diff options
Diffstat (limited to 'docs/configuration/firewall/index.rst')
-rw-r--r-- | docs/configuration/firewall/index.rst | 258 |
1 files changed, 258 insertions, 0 deletions
diff --git a/docs/configuration/firewall/index.rst b/docs/configuration/firewall/index.rst new file mode 100644 index 00000000..ed255916 --- /dev/null +++ b/docs/configuration/firewall/index.rst @@ -0,0 +1,258 @@ +.. _firewall: + +Firewall +======== + +VyOS makes use of Linux `netfilter <http://netfilter.org/>`_ for packet filtering. + +The firewall supports the creation of groups for ports, addresses, and networks +(implemented using netfilter ipset) and the option of interface or zone based +firewall policy. + +**Important note on usage of terms:** The firewall makes use of the terms +`in`, `out`, and `local` for firewall policy. Users experienced with netfilter +often confuse `in` to be a reference to the `INPUT` chain, and `out` the +`OUTPUT` chain from netfilter. This is not the case. These instead indicate the +use of the `FORWARD` chain and either the input or output interface. The +`INPUT` chain, which is used for local traffic to the OS, is a reference to +as `local` with respect to its input interface. + +Groups +------ + +Firewall groups represent collections of IP addresses, networks, or ports. Once +created, a group can be referenced by firewall rules as either a source or +destination. Members can be added or removed from a group without changes to +or the need to reload individual firewall rules. + +.. note:: Groups can also be referenced by NAT configuration. + +While **network groups** accept IP networks in CIDR notation, specific IP addresses +can be added as a 32-bit prefix. If you foresee the need to add a mix of +addresses and networks, the network group is recommended. + +Here is an example of a network group for the IP networks that make up the +internal network: + +.. code-block:: none + + set firewall group network-group NET-INSIDE network 192.168.0.0/24 + set firewall group network-group NET-INSIDE network 192.168.1.0/24 + +Groups need to have unique names. Even though some contain IPv4 addresses and others contain IPv6 addresses, they still need to have unique names, so you may want to append "-v4" or "-v6" to your group names. + +.. code-block:: none + + set firewall group network-group NET-INSIDE-v4 network 192.168.1.0/24 + set firewall group ipv6-network-group NET-INSIDE-v6 network 2001:db8::/64 + + +A **port group** represents only port numbers, not the protocol. Port groups can +be referenced for either TCP or UDP. It is recommended that TCP and UDP groups +are created separately to avoid accidentally filtering unnecessary ports. +Ranges of ports can be specified by using `-`. + +Here is an example of a port group a server: + +.. code-block:: none + + set firewall group port-group PORT-TCP-SERVER1 port 80 + set firewall group port-group PORT-TCP-SERVER1 port 443 + set firewall group port-group PORT-TCP-SERVER1 port 5000-5010 + +Rule-Sets +--------- + +A rule-set is a named collection of firewall rules that can be applied to an +interface or zone. Each rule is numbered, has an action to apply if the rule +is matched, and the ability to specify the criteria to match. + +Example of a rule-set to filter traffic to the internal network: + +.. code-block:: none + + set firewall name INSIDE-OUT default-action drop + set firewall name INSIDE-OUT rule 1010 action accept + set firewall name INSIDE-OUT rule 1010 state established enable + set firewall name INSIDE-OUT rule 1010 state related enable + set firewall name INSIDE-OUT rule 1020 action drop + set firewall name INSIDE-OUT rule 1020 state invalid enable + +Applying a Rule-Set to an Interface +----------------------------------- + +Once a rule-set is created, it can be applied to an interface. + +.. note:: Only one rule-set can be applied to each interface for `in`, `out`, + or `local` traffic for each protocol (IPv4 and IPv6). + +.. code-block:: none + + set interfaces ethernet eth1 firewall out name INSIDE-OUT + +.. _routing-mss-clamp: + +TCP-MSS Clamping +---------------- + +As Internet wide PMTU discovery rarely works, we sometimes need to clamp +our TCP MSS value to a specific value. This is a field in the TCP +Options part of a SYN packet. By setting the MSS value, you are telling +the remote side unequivocally 'do not try to send me packets bigger than +this value'. + +Starting with VyOS 1.2 there is a firewall option to clamp your TCP MSS +value for IPv4 and IPv6. + + +.. note:: MSS value = MTU - 20 (IP header) - 20 (TCP header), resulting + in 1452 bytes on a 1492 byte MTU. + + +IPv4 +^^^^ + +.. cfgcmd:: set firewall options interface <interface> adjust-mss <number-of-bytes> + + Use this command to set the maximum segment size for IPv4 transit + packets on a specific interface (500-1460 bytes). + + +Example +""""""" + +Clamp outgoing MSS value in a TCP SYN packet to `1452` for `pppoe0` and +`1372` +for your WireGuard `wg02` tunnel. + +.. code-block:: none + + set firewall options interface pppoe0 adjust-mss '1452' + set firewall options interface wg02 adjust-mss '1372' + +IPv6 +^^^^^ + +.. cfgcmd:: set firewall options interface <interface> adjust-mss6 <number-of-bytes> + + Use this command to set the maximum segment size for IPv6 transit + packets on a specific interface (1280-1492 bytes). + + +Example +""""""" + +Clamp outgoing MSS value in a TCP SYN packet to `1280` for both `pppoe0` and +`wg02` interface. + +.. code-block:: none + + set firewall options interface pppoe0 adjust-mss6 '1280' + set firewall options interface wg02 adjust-mss6 '1280' + + + +.. hint:: When doing your byte calculations, you might find useful this + `Visual packet size calculator <https://baturin.org/tools/encapcalc/>`_. + + +Applying a Rule-Set to a Zone +----------------------------- + +A named rule-set can also be applied to a zone relationship (note, zones must +first be created): + +.. code-block:: none + + set zone-policy zone INSIDE from OUTSIDE firewall name INSIDE-OUT + +How VyOS replies when being pinged +---------------------------------- + +By default, when VyOS receives an ICMP echo request packet destined for itself, it will answer with an ICMP echo reply, unless you avoid it through its firewall. + +With the firewall you can set rules to accept, drop or reject ICMP in, out or local traffic. You can also use the general **firewall all-ping** command. This command affects only to LOCAL (packets destined for your VyOS system), not to IN or OUT traffic. + +.. note:: **firewall all-ping** affects only to LOCAL and it always behaves in the most restrictive way + +.. code-block:: none + + set firewall all-ping enable + +When the command above is set, VyOS will answer every ICMP echo request addressed to itself, but that will only happen if no other rule is applied droping or rejecting local echo requests. In case of conflict, VyOS will not answer ICMP echo requests. + +.. code-block:: none + + set firewall all-ping disable + +When the comand above is set, VyOS will answer no ICMP echo request addressed to itself at all, no matter where it comes from or whether more specific rules are being applied to accept them. + +Example Partial Config +---------------------- + +.. code-block:: none + + firewall { + all-ping enable + broadcast-ping disable + config-trap disable + group { + network-group BAD-NETWORKS { + network 198.51.100.0/24 + network 203.0.113.0/24 + } + network-group GOOD-NETWORKS { + network 192.0.2.0/24 + } + port-group BAD-PORTS { + port 65535 + } + } + name FROM-INTERNET { + default-action accept + description "From the Internet" + rule 10 { + action accept + description "Authorized Networks" + protocol all + source { + group { + network-group GOOD-NETWORKS + } + } + } + rule 11 { + action drop + description "Bad Networks" + protocol all + source { + group { + network-group BAD-NETWORKS + } + } + } + rule 30 { + action drop + description "BAD PORTS" + destination { + group { + port-group BAD-PORTS + } + } + log enable + protocol all + } + } + } + interfaces { + ethernet eth1 { + address dhcp + description OUTSIDE + duplex auto + firewall { + in { + name FROM-INTERNET + } + } + } + } |