Differences between revisions 6 and 50 (spanning 44 versions)
Revision 6 as of 2015-02-17 21:03:03
Size: 4229
Comment: typos
Revision 50 as of 2022-01-09 21:43:49
Size: 6348
Editor: donfede
Comment: add version number to Debian 10 Buster references
Deletions are marked like this. Additions are marked like this.
Line 1: Line 1:
#language en

~-[[DebianWiki/EditorGuide#translation|Translation(s)]]:
English - [[it/nftables|Italiano]] - [[ru/nftables|Русский]] -~

----
Line 3: Line 10:
Two of the most common uses of nftables is to provide firewall support and NAT. Two of the most common uses of nftables is to provide firewall support and Network Address Translation (NAT).
Line 5: Line 12:
nftables is meant to replace the [[iptables|iptables]] framework. nftables is the default and recommended firewalling framework in Debian, and it replaces the old [[iptables|iptables]] (and related) tools.

<<TableOfContents>>
Line 9: Line 18:
nftables is under heavy development. '''NOTE: Debian 10 Buster and later use the nftables framework by default'''.
Line 11: Line 20:
For a production firewall, you should keep using [[iptables]]. Starting with Debian 10 Buster, nf_tables is the default backend when using iptables, by means of the iptables-nft layer (i.e, using iptables syntax with the nf_tables kernel subsystem).
This also affects ip6tables, arptables and ebtables.
Line 13: Line 23:
However, you would like to start testing nftables: = Hints =
Line 15: Line 25:
 * the new syntax and engine
 * get in touch with new workflows
 * report bug and request features
Some hints folks might find interesting in some situations.
Line 19: Line 27:
Currently, lots of bugs-fixes and features are added in each new release. == Use firewalld ==
Line 21: Line 29:
'''Why nftables is not in Debian Jessie?''' You should consider using a wrapper instead of writing your own firewalling scripts. It is recommended to run [[firewalld]], which integrates pretty well into the system. See also https://firewalld.org/

The firewalld software takes control of all the firewalling setup in your system, so you don't have to know all the details of what is happening in the underground.
There are many other system components that can integrate with firewalld, like NetworkManager, libvirt, podman, fail2ban, docker, etc.

== Reverting to legacy xtables ==

You can switch back and forth between iptables-nft and iptables-legacy by means of update-alternatives (same applies to arptables and ebtables).

The default starting with Debian 10 Buster:
{{{
# update-alternatives --set iptables /usr/sbin/iptables-nft
# update-alternatives --set ip6tables /usr/sbin/ip6tables-nft
# update-alternatives --set arptables /usr/sbin/arptables-nft
# update-alternatives --set ebtables /usr/sbin/ebtables-nft
}}}

Switching to the legacy version:
{{{
# update-alternatives --set iptables /usr/sbin/iptables-legacy
# update-alternatives --set ip6tables /usr/sbin/ip6tables-legacy
# update-alternatives --set arptables /usr/sbin/arptables-legacy
# update-alternatives --set ebtables /usr/sbin/ebtables-legacy
}}}

== nftables in Debian the easy way ==

If you want to enable a default firewall in Debian, follow these steps:

{{{
# aptitude install nftables
# systemctl enable nftables.service
}}}

This way, nftables is active at boot. By default, rules are located in '''/etc/nftables.conf'''.

To stop nftables from doing anything, just drop all the rules:

{{{
# nft flush ruleset
}}}

To prevent nftables from starting at boot:

{{{
# systemctl mask nftables.service
}}}

To uninstall it and purge any traces of nftables in your system:

{{{
# aptitude purge nftables
}}}
Line 24: Line 84:
The short answer is: nftables is not a stable software, and it has nothing to do in a stable operating system.

Jessie was about to be released with nftables v0.3. Which is a very early release. Some important changes happened in v0.4, and in Linux kernel 3.18.

However, once Jessie is released the maintainer of nftables expect to keep jessie-backports as updated as possible :-)

= Requirements =

nftables requires a linux '''kernel >= 3.13''', but running a newer '''kernel >= 3.18''' is recommended.

Also, nftables requires libnftnl, a public library which provides a low level interface to the kernel subsystem.

Aptitude will take care of all dependencies.
Line 40: Line 87:
=== What is nftables? === == What is nftables? ==
Line 43: Line 90:
=== Why a new framework? === == Should I build a firewall using a nftables? ==

Yes. Building new firewalls on top of iptables is discouraged.

== Should I replace an iptables firewall with a nftables one? ==

Yes, nftables is the replacement for iptables. There are some tools in place to ease in this task.

Please read: https://wiki.nftables.org/wiki-nftables/index.php/Moving_from_iptables_to_nftables

== Why a new framework? ==
Line 46: Line 104:
=== Should I stop working with iptables to build a firewall? ===
No. Currently, nftables is in an early development state. Nowadays, iptables is more stable.
== What are the major differences? ==
Line 49: Line 106:
=== Then, Why should I use nftables? ===
You can start testing what is meant to become the future of firewalls on Linux. Report bugs, request features and get in touch with the latest in this technology.

=== I knew the iptables syntax. Is there a new syntax is nftables? ===
Yes, but the nftables is better :-)

=== What are the major differences? ===
Line 64: Line 114:
This new framework features a new linux kernel subsystem, known as nf_tables.
The new engine mechanism is inspired by BPF-like systems, with a set of basic expressions, which can be combined to build complex filtering rules.

== Should I mix nftables and iptables/ebtables/arptables rulesets? ==

No, unless you know what you are doing.

== I knew the iptables syntax. Is there a new syntax in nftables? ==

Yes, but the nftables one is better :-)

Help in migrating to nftables: https://wiki.nftables.org/wiki-nftables/index.php/Moving_from_iptables_to_nftables
Line 65: Line 128:
Create a basic IPv4 table:
Create a basic IPv4/IPv6 dual-stack table:
Line 67: Line 131:
# nft add table filter # nft add table inet filter
Line 70: Line 134:
Create a chain for input traffic IPv4: Create a chain for input IPv4/IPv6 dual-stack traffic:
Line 72: Line 136:
# nft add chain filter input { type filter hook input priority 0; } # nft add chain inet filter input { type filter hook input priority 0\; }
Line 75: Line 139:
A rule to check that all is fine (IPv4): A rule to check that all is fine (IPv4/IPv6 dual-stack):
Line 77: Line 141:
# nft add rule filter input counter accept # nft add rule inet filter input counter accept
Line 82: Line 146:
# nft list table filter # nft list table inet filter
Line 87: Line 151:
# nft flush chain filter input # nft flush chain inet filter input
Line 92: Line 156:
# nft delete chain filter input # nft delete chain inet filter input
Line 97: Line 161:
# nft delete table filter # nft delete table inet filter
Line 100: Line 164:
The family parameter is optional. The default is 'ip': The family parameter is optional. The default is 'ip'. Other families are 'inet', 'ip6', 'arp', 'bridge' or 'netdev':
Line 107: Line 171:
Debian ships an example configuration: Debian ships example configurations in:
Line 109: Line 173:
# nft -f /usr/share/doc/nftables/examples/basic.nft #/usr/share/doc/nftables/examples/
Line 112: Line 176:
Count traffic on destination port tcp/22: Count traffic on destination port tcp/22 (IPv4/IPv6 dual-stack):
Line 114: Line 178:
# nft add rule filter input tcp dport 22 counter # nft add rule inet filter input tcp dport 22 counter
Line 117: Line 181:
Count and accept traffic in 80/tcp and 443/tcp in new an establised state: Count and accept traffic in 80/tcp and 443/tcp in new and established state (IPv4/IPv6 dual-stack):
Line 119: Line 183:
# nft add rule filter input tcp dport {80, 443} ct state new,established counter accept
}}}

Export the ruleset in XML format (importing not yet supported):
{{{
nft export xml
# nft add rule inet filter input tcp dport {80, 443} ct state new,established counter accept

Translation(s): English - Italiano - Русский


nftables is a framework by the Netfilter Project that provides packet filtering, network address translation (NAT) and other packet mangling.

Two of the most common uses of nftables is to provide firewall support and Network Address Translation (NAT).

nftables is the default and recommended firewalling framework in Debian, and it replaces the old iptables (and related) tools.

Current status

NOTE: Debian 10 Buster and later use the nftables framework by default.

Starting with Debian 10 Buster, nf_tables is the default backend when using iptables, by means of the iptables-nft layer (i.e, using iptables syntax with the nf_tables kernel subsystem). This also affects ip6tables, arptables and ebtables.

Hints

Some hints folks might find interesting in some situations.

Use firewalld

You should consider using a wrapper instead of writing your own firewalling scripts. It is recommended to run ?firewalld, which integrates pretty well into the system. See also https://firewalld.org/

The firewalld software takes control of all the firewalling setup in your system, so you don't have to know all the details of what is happening in the underground. There are many other system components that can integrate with firewalld, like NetworkManager, libvirt, podman, fail2ban, docker, etc.

Reverting to legacy xtables

You can switch back and forth between iptables-nft and iptables-legacy by means of update-alternatives (same applies to arptables and ebtables).

The default starting with Debian 10 Buster:

# update-alternatives --set iptables /usr/sbin/iptables-nft
# update-alternatives --set ip6tables /usr/sbin/ip6tables-nft
# update-alternatives --set arptables /usr/sbin/arptables-nft
# update-alternatives --set ebtables /usr/sbin/ebtables-nft

Switching to the legacy version:

# update-alternatives --set iptables /usr/sbin/iptables-legacy
# update-alternatives --set ip6tables /usr/sbin/ip6tables-legacy
# update-alternatives --set arptables /usr/sbin/arptables-legacy
# update-alternatives --set ebtables /usr/sbin/ebtables-legacy

nftables in Debian the easy way

If you want to enable a default firewall in Debian, follow these steps:

# aptitude install nftables
# systemctl enable nftables.service

This way, nftables is active at boot. By default, rules are located in /etc/nftables.conf.

To stop nftables from doing anything, just drop all the rules:

# nft flush ruleset

To prevent nftables from starting at boot:

# systemctl mask nftables.service

To uninstall it and purge any traces of nftables in your system:

# aptitude purge nftables

FAQ

What is nftables?

Is the new framework by the Netfilter Project, allowing you to perform packet filtering (firewalling), NAT, mangling and packet classification.

Should I build a firewall using a nftables?

Yes. Building new firewalls on top of iptables is discouraged.

Should I replace an iptables firewall with a nftables one?

Yes, nftables is the replacement for iptables. There are some tools in place to ease in this task.

Please read: https://wiki.nftables.org/wiki-nftables/index.php/Moving_from_iptables_to_nftables

Why a new framework?

The previous framework (iptables) has several problems hard to address, regarding scalability, performance, code maintenance, etc..

What are the major differences?

In iptables there are several tables (filter, nat) and chains (FORWARD, INPUT...) by default. In nftables, there are no default tables/chains.

Also, in iptables you only have one target per rule (-j ACCEPT, -j LOG ...). In nftables, you can perform several actions in one single rule.

nftables includes built-in data sets capabilities. In iptables this is not possible, and there is a separated tool: ?ipset.

In the iptables framework there are tools per family: iptables, ip6tables, arptables, ebtables. Now, nftables allows you to manage all families in one single CLI tool.

This new framework features a new linux kernel subsystem, known as nf_tables. The new engine mechanism is inspired by BPF-like systems, with a set of basic expressions, which can be combined to build complex filtering rules.

Should I mix nftables and iptables/ebtables/arptables rulesets?

No, unless you know what you are doing.

I knew the iptables syntax. Is there a new syntax in nftables?

Yes, but the nftables one is better :-)

Help in migrating to nftables: https://wiki.nftables.org/wiki-nftables/index.php/Moving_from_iptables_to_nftables

new syntax

Create a basic IPv4/IPv6 dual-stack table:

# nft add table inet filter

Create a chain for input IPv4/IPv6 dual-stack traffic:

# nft add chain inet filter input { type filter hook input priority 0\; }

A rule to check that all is fine (IPv4/IPv6 dual-stack):

# nft add rule inet filter input counter accept

Show all the previous:

# nft list table inet filter

Flush rules in chain filter/input:

# nft flush chain inet filter input

Delete the chain filter/input:

# nft delete chain inet filter input

Delete the table filter:

# nft delete table inet filter

The family parameter is optional. The default is 'ip'. Other families are 'inet', 'ip6', 'arp', 'bridge' or 'netdev':

# nft add table ip6 filter
# nft add chain ip6 filter input
# nft add rule ip6 filter input counter accept

Debian ships example configurations in:

#/usr/share/doc/nftables/examples/

Count traffic on destination port tcp/22 (IPv4/IPv6 dual-stack):

# nft add rule inet filter input tcp dport 22 counter

Count and accept traffic in 80/tcp and 443/tcp in new and established state (IPv4/IPv6 dual-stack):

# nft add rule inet filter input tcp dport {80, 443} ct state new,established counter accept

external resources

Check out the official nftables wiki: http://wiki.nftables.org/


CategorySystemAdministration