Tuesday, 3 September 2013

Check Point - Clear Identity Awareness user to IP mappings

It doesn't appear that there's an easy way on the Check Point CLI to remove all user to IP address mappings. You can revoke a single IP at a time - but for troubleshooting you might want to wipe out the whole lot.

Sounds like a job for a dirty bash one-liner!

(From expert mode of course)

 pep show user all | egrep -o '[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}' | grep -v '' | xargs -i -p pdp revoke_ip {}  

Friday, 2 August 2013

Check Point - Linux Remote Access VPN with Shrew VPN client

A few years back I tried to get any sort of VPN client working on Ubuntu that would connect to a Check Point firewall. SSL Network Extender (SNX) works but requires additional configuration on the gateway, so I gave up.

Between then and now, Shrew Soft VPN client has added support for Check Point firewalls and works pretty well. I found the client crashed intermittently when setting up the profiles but after that the tunnel seemed stable. I've tested it on Ubuntu 12 and 13 (on 13 you have to compile from source as it's not in the Apt repositories yet) and both work ok.

The client is pretty straightforward to setup - once you know which options to use of course!

Hostname / IP address - IP or DNS name for your firewall
Auto Configuration - Leave this at 'ike config pull'.
Local Host - If you're using office mode with DHCP addresses, this will take care of picking up the address once the tunnel is up.

Client and name resolution tabs - leave these settings as default.
Authentication - assuming you haven't changed any of your remote access settings, Hybrid RSA + XAuth is what you need.
Authentication method:
Local Identity - User Fully Qualified Domain Name (make sure you leave the value blank however)
Remote Identity - Any
Credentials -> Server Certificate Authority File - This needs to be the Check Point internal CA certificate that issues the VPN certificate for your gateway. Setting it to 'any' doesn't appear to work.


Phase 2

Policy - Leave policy generation set to 'auto' and untick the following two boxes.
Then, make sure you add in all of the remote networks you want to access over the tunnel and the software will add in the correct routes for your remote resources. 

Then you're all set to connect!

Friday, 21 June 2013

Fortinet PPTP VPN with LDAP authentication

With a somewhat fundamental documentation failure, it looks like FortiGate PPTP VPN do not support CHAP/MSCHAPv2 when you are authenticating your user groups via LDAP. According to Fortinet - this is something they're aware of and works using PAP. When I spoke with them this morning, I advised this is unacceptable as a workaround because it disables encryption! I'll update this when I get a real solution.

Update: Official answer, it's not supported. http://kb.fortinet.com/kb/microsites/search.do?cmd=displayKC&docType=kc&externalId=10718&sliceId=1&docTypeID=DT_KCARTICLE_1_1&dialogID=51071690&stateId=0%200%2051073253 . I guess the solution is to use FortiClient (or Check Point...)

Wednesday, 22 May 2013

Blue Coat Troubleshooting Basics

A few useful 'set piece' procedures for gathering diagnostic information from a Blue Coat Proxy SG appliance.

Full policy trace

Enable this under Configuration > Policy > Policy options > enable full policy execution. Then, go to https://x.x.x.x:8082/policy (where x.x.x.x is the IP of your proxy) and delete the default trace. Also make sure there are no tracing rules in CPL or the VPM otherwise the trace will not work. Then, replicate the problem by going to the problematic website a few times (making a note of the IP address the client is using) and then turn off the policy trace.

Filtered policy trace

Add the following to your local policy file (replacing z.z.z.z with the IP address of the client you are testing from).
client.address=z.z.z.z trace.rules(all) trace.request(yes) trace.destination(trace.html)
Then, replicate the issue.
Then go to https://x.x.x.x:8082/policy (where x.x.x.x is the IP of your proxy) and download the file trace.html


Go to https://x.x.x.x:8082/sysinfo (replacing x.x.x.x with the IP address of your proxy) and save the page as a text file. Then please zip and email it to me.

Sysinfo stats

Go to https://x.x.x.x:8082/Diagnostics/Snapshot/sysinfo_stats/download/all to download an archive of all the available snapshots on the SG.

HTTP Debug:

The HTTP debug can be done using the advanced URL https://xx.xx.xx.xx:8082/HTTP/Debug. Set the mask enabling all options and clear the log just before browsing to the site both times. Saving the output from each.

Packet capture

Login to https://x.x.x.x:8082/PCAP/statistics and make sure filtering is off. Then, start the packet capture, try to access the problematic site through the proxy and then stop the capture. Then please download and compress the packet capture and send it to me with a note of the client IP address, proxy IP address and the IP address and URL of the website you tried to view.

Blue Coat - SSL reverse proxy for Exchange Active Sync (with client certificates)

I've recently had the pleasure of setting up a reverse SSL proxy for Exchange Active Sync (EAS) with the requirement of SSL on both sides of the connection, with client certificates for mobile devices.
After some coffee, some 'what have I done to deserve this' moments and a bit of reading - it's actually not quite as painful as you might imagine. There's a good amount of documentation on how to achieve this with Microsoft ISA / TMG but not much for other vendors. Here's how I set it up with a Blue Coat  Proxy SG running SGOS

1. To make things a bit easier, you'll need all of your certificates in PEM format (Base 64 encoded - the ones if you open in a text edit start with ===BEGIN CERTIFICATE/KEY===). If you generate them from your Microsoft enterprise CA you'll most likely be given them in PFX or P12 unless you want to generate CSRs from the Blue Coat device. You'll need to use openssl to convert them into a PEM file containing the unencrypted private key and the certificate.
You will need the certificates for your enterprise CA, exchange server if not issued by your CA and one for the Blue Coat device itself  to identify to the clients (issued by your enterprise CA).

2. Add the certificate for the Blue Coat device to your proxy as a keyring (configuration -> SSL -> Keyrings). This should be the only certificate with a private key contained in the PEM file.

3. Add the other CA certificates to the proxy in  CA certificates section (configuration -> SSL -> CA certificates). This will ensure the proxy trusts any certificates issued by your enterprise CA (clients and the Exchange server). It's quite common for Exchange to use a self-signed certificate - if this is the case you'll need to export this certificate and add this to the Blue Coat as well (otherwise you'll need to add policy later on to ignore the fact it's issuer isn't trusted.

4. Create a new CA Certificate list and add in the CA certificates you created above. We'll use this shortly on our proxy service listener.

5. Under Configuration -> Services -> Proxy Services, create a new service group and named it 'Reverse proxy'. (this is optional but keeps things more organised). Create a new and fill in the details as below.
Name: Something meaningful to you
Service group: Group you created above
Proxy: HTTPS Reverse Proxy
Keyring: This is the keyring for the Blue Coat appliance you created previously (the one with the private key).
CCL: This is the CCL you created in step 4. This will make sure the proxy only accepts certificates which were issued by your enterprise CA.
SSL versions: Enable whatever you require here - but make sure you tick Verify Client and don't enable Forward Client Cert (it's impossible to forward the actual client certificate so the Blue Coat forwards some information from the certificate in the HTTP headers. This doesn't work for our goal which is why we're using KCD).
ADN: unticked - it's not relevant here.
Listeners: This is the address that you will NAT from your firewall (or a public address if your proxy is deployed in that way). This will be where clients connect. You can use an IP that isn't in use by the proxy so long as you add it as a Virtual IP.

6. Join the proxy to your domain (you'll need a domain administrator account for this). Under Authentication -> Windows Domain create a new domain and then join it. Make a note of the name you join to the domain with as you'll need this for the KCD settings. If you have problems joining - make sure the DNS on your proxy is pointing toward your Active Directory server. Once joined, go to Authentication -> IWA -> IWA servers and make sure under Client Authentication Modes - 'Kerberos credentials' is ticked.

7. Create a certificate authentication realm under Authentication -> Certificate. Then under Certificate Main, configure the extractors so the proxy is able to obtain the username from the clients certificate.
Your certificates might vary so you might need to try a few options to find what works for you. Our AD / CA setup is mostly standard so far as I know so these settings should work. You'll get errors in the event and access logs saying 'unable to extract username' if these settings aren't relevant to your environment. The options on the other tabs can be left at default.

8. Over to your AD setup for a moment now to enable your Blue Coat device to perform KCD for authentication to your Exchange server. In the AD 'Users And Computers' view - find the new device representing the proxy (the name you chose in step 6)and edit the properties. Under the delgation tab, configure your options as in the screenshot.
For the user / computer options which are blanked out in the screenshot - this needs to be the object for your Exchange server. Click apply / OK and close this window.

9. Back to the proxy now. Go to configuration -> Forwarding -> Forwarding Hosts and create a new forwarding host. This needs to have the below settings.
Host: IP address of your Exchange server
Type: Server
Ports: 443
Others left unticked as per the default.

10. Open up the VPM via Configuration -> Policy -> Visual Policy Manager.
For the purpose of this guide - I'll be working with a totally blank policy to keep things simple. The policy layers should be setup as below.

Layer 1 - Web Authentication Layer.
Destination: Request URL object. Simple match - value should be the DNS name clients will be accessing for your reverse proxy. Eg, activesync.mydomain.co.uk
Action: Combination object containing an Authenticate Object (Realm should be your certificate authentication realm from step 7. Mode: Auto) and a Kerberos Constrained Delegation object. (Authentication type should be origin, IWA Realm should be the IWA direct realm from step 6. Service Principal Name should be 'host/' then the exact name of your Exchange server without the domain).

Layer 2 - Web Access Layer.
Source: Authenticated user (no configuration for this object)
Destination: Request URL object (Advanced match - scheme any, Path "/Microsoft-Server-ActiveSync"). This will ensure only EAS connections are allowed to this reverse proxy connection. 

Layer 3 - SSL Access Layer
NOTE: This layer isn't strictly necessary but it's useful to troubleshoot SSL problems whilst you're setting up. The destination is the IP of your Exchange server and the Action is an SSL certificate validation action to ignore untrusted issuer.

Layer 4 - Forwarding Layer
Source: Authenticated User (no options again)
Destination: Server URL object (simple match again with the DNS name of your reverse proxy connection).
Action: Select Forwarding Object. Move the object name you created in step 9 into the right hand column and then ok.

11. We're done! Install the VPM policy and then you're ready to test. Make sure the client certificates you deploy contain the full certificate chain or you'll get SSL errors when your devices try connecting to the proxy. If you use Active Directory to issue user certificates you'll probably find they're in PFX format. In this case, they should contain the full chain by default.

Friday, 8 February 2013

Check Point dynamic object for DNS

Although Check Point use domain objects for resolving DNS entries to IP addresses in the rulebase - the documentation recommends not to use them for various good reasons. I've written a simple script to make use of Check Points rarely used dynamic objects. The script pings a host by DNS, takes the IP address then updates a dynamic object you can then use in your firewall policy. Add this to a cronjob and then you can have a dynamic dynamic object(!?).
Download here

#!/bin/bash -f
#Dynamic Object update script - dyno.sh
#Stuart Green - 08/02/2013
# This script will create and subsequently update a Check Point dynamic
# object which can be used in the rulebase in place of a domain object which is
# known to be unreliable and resource intensive. The script should be placed in
# a crontab and run at the interval you choose.
# DNS needs to be configured on your gateway.
# ICMP and DNS need to be allowed outbound from your gateway / cluster object.
# Tested on R75.40 SecurePlatform and Gaia - should be fine with other versions.
# Copy this script somewhere on your firewall.
# Edit the value of CDIR in this script to be the value of $CPDIR from your 
# shell. Eg, the value you get when you run 'echo $CPDIR'.
# Edit the value of DNSNAME to be the domain you want to resolve to an IP.
# Edit the value of CPOBJECT to represent a UNIQUE Check Point dynamic object.
# Add a cronjob to execute the script at your desired time interval (~10min).
# run 'crontab -e' to edit your crontab
# paste in the following line (minus quotes) to run the script every 10 minutes: 
# '10 * * * * /home/admin/dyno.sh'
# Create a dynamic object in your policy with the EXACT same name as CPOBJECT.
# Use the object where required.
# If you need multiple dynamic objects, you can copy the script with a 
# different name, modify the values of DNSNAME and CPOBJECT then add the new
# script to a new cronjob.
# Domains that return multiple IPs will only update the dynamic object with the
# FIRST IP address!
source $CDIR/tmp/.CPprofile.sh
IP=$(/bin/ping -c1 $DNSNAME | /bin/egrep -oh -m1 "([0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3})")
$FWDIR/bin/dynamic_objects -do $CPOBJECT > /dev/null 2>&1
$FWDIR/bin/dynamic_objects -n $CPOBJECT > /dev/null 2>&1
$FWDIR/bin/dynamic_objects -o $CPOBJECT -r $IP $IP -a > /dev/null 2>&1
if [[ $EXIT_STATUS -eq 0 ]]; then
  echo "Dynamic object $CPOBJECT updated. $DNSNAME resolves to $IP" >> /var/log/messages
if [[ $EXIT_STATUS -ne 0 ]]; then
  echo "Dynamic object $CPOBJECT not updated successfully" >> /var/log/messages

Friday, 4 January 2013

Check Point - upgrade_export and the migrate command

What's the difference between the two you may ask. In true Check Point style the documentation implies they're different commands with different functions.

 Let's compare the binaries...

md5sum migrate
d9889562e3093d70669ff511b8fbc9d4  migrate

md5sum upgrade_export
d9889562e3093d70669ff511b8fbc9d4  upgrade_export

So the difference - nothing. They're exactly the same binary. The only apparent difference is you get a deprecation warning if you use upgrade_export but not migrate. Thanks for the misleading documentation Check Point...