This is a quick guide to
and then setting up some common configurations:
This should cover everything you need to set up
More complex requirements are covered elsewhere:
However, please read this quick start section first, before tackling the others.
There are two easy ways to install FreeS/WAN:
If you are using one of them, just include FreeS/WAN in the choices you make during installation, or add it to your configuration later using the distribution's tools.
If your FreeS/WAN came on your machine, and is pre-1.98, you will need to hand generate an RSA key pair for authentication, using these instructions.
Sources for RPM packages of FreeS/WAN are:
Note that:
You need to download at least two RPMs:
You do not need the kernel header and kernel source RPMs to install FreeS/WAN, but Murphy's Law suggests you should download them so that the kernel source and headers on your system match the kernel actually in use.
Once you have them, install the RPMs with rpm -i commands. You will need to be root to install a module or kernel.
Then, start FreeS/WAN:
service ipsec start
If your distribution does not include FreeS/WAN and no RPMs are available, see our installation from source document.
To check that you have a successful install, run:
ipsec verify
or, on older FreeS/WANs:
ipsec whack --status
ipsec verify(8) checks whether:
It also anticipates future problems. For example, it tests if:
If your checks fail, see our troubleshooting guide.
That's it. FreeS/WAN is installed.
If you are running a firewall, you must now allow FreeS/WAN traffic through. You will also wish to consider how to protect your machine from unwanted traffic which might enter through your FreeS/WAN tunnels. There are three steps to firewalling with FreeS/WAN.
The first step is to allow IPsec packets in and out of your machine. Allow:
If that machine is a gateway, be sure that packets emerging from IPSec processing are correctly forwarded.
A second firewalling step -- access controls built into FreeS/WAN -- is automatically applied. For example, in some situations KLIPS determines that it should not have received a certain packet, and drops it.
Optionally, you may wish to add a third step, to filter packets emerging from your IPSec tunnels. This is a sensible precaution at any time, but becomes more important if you employ full opportunistic encryption, since that can allow strangers to bypass your usual firewall rules.
More detail is here , along with sample firewall scripts.
FreeS/WAN relies on two configuration files:
Note that Mandrake puts them in /etc/freeswan.
The remainder of this document shows you how to configure these common setups:
Opportunistic encryption makes many aspects of the setup and administration of IPsec easier.
For opportunistic encryption, you do not need to communicate with the administrator of a site before establishing secure communications to that site. In particular, you do not have to send them your keys or collect and authenticate theirs. You also do not have to configure IP information for each connection.
Instead, you need to place some information in DNS, and create a simple configuration that will enable your system to connect securely with every other system set up for OE. This presents a clear advantage over manually setting up a large number of VPN connections.
A major goal of the FreeS/WAN project is to get opportunistic encryption widely enough deployed that a "FAX effect" comes into play. Neither a FAX machine nor opportunistic encryption is of much value if there are only a few installed, but both become much more useful as the installed base increases.
Widespread deployment of opportunistic encryption appears to be our best hope for making the Internet more secure. See discussion in our introduction.
Opportunistic encryption relies on DNS:
To set up full opportunistic encryption, you must be able to insert resource records in the DNS reverse maps for:
For a standalone machine or a gateway using NAT, these two IPs will be identical, so you will only need one RR.
This requires a static IP, at least on the machine running IPSec.
Normally the reverse map is controlled by your Internet Service Provider (ISP). Ask your ISP's staff if they are willing to publish a resource record, which you would provide, in your IP's reverse map.
If you cannot, you can still protect connections which you initiate. This requires simply that you can put a record in some domain's forward DNS.
Directions for initiate-only opportunism are just below; those for full opportunism are in the following section .
Because opportunistic encryption relies on DNS:
its authentication is only as strong as your DNS is secure .
Without secure DNS, opportunistic connections protect against passive snooping, but not active man-in-the-middle attacks.
You can make your DNS entries more trustworthy by serving them within your security perimeter. We recommend running a nameserver on your gateway, as described in our opportunism HOWTO ("Getting DNS Through").
Opportunistic encryption is new technology and we are still working out some fine points. Please see this list of known issues.
In the next sections, we will tackle three typical setups for opportunistic encryption:
One example does build on the previous one, but you can skip ahead to get an idea of what your situation entails.
In this section, we the case of opportunistic encryption that has the fewest requirements:
This would apply to a standalone machine, or to a home gateway with some invisible NAT clients.
Given the above conditions, you can set up opportunistic encryption without having to add resource records to the DNS reverse map for your public IP address. Sections after this one cover situations where one or more of the above restrictions do not apply.
There are two steps:
Once this is done, your system will automatically encrypt whenever it can.
You need to put your system's RSA public key in a forward DNS record so that systems you communicate with can find it.
Dynamic IP users take note: the domain where you place your key need not be associated with the IP address for your system, or even with your system's usual hostname.
When negotiating a connection, your FreeS/WAN will identify itself by a fully qualified domain name (FQDN) which tells other FreeS/WANs where to find its KEY. Choose a name, within the domain you have access to, that you will use for this purpose.
For example, if you have access to example.com's DNS, you could choose xy.example.com, even if your hostname is xyz.instance.com and the reverse map for your IP points to 12345678-ourtowns-cableco.com.
Note that xy.example.com need not be associated with your current IP address, or any IP for that matter.
You can generate a DNS KEY record containing your system's public key with the command:
ipsec showhostkeyThe result should look like this (with the key data trimmed down for clarity):
; RSA 2048 bits xy.example.com Sat Apr 15 13:53:22 2000 xy.example.com. IN KEY 0x4200 4 1 AQOF8tZ2...+buFuFn/
Change xy.example.com to your FQDN.
Insert the record into DNS, or have a helpful system adminstrator do so for you. Remember that it will take time (up to 48 hours, but often a lot less) for any new DNS information to propagate, so OE won't work immediately.
By default, Linux FreeS/WAN ships with an opportunistic connection, conn me-to-anyone. It's easy to customize this.
Add a line to me-to-anyone which sets the parameter leftid to your FQDN, preceded by an @ sign. In our example:
conn me-to-anyone
...
leftid=@xy.example.com
...
The @ causes FreeS/WAN to identify itself by your FQDN, rather than by an IP which corresponds to that name.
Note that the left and right designations in ipsec.conf are arbitrary. We follow a convention of using left for local and right for remote.
Uncomment the line auto=route to enable the connection.
Normally, this is all the configuration you will need to do. However, if your FreeS/WAN is protecting an interface other than the one on your machine's default route, you will need to adjust the interfaces= line in config setup.
A full ipsec.conf(5) file for this setup is available here.
The quick method is to point a browser to oetest.freeswan.org. A link there will tell you whether or not you have an encrypted connection.
For more detail, take these steps:
You should see a tunnel to the opportunistic host. It will look something like:
1.2.3.4/32 -> 5.6.7.8/32 => tun0x146c@5.6.7.8
When FreeS/WAN cannot set up an opportunistic connection, and no explicit tunnel has been configured, its default is to allow the traffic through in the clear. For the non-opportunistic host, you should see a %pass eroute (IPsec route), the FreeS/WAN mechanism that implements that default. This looks something like:
1.2.3.4/32 -> 9.10.11.12/32 => %pass
To enable full opportunism on your standalone system (or NATting gateway), you need to do a bit more. This will allow others to initiate encrypted connections to your system -- for example, if you run services on your machine and want remote clients to be able to access them securely.
There are two steps in the setup.
Both need to be a little different than in the initiate-only case.
Most people can use the conn me-to-anyone exactly as it ships in the default ipsec.conf(5) file. Just uncomment auto=route to enable it.
Unlike the configuration above, do not set leftid=. By default, FreeS/WAN will use the IP address of your public interface instead of a name as your identity. Once again, if FreeS/WAN is protecting a different interface, you will need to adjust interfaces= in the config setup section.
You can refer to this sample configuration.
FreeS/WAN must also use the secret key from ipsec.secrets(5) that matches this IP address. Normally you have only one RSA key, so there is no need to alter ipsec.secrets.
You need to put two records, a KEY record and a TXT record, in the DNS reverse map for your public IP. They need to be here because the initiating FreeS/WAN must look up your data knowing only your machine's IP address, not its name. As mentioned above, this requires your ISP's co-operation.
The KEY record you need looks like this:
; RSA 2048 bits xy.example.com Sat Apr 15 13:53:22 2000 4.3.2.1.in-addr.arpa. IN KEY 0x4200 4 1 AQOF8tZ2...+buFuFn/
Generate it with
You also need to create a TXT record, to let others know that this machine can receive opportunistic connections. It also lets them know that the machine is authorized to encrypt on its own behalf.
Use the command:
ipsec showhostkey --txt 1.2.3.4
where you replace 1.2.3.4 with your public IP.
The record (with key shortened) looks like this:
; RSA 2048 bits xy.example.com Sat Apr 15 13:53:22 2000
IN TXT "X-IPsec-Server(10)=1.2.3.4 AQOF8tZ2...+buFuFn/"
Send both records to your ISP, to be published in your IP's reverse map.
There is a particular security concern when you allow incoming opportunism.
Incoming opportunistic packets enter your machine via an IPSec tunnel. That is, they all appear as ESP (protocol 50) packets, concealing whatever port and protocol characteristics the packet within the tunnel has. Contained in the tunnel as they pass through ppp0 or eth0, these packets can bypass your usual firewall rules on these interfaces.
Since you will be exchanging opportunistic packets with peers who are not familiar to you, you will want to firewall your ipsec interfaces the way you would any publicly accessible interface.
A simple way to do this is to create one iptables(8) table with all your filtering rules for incoming packets, and apply the entire table to all public interfaces, including ipsec interfaces.
Here's more on firewalling with opportunistic encryption.Next we expand from a standalone system (which protects only its own traffic) to a gateway (which protects traffic for other systems).
There is one special case in which gateway configuration is quite simple -- if all the machines behind the gateway are hidden from the Internet. We describe that first, then go on to describe gateways for visible clients.
If your gateway uses NAT to allow machines to access the Internet without having their own routable IP addresses, then from the point of view of anyone else on the Internet:
For purposes of IPsec across the Internet, your gateway can be treated as a standalone machine. Consequently,
For a more detailed discussion of NAT, see our background section.
Many gateways will need to support client systems that have routable addresses and are thus visible to the Internet. To equip such a gateway with opportunistic encryption, you'll need a co-operative ISP that will allow you to insert resource records into reverse DNS.
Setup for a gateway involves:
You need only make a few additions to in the ipsec.conf(5) file:
In ipsec.conf(5), a new conn is needed for each local subnet to be protected by opportunistic encryption. If you are also protecting the gateway, most of the new conn can be borrowed from the old one. Place this before me-to-anyone:
conn subnet-to-anyone
also=me-to-anyone
leftsubnet=42.42.42.0/24
Note that a subnet described in ipsec.conf(5) need not correspond to a physical network segment. This is discussed in more detail in our advanced configuration document.
If required, a gateway can easily provide this service for more than one subnet. You just add a connection description for each.
We assume you already have a KEY record in the reverse map for your gateway's public IP address, so your gateway can accept incoming connections. This is described above.
For the gateway to provide an opportunistic encryption service for other systems, it must be possible for the initiator of an IPsec connection to:
This is done by adding a TXT record to the reverse map for the endpoint. The record (with key shortened) looks like this:
; RSA 2048 bits gateway.example.com Sat Apr 15 13:53:22 2000
IN TXT "X-IPsec-Server(10)=1.2.3.4 AQOF8tZ2...+buFuFn/"
This record must be generated on the gateway so it can get the key from ipsec.secrets(5). The command is:
ipsec showhostkey --txt 1.2.3.4
You must supply the gateway IP address on the command line.
One of these records is required in the reverse map for each system using this gateway for opportunistic IPsec. You insert it in the reverse map part of the zone file right after the line for that system's IP address, so part of the file might look like this:
1.42.42.42.in-addr.arpa. IN PTR arthur.example.com
; RSA 2048 bits gateway.example.com Sat Apr 15 13:53:22 2000
IN TXT "X-IPsec-Server(10)=1.2.3.4 AQOF8tZ2...+buFuFn/"
2.42.42.42.in-addr.arpa. IN PTR ford.example.com
; RSA 2048 bits gateway.example.com Sat Apr 15 13:53:22 2000
IN TXT "X-IPsec-Server(10)=1.2.3.4 AQOF8tZ2...+buFuFn/"
3.42.42.42.in-addr.arpa. IN PTR trillian.example.com
; RSA 2048 bits gateway.example.com Sat Apr 15 13:53:22 2000
IN TXT "X-IPsec-Server(10)=1.2.3.4 AQOF8tZ2...+buFuFn/"
You need one TXT record per client, but the TXT records can all be identical.
Some circumstances call for pre-configured IPSec connections, also known as VPNs (Virtual Private Networks). In particular, when you need strong authentication, ie. to access an office network, VPN is your best current choice. This may change when secure DNS arrives, since secure DNS will strengthen the authentication system for opportunistic encryption.
Below are instructions for pre-configuring:
as well as tips on:
A common requirement is for pre-configured connections between a specific network and some set of remote machines. For example, an office network will often need to provide remote access services for:
We refer to the remote machines as "road warriors". For purposes of IPsec, anyone with a dynamic IP address is a road warrior.
Of course, if both the warrior and the gateway at the office are set up for opportunistic encryption, then you may not need the pre-configured connection. Here we assume that you do need it. For example:
This section has three sub-sections:
On either end, the opportunistic setup is unaffected by this. You leave it in place so both systems can continue to do opportunistic encryption with everyone but each other.
To set up an explicitly configured connection, you need some information about the system on the other end.
Connection descriptions use left and right to designate the two ends. We adopt the convention that, from the gateway's point of view left=local and right =remote.
The gateway administrator needs to know some things about each road warrior:
To get this information, in a format suitable for insertion directly into the gateway's ipsec.conf(5) file, issue this command on the warrior machine:
ipsec showhostkey --right
The output should look like this (with the key shortened for easy reading):
rightid=@xy.example.com
rightrsasigkey=0s1LgR7/oUM...
The road warrior needs to know:
which can be generated by running ipsec showhostkey --left on the gateway. Each warrior must also know:
This information should be provided in a convenient format, ready for insertion in the warrior's ipsec.conf(5) file. For example:
left=1.2.3.4
leftsubnet=42.42.42.0/24
leftid=@gateway.example.com
leftrsasigkey=0s1LgR7/oUM...
The gateway administrator typically needs to generate this only once. The same file can be given to all warriors.
Of course it is also possible to provide different versions, with access to different subnets, to different groups of warriors. See our advanced configuration document.
To set up a road warrior machine, we start from the opportunistic initiator setup shown above.
Simply add a connection description us-to-office, with the left and right information you gathered above. This might look like:
# pre-configured link to office network
conn us-to-office
# information obtained from office system admin
left=1.2.3.4 # gateway IP address
leftsubnet=42.42.42.0/24 # the office network
leftid=@gateway.example.com
# real keys are much longer than shown here
leftrsasigkey=0s1LgR7/oUM...
# our stuff, same when we are opportunistic initiator
right=%defaultroute
rightid=@xy.example.com
Here's more detail on this configuration.
We could easily add more connections as required, perhaps one each for his office, her office, the kid's school,... The file would grow longer, but nothing already in the file would need to change.
Once you have a number of connections, you may like to make your ipsec.conf modular, to save typing. See these instructions.
Adding road warrior support so people can connect remotely to your office network is straightforward.
We start from the opportunistic gateway setup shown above.
You could put a complete connection description for each warrior in your ipsec.conf(5) file, but this makes for a rather unmanageable file if you have many warriors.
Instead, we suggest you give each warrior its own file, choosing some directory and naming convention that suits your system and style.
For this example, we use the directory /etc/ipsec.road and use filenames based on IPsec ID, so the warrior using ID xy.example.com gets a file named xy.conf.
Using such files, you need add only one line to ipsec.conf(5). With our naming convention, the line is:
include /etc/ipsec.road/*.conf
FreeS/WAN will then read all those files and behave as if they were part of the ipsec.conf(5) file.
We suggest you make your ipsec.conf modular, so that each file does not need to include information for the gateway side. Create a separate connection conn gate_stuff , containing this common information. For example:
conn gate_stuff
left=1.2.3.4
leftsubnet=42.42.42.0/24
leftid=@gateway.example.com
leftrsasigkey=0s1LgR7/oUM...
You will reference this in each road warrior connection.
Your include line needs to be before conn gate_stuff . A convenient place for the line is right after the conn %default section.
Each of the road warrior files then contains a connection description for that warrior. For example:
# connection description for road warrior "xy"
conn gate-xy
# use the gateway description in ipsec.conf(5)
also=gate_stuff
# allow connection attempt from any address
# attempt fails if caller cannot authenticate
right=%any
# authentication information
rightid=@xy.example.com
rightrsasigkey=0s1LgR7/oUM...
With this technique, it becomes fairly simple to administer a gateway that supports many road warriors. For example:
To add a new user, simply add a suitable file.
To disable an account -- for example if a key is compromised -- take any existing connection down and delete it from Pluto's internal database with:
ipsec auto --delete connection
and remove or correct the affected file.
If you have many users, it would be worthwhile to write scripts to automate such tasks.
Often it is useful to have explicitly configured IPsec tunnels between different offices of an organisation, or between organisations that have joint projects.
Of course, if both offices are set up for opportunistic encryption and the security policies in place allow you to use that, explicitly configured tunnels become unnecessary. However, this will not always be the case.
Adding up a network-to-network tunnel does not require any change to the opportunistic or road warrior parts of your ipsec.conf(5). You can keep those parts exactly as shown above.
Of course, a network-to-network tunnel requires its own connection description, so you have to add that. There are two ways to do this.
Choose whichever is more convenient to administer in your environment.
Here is a network-to-network tunnel description from our examples file:
# sample tunnel
# The network here looks like:
# leftsubnet====left----leftnexthop......rightnexthop----right====rightsubnet
# If left and right are on the same Ethernet, omit leftnexthop and rightnexthop.
conn sample
# left security gateway (public-network address)
left=10.0.0.1
# next hop to reach right
leftnexthop=10.44.55.66
# subnet behind left (omit if there is no subnet)
leftsubnet=172.16.0.0/24
# right s.g., subnet behind it, and next hop to reach left
right=10.12.12.1
rightnexthop=10.88.77.66
rightsubnet=192.168.0.0/24
auto=start
If you give an explicit IP address for left (and left and right are not directly connected), then you must specify leftnexthop (the router which left sends packets to in order to get them delivered to right). Similarly, you may need to specify rightnexthop (vice versa).
The *nexthop parameters are needed because of an unfortunate interaction between FreeS/WAN and the kernel routing code. They will be eliminated in a future release, but perhaps not soon. We know they should go, but getting them out is not a simple problem.
This description can be generated on either machine and simply inserted in the ipsec.conf(5) file on the other. No change is required or desired.
Provided both machines do IPsec over the interface that is their default route to the Internet (a common case, but by no means the only one) you can simplify the description somewhat.
When using left=%defaultroute, you do not need to specify leftnexthop. left does not need to know rightnexthop either, so on left the connection description can be:
conn sample
# left security gateway (public-network address)
left=%defaultroute
# subnet behind left (omit if there is no subnet)
leftsubnet=172.16.0.0/24
# right s.g., subnet behind it
right=10.12.12.1
rightsubnet=192.168.0.0/24
auto=start
On right it is:
conn sample
# left security gateway (public-network address)
left=10.0.0.1
# subnet behind left (omit if there is no subnet)
leftsubnet=172.16.0.0/24
# right s.g., subnet behind it
right=%defaultroute
rightsubnet=192.168.0.0/24
auto=start
At this point, we have covered setup for opportunistic encryption and for simple cases of road warrior and VPN connections. You have several choices for what to look at next: