------------------------- No file to patch. Skipping... Hunk #1 ignored at 1236. 1 out of 1 hunks ignored while patching pfSense-pkg-squid/files/usr/local/pkg/squid.inc done

Got damn, there are one unused backend in HAproxy package configuration! How I miss it?…

After delete AND cold pfSense-based server restart - error not appear at all.

@TheCyborgWeasel said in haproxy 0.63_2 weird behavior, edits not working:

Yes, I can confirm that, at least partially. I simply had to enable SSL on the backend and thought the connection was secured. The SSL server was not available so I should have seen a 503. But haProxy did continue to gain access to the webserver which means that the settings were not applied correctly. Also a restart of the service did not help. Only a restart of the whole pfSense worked and applied the new SSL setting. I got a 503 then.

Strange: The other way around worked directly. Disabling SSL and hitting the apply button showed immediately the old website again, no need to restart the haProxy service or even the whole pfSense. So it is partially buggy :-)

This is reproducable!

Is this enough to open a case?

Thank you for confirming that there is a bug in haproxy.

Did'nt you forget the ACL? The action will never be triggered...?!

Bumping this as I am experiencing the exact same issue with the exact same behavior. I have even tried putting a transparent bypass for 127.0.0.1 as the source and destination, the hostname of the firewall, and the firewall's own public address as a source with no success.

@sensewolf said in Can't protect certain path only with client certificate:

-- The expected outcome is that in order to access the specific path, a client certificate is required. Surprisingly, however, the path becomes publicly accessible again without the client certificate --

I don't understand why this doesn't work. The setup is basically the same as for my other accessible and protected domains with the only difference that in this case only a certain path should be protected.

Did you put this rule to the top, so that it is probed and executed before the other one?

For testing the ACLs just use a simple rule, which give a clear result like "http request deny".

Why isn't this working? What am I missing?

Maybe someone will see it if you post the whole configuration.

Hi there,

Thanks for the detailed explanation—this is a good catch. Your suspicion is right: this issue isn’t caused by pfSense or HAProxy, but by the web server behind Nextcloud (likely Apache or Nginx) not serving .mjs files with the correct MIME type.

HAProxy simply acts as a reverse proxy and does not handle or modify MIME types. It forwards the request to your backend server (10.0.24.10:3334 in this case), which is responsible for serving static content like .mjs files. To fix it:

For Nginx, modify mime.types to include: application/javascript js mjs; For Apache, add: AddType application/javascript .mjs

Once this change is in place, the viewer and other JS modules should load properly through HAProxy.

As a side note, if you're ever doing advanced tasks like proxy rotation for web scraping or external API access, that’s something HAProxy can help with—but it doesn’t apply here.

Let me know if you'd like help locating your web server config!

@sensewolf All of the applications I have that run websockets use the same port, so I do not create separate backends for them and it works fine. If your application does use separate ports then you will need to create a separate backend. I'm not sure how common this is, but I have I think 5 domains with websockets and none of them use a separate port.

Did you manage to get it working as I am doing the same thing but have noticed Cloudflare Proxied traffic seems to really be slow......not sure if there is something in Cloudflare that needs tweaking but it is pretty much unusable

@viragomann said in Nextcloud und SSL über pfSense Configurieren:

https://<WAN IP>

Sooo Fehler gefunden, es waren die IPv6 Einträge, gelöscht und es funktioniert!

Über DSL von Zuhause konnten alle anfragen Aufgelöst werden, über bein Test gerät welches über Mobiele Daten ging nicht... da dieses vermutlich IPv6 genutzt hatte.

Besten Danke für die unterstützung

@ngr2001 I don’t know try the directive to see it if works, I do not have that crypto chip

@tiago-duarte squid + pfSense plus

@Sebastian74

If you're using TCP mode, make sure that SSL passthrough is configured correctly for both backends.

If you're offloading SSL at HAProxy, ensure that the certificates are properly assigned and that backend communication is happening over HTTP or correctly re-encrypted HTTPS.

I'm using every of this two option because one server have the certificate inside it, the other one is certified by HA Proxy and the cert is on Pfsense.

I need to mix this two mode or tell me another way to do this correctly, i can't find anything for this case scenario.

Thank you

Wow, I am some glad I finally found this. I've been racking my brain on it since December, and holding off upgrading 2 production instances of pfSense to 24.11 as I thought the problem was related to 24.11, and not MIM. I'll now be able to do the upgrades on those 4 other units(2x HA pairs in remote datacenter sites).

CPU and memory all good. This is a brand new install. Why would I need to rebuild the cache?

Thank you!