utm 9 reverse proxy configuration

I have the same problem with my installation. Reverse Proxy for login in and file exchange works fine, but i can not open or edit files with collabora. Every time i try to open a file i get a white screen but the document is not opened.

I think ssl is also on my side not a problem. But i think the utm has a problem with the different names - cloud.... for my nextcloud installation and office.... for the collabora installation. Do you think this is possible?

Here are some logfile entrys:

2017:06:25-22:47:07 remote httpd[31268]: [core:notice] [pid 31268:tid 3995577200] [client 91.17.50.163:59112] AH00026: found %2f (encoded '/') in URI (decoded='/lool/cloud.server.com/.../7_ocukbswiqfwn, returning 404

2017:06:25-22:47:07 remote httpd: id="0299" srcip="91.17.50.163" localip="172.20.96.1" size="373" user="-" host="91.17.50.163" method="GET" statuscode="404" reason="-" extra="-" exceptions="-" time="1568" url="/lool/cloud.server.com/.../7_ocukbswiqfwn server="office.server.com" port="443" query="" referer="-" cookie="-" set-cookie="-" uid="WVAhS6wUYAEAAHokKKgAAACm"

2017:06:25-22:47:08 remote httpd[31268]: [core:notice] [pid 31268:tid 3978791792] [client 91.17.50.163:59113] AH00026: found %2f (encoded '/') in URI (decoded='/lool/cloud.server.com/.../7_ocukbswiqfwn, returning 404

2017:06:25-22:47:08 remote httpd: id="0299" srcip="91.17.50.163" localip="172.20.96.1" size="373" user="-" host="91.17.50.163" method="GET" statuscode="404" reason="-" extra="-" exceptions="-" time="1620" url="/lool/cloud.server.com/.../7_ocukbswiqfwn server="office.server.com" port="443" query="" referer="-" cookie="-" set-cookie="-" uid="WVAhTKwUYAEAAHokKKkAAACo"

Best regards

André

It seems,clear that it is objecting to the url because it contains %2f.

I think if you check adjacent log entries for one starting

"[Modsecurity:", that it will contain a sction of the form [id 123456]    Put that number into the rigid filter exception list.  Or turn off rigid filtering completely (which weakens security more.)

Hi Douglas,

thanks for your quick answer. At the moment i have no filter active, no firewall profile in my virtual servers. This is only for testing purposes, i know that this is not safe ;-). What to do you mean with "adjacent log entries"?

Maybe the .conf for my collabora installation is helpful:

<VirtualHost *:443>
ServerName office.server.com:443

<Directory /var/www>
Options -Indexes
</Directory>

# SSL configuration, you may want to take the easy route instead and use Lets Encrypt!
SSLEngine on
SSLCertificateChainFile /etc/letsencrypt/live/office.server.com/chain.pem
SSLCertificateFile /etc/letsencrypt/live/office.server.com/cert.pem
SSLCertificateKeyFile /etc/letsencrypt/live/office.server.com/privkey.pem
SSLOpenSSLConfCmd DHParameters /etc/letsencrypt/live/office.server.com/dhparam.pem
SSLProtocol all -SSLv2 -SSLv3
SSLCipherSuite ECDHE-ECDSA-CHACHA20-POLY1305:ECDHE-RSA-CHACHA20-POLY1305:ECDHE-ECDSA-AES128-GCM-SHA256:ECDHE-RSA-AES128-GCM-SHA256:ECDHE-ECDSA-AES256-GCM-SHA384:ECDHE-RSA-AES256-GCM-SHA384:DHE-RSA-AES128-GCM-SHA256:DHE-RSA-AES256-GCM-SHA384:ECDHE-ECDSA-AES128-SHA256:ECDHE-RSA-AES128-SHA256:ECDHE-ECDSA-AES128-SHA:ECDHE-RSA-AES256-SHA384:ECDHE-RSA-AES128-SHA:ECDHE-ECDSA-AES256-SHA384:ECDHE-ECDSA-AES256-SHA:ECDHE-RSA-AES256-SHA:DHE-RSA-AES128-SHA256:DHE-RSA-AES128-SHA:DHE-RSA-AES256-SHA256:DHE-RSA-AES256-SHA:ECDHE-ECDSA-DES-CBC3-SHA:ECDHE-RSA-DES-CBC3-SHA:EDH-RSA-DES-CBC3-SHA:AES128-GCM-SHA256:AES256-GCM-SHA384:AES128-SHA256:AES256-SHA256:AES128-SHA:AES256-SHA:DES-CBC3-SHA:!DSS
SSLHonorCipherOrder on
SSLCompression off

# Encoded slashes need to be allowed
AllowEncodedSlashes NoDecode

# Container uses a unique non-signed certificate
SSLProxyEngine On
SSLProxyVerify None
SSLProxyCheckPeerCN Off
SSLProxyCheckPeerName Off

# keep the host
ProxyPreserveHost On

# static html, js, images, etc. served from loolwsd
# loleaflet is the client part of LibreOffice Online
ProxyPass /loleaflet https://127.0.0.1:9980/loleaflet retry=0
ProxyPassReverse /loleaflet https://127.0.0.1:9980/loleaflet

# WOPI discovery URL
ProxyPass /hosting/discovery https://127.0.0.1:9980/hosting/discovery retry=0
ProxyPassReverse /hosting/discovery https://127.0.0.1:9980/hosting/discovery

# Main websocket
ProxyPassMatch "/lool/(.*)/ws$" wss://127.0.0.1:9980/lool/$1/ws nocanon

# Admin Console websocket
ProxyPass /lool/adminws wss://127.0.0.1:9980/lool/adminws

# Download as, Fullscreen presentation and Image upload operations
ProxyPass /lool https://127.0.0.1:9980/lool
ProxyPassReverse /lool https://127.0.0.1:9980/lool
</VirtualHost>

Regards

André

I have been looking at WAF logs recently, and they are  hard to parse.  Suggest you open a prior day log in a text efitor for review.   Some lines start with a timestamp and have basic source-target information, similar to web filter logs.  Other lines do not have that information but do have data about alarms that are triggered.  But any line with [id value] represents a rigid filtering rule that fired.  If you do not want that rule enforced against this website, you put that id number into the rigid filtering exceptions.  There is no master list of rule ids published, you just handle them as they come.   Hopefully rhe source-target information allows you to distinguish known-good traffic from possubly-hostile traffic.

Hi Douglas,

i have been looking in the log with activated firewall profile. I found the 3 following ids and made exclusions - 950120, 960032, 981203. I was getting a warning...

"The list of skipped filter rules contains the following required infrastructure rules: 981203. Disabling a required infrastructure rule can lead to attacks not being blocked by the Web Application Firewall."

...and saved the settings.

But opening a document with collabora is also not working, see logs down below:

2017:06:26-22:36:56 remote httpd: id="0299" srcip="91.17.50.163" localip="172.20.96.1" size="1523" user="-" host="91.17.50.163" method="POST" statuscode="200" reason="-" extra="-" exceptions="-" time="242458" url="/loleaflet/4f4593a/loleaflet.html" server="office.server.com" port="443" query="?WOPISrc=https%3A%2F%2Fcloud.server.com%2Fapps%2Frichdocuments%2Fwopi%2Ffiles%2F262_ocukbswiqfwn&title=About.odt&lang=de-DE&closebutton=1&revisionhistory=1" referer="-" cookie="-" set-cookie="-" uid="WVFwaKwUYAEAAGXQps4AAAAH"

2017:06:26-22:36:56 remote httpd[26064]: [core:notice] [pid 26064:tid 4062718832] [client 91.17.50.163:52232] AH00026: found %2f (encoded '/') in URI (decoded='/lool/cloud.server.com/.../262_ocukbswiqfwn, returning 404

2017:06:26-22:36:56 remote httpd: id="0299" srcip="91.17.50.163" localip="172.20.96.1" size="375" user="-" host="91.17.50.163" method="GET" statuscode="404" reason="-" extra="-" exceptions="-" time="3543" url="/lool/cloud.server.com/.../262_ocukbswiqfwn server="office.server.com" port="443" query="" referer="-" cookie="-" set-cookie="-" uid="WVFwaKwUYAEAAGXQps8AAAAI"

2017:06:26-22:36:57 remote httpd[26064]: [core:notice] [pid 26064:tid 4054326128] [client 91.17.50.163:52233] AH00026: found %2f (encoded '/') in URI (decoded='/lool/cloud.server.com/.../262_ocukbswiqfwn, returning 404

2017:06:26-22:36:57 remote httpd: id="0299" srcip="91.17.50.163" localip="172.20.96.1" size="375" user="-" host="91.17.50.163" method="GET" statuscode="404" reason="-" extra="-" exceptions="-" time="2975" url="/lool/cloud.server.com/.../262_ocukbswiqfwn server="office.server.com" port="443" query="" referer="-" cookie="-" set-cookie="-" uid="WVFwaawUYAEAAGXQptAAAAAJ"

Is it possible that the utm has a problem with the redirection and the 2 different domain names? office.server.com + cloud.server.com

Any other idea?

Regards

André

Hi Douglas,

i have been looking in the log with activated firewall profile. I found the 3 following ids and made exclusions - 950120, 960032, 981203. I was getting a warning...

"The list of skipped filter rules contains the following required infrastructure rules: 981203. Disabling a required infrastructure rule can lead to attacks not being blocked by the Web Application Firewall."

...and saved the settings.

But opening a document with collabora is also not working, see logs down below:

2017:06:26-22:36:56 remote httpd: id="0299" srcip="91.17.50.163" localip="172.20.96.1" size="1523" user="-" host="91.17.50.163" method="POST" statuscode="200" reason="-" extra="-" exceptions="-" time="242458" url="/loleaflet/4f4593a/loleaflet.html" server="office.server.com" port="443" query="?WOPISrc=https%3A%2F%2Fcloud.server.com%2Fapps%2Frichdocuments%2Fwopi%2Ffiles%2F262_ocukbswiqfwn&title=About.odt&lang=de-DE&closebutton=1&revisionhistory=1" referer="-" cookie="-" set-cookie="-" uid="WVFwaKwUYAEAAGXQps4AAAAH"

2017:06:26-22:36:56 remote httpd[26064]: [core:notice] [pid 26064:tid 4062718832] [client 91.17.50.163:52232] AH00026: found %2f (encoded '/') in URI (decoded='/lool/cloud.server.com/.../262_ocukbswiqfwn, returning 404

2017:06:26-22:36:56 remote httpd: id="0299" srcip="91.17.50.163" localip="172.20.96.1" size="375" user="-" host="91.17.50.163" method="GET" statuscode="404" reason="-" extra="-" exceptions="-" time="3543" url="/lool/cloud.server.com/.../262_ocukbswiqfwn server="office.server.com" port="443" query="" referer="-" cookie="-" set-cookie="-" uid="WVFwaKwUYAEAAGXQps8AAAAI"

2017:06:26-22:36:57 remote httpd[26064]: [core:notice] [pid 26064:tid 4054326128] [client 91.17.50.163:52233] AH00026: found %2f (encoded '/') in URI (decoded='/lool/cloud.server.com/.../262_ocukbswiqfwn, returning 404

2017:06:26-22:36:57 remote httpd: id="0299" srcip="91.17.50.163" localip="172.20.96.1" size="375" user="-" host="91.17.50.163" method="GET" statuscode="404" reason="-" extra="-" exceptions="-" time="2975" url="/lool/cloud.server.com/.../262_ocukbswiqfwn server="office.server.com" port="443" query="" referer="-" cookie="-" set-cookie="-" uid="WVFwaawUYAEAAGXQptAAAAAJ"

Is it possible that the utm has a problem with the redirection and the 2 different domain names? office.server.com + cloud.server.com

Any other idea?

Regards

André

Yes, it seems plausible that a url-within-a-url could cause UTM WAF to get confused.   Suggest you open a case with Sophos Support, if you have not done so already, as it is the only way to get bugs documented and fixed.

yes. seem to be a bug. i opend a support ticket and refered to this thread.

Case 8145469

André and Stefan, what happens if you select 'Pass host header' in the Virtual Server?

Cheers - Bob

Hey Bob,

in my configuration the option "Pass host header" has always been checked. Last time i tested is more than 8 weeks ago and my VM with nextcloud is not running at the moment. I think the certificates are already expired.

Regards André

Andre Winkler said:

Hey Bob,

in my configuration the option "Pass host header" has always been checked. Last time i tested is more than 8 weeks ago and my VM with nextcloud is not running at the moment. I think the certificates are already expired.

Regards André

 

Same here. But i tested with this option and wihout. no change. but without this option i don't see the taskbar from Collabora. So the option should be checked. that is important.

Hi,

In case you're talking about the error "AH00026: found %2f (encoded '/') in ... returning 404": That's because Apache by default won't allow this. The solution is to add "AllowEncodedSlashes On" within the vhost config of reverseproxy.conf ; However, this can only done manually. But as soon as you make a change in the GUI, it's overwritten.

Sophos had this feature implemented (according to support) in NUTM-2373, but the removed it again because it was leading to "other issues". The manual change was provided as workaround.

Because I can't ask my customers to call me everytime they make a change in the GUI so I can update the reverseproxy.conf, and I wanted to create a way so my customers can "add" it them self, I've updated the script which (re)starts the reverseproxy service:

=== START PATCH
<M> fw:/var/mdw/scripts # diff -Nura reverseproxy.orig reverseproxy
--- reverseproxy    2018-07-06 11:38:01.464266168 +0200
+++ reverseproxy.new    2018-07-06 13:47:37.722784507 +0200
@@ -35,7 +35,20 @@
     ${APACHE2CTL_LOG}
 }
 
+apache_allowslashes() {
+    grep -q 'PATH "/AllowEncodedSlashes/"' ${CHROOT}/usr/apache/conf/reverseproxy.conf
+    MYRESULT=$?
+
+    if [ ${MYRESULT} -eq 0 ]; then
+        echo "AllowEncodedSlashes found - fixing" | log
+        sed -i 's|WAFExceptions PATH "/AllowEncodedSlashes/" SkipAntiVirus|AllowEncodedSlashes On|g' ${CHROOT}/usr/apache/conf/reverseproxy.conf
+    else
+        echo "AllowEncodedSlashes not found - skipping" | log
+    fi
+}
+
 apache_conftest() {
+    apache_allowslashes
     chroot ${CHROOT} ${APACHE2CTL} configtest 2>&1 | log
     if [ $? -ne 0 ]; then
         return 1

=== END PATCH

This way, the only thing my customer has to do when I wants 'AllowEncodedSlashes On' enabled for a virtual server, is create an exception under "Webserver Protection" > Firewall Policies > [Exceptions]. The param of the exception needs to be excluding anti-virus for the path /AllowEncodedSlashes/

That's all! Works like a charm. Only thing to keep in mind is when you update the appliance, you need to check if the start script has been changed and if so, re-apply the patch. And I added the hook right before the actual config test, so when (in any strange way) my hook renders the config invalid, it keeps working because the reload/restart won't be executed.

I hope this is of some help to anyone.

Cheers,
Onno.

Hoi Onno and welcome to the UTM Community!

Thanks for joining us - it's very rare that a first post here is a solution to a problem.

The OP was running in Docker.  Is your solution one that will allow WAF to work well with apps in Docker?

Cheers - Bob

Hi there,

no. Doesn't work. or i did something wrong.

Here are some screenshots:

Hi Stefan,

The image of the Exception is not legible.  Instead of the screencap of the code, please copy and paste the text.

Cheers - Bob

Hi Stefan,

Hmm interesting. Everything looks OK. What you could double check, in the reverseproxy.conf file, is the specific vhost and if it contains the line: WAFExceptions PATH "/AllowEncodedSlashes/" SkipAntiVirus

The virtual server of my customer has also enabled the RewriteHTML and RewriteCookies, but I'm not sure if that makes any difference.
Also, the init script logs to reverseproxy.log: when you grep on "AllowEncodedSlashes" you should at least see something like "AllowEncodedSlashes found - fixing" or "AllowEncodedSlashes not found - skipping". If not, it looks like the hook is not being triggered... somehow ..

KR,
Onno.