Ipv6 Cheat Sheet

Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

   Internet Protocol Version 6 (IPv6) Basics Cheat Sheet IPv6 Addresses

by Jens Roesen
/64 – lan segment, 18,446,744,073,709,551,616 v6 IPs
IPv6 quick facts /48 – subscriber site, 65536 /64 lan segments
successor of IPv4 • 128-bit long addresses • that's 296 times the IPv4 address space • that's 2128 or 3.4x1038 or over 340 /32 – minimum allocation size, 65536 /48 subscriber sites, allocated to ISPs
undecillion IPs overall • customer usually gets a /64 subnet, which yields 4 billion times the Ipv4 address space • no
need for network address translation (NAT) any more • no broadcasts any more • no ARP • stateless address 2001:0db8:0f61:a1ff:0000:0000:0000:0080
configuration without DHCP • improved multicast • easy IP renumbering • minimum MTU size 1280 • mobile IPv6 • global routing prefix subnet ID interface ID
mandatory IPsec support • fixed IPv6 header size of 40 bytes • extension headers • jumbograms up to 4 GiB
subnet prefix /64

IPv6 & ICMPv6 Headers


IPv6 addresses are written in hexadecimal and divided into eight pairs of two byte blocks, each containing four hex
IPv6 header digits. Addresses can be shortened by skipping leading zeros in each block. This would shorten our example address to
0 8 16 24 32 2001:db8:f61:a1ff:0:0:0:80. Additionally, once per IPv6 IP, we can replace consecutive blocks of zeros with a
version traffic class flow label double colon: 2001:db8:f61:a1ff::80.
The 64-bit interface ID can/should be in modified EUI-64 format. A MAC 00 03 ba 24 a9 6c
payload length next header hop limit 48-bit MAC can be transformed to an 64-bit interface ID by
inverting the 7 (universal) bit and inserting a ff and fe byte after
th

source IPv6 address the 3rd byte. So the MAC 00:03:ba:24:a9:c6 becomes EUI-64 02 03 ba ff fe 24 a9 6c
0203:baff:fe24:a9c6. See RFC 4291 Appendix A and RFC 4941.

IPv6 Address Scopes


destination IPv6 address ::/128 unspecified address
::1/128 localhost
fe80::/10 link local scope
Version (4 bits): IP version. Always 6.
Traffic class (8 bits): Used for QoS. Like the TOS field in IPv4. RFC 2474. fec0::/10 site local scope, intended as RFC 1918 successor, deprecated in RFC 3879
Flow label (20 bits): Used for packet labelling, End-to-end QoS. RFC 6437. fc00::/7 unique local unicast scope, RFC 4193, divided into:
Payload length (16 bits): Length of the payload following the header in bytes. Limits packet size to 64 KB. fc00::/8 centrally assigned by unknown (see http://bit.ly/IETFfc00), routed within a site
Next header (8 bits): Code for the following extension header or UL protocol. Like protocol type field in IPv4. fd00::/8 free for all, global ID must be generated randomly, routed within a site
Hop limit (8 bits): Number of hops until the packet gets discarded. TTL in IPv4.
ff00::/8 multicast scope, after the prefix ff there are 4 bits for flags (0RPT) and 4 bits for the scope
Source address (128 bit): IPv6 source address.
Destination address (128 bits): IPv6 destination address. ::/96 IPv4-compatible IPv6 address, example: ::192.168.1.2, deprecated with RFC 4291
::ffff:0:0/96 IPv4-mapped IPv6 address, example: ::ffff:192.168.2.1, see RFC 4038
ICMPv6 header
2000::/3 global unicast scope, divided into:
0 8 16 24 32
2001::/16 /32 subnets assigned to providers, they assign /48, /56 or /64 to the customer
ICMPv6 type ICMPv6 code ICMPv6 checksum
2001::/32 Global Teredo IPv6 Service Prefix.
ICMPv6 data
2001:db8::/32 Reserved for use in documentation. See RFC 3849.
ICMP type (8 bits): Error messages have a 0 high-order-bit (types 0 to 127), info messages have a 1 high-order-bit 2001:678::/29 Provider Independent (PI) space and anycasting TLD nameservers (f.i. 2001:678:2::/48 for DENIC)
(types 128 to 255). 2002::/16 6to4 scope, 2002:c058:6301:: is the 6to4 public router anycast (deprecated with RFC 7526)
ICMP code (8 bits): Further specifies the kind of message along with the type. F.i. type 1 code 4 is “destination port 3ffe::/16 6Bone scope, returned to IANA with RFC 3701, you should not see these
unreachable”.
64:ff9b::/96 prefix used for representing IPv4 addresses in the IPv6 address space, see RFC 6052
ICMP checksum (16 bits): Checksum to prevent data corruption.
Well Known Multicast Addresses (T-Flag = 0) Multicast Scopes
IPv6 Extension Headers (RFC 2460 and it's updates)
ff0X::1 all nodes address (scopes 1 and 2) 1 Interface-local 5 Site-local
Because of the IPv6 header simplification and fixed size of 40 bytes IPv6 Header NH 0
ff0X::2 all routers address (scopes 1, 2 and 5) 2 Link-local 8 Organization-Local
(compared to the IPv4 header with more fields and options and 20 to 60
order as suggested in RFC 2460

Hop-by-Hop Options (0) NH 60


bytes in size) additional IP options were moved from the main IPv6 header Destination Options (60) NH 43 ff05::1:3 all site-local DHCP servers 3 Admin-local e Global
into additional headers. These extension headers (EH) will be appended to Routing Header(43) NH 44 ff02::9 all link-local RIP routers ← A “X” in the prefix is a place holder for the scope ↑
the main header as needed. The first 8 bit of each EH identify the next Fragment Header(44) NH 51
header (another EH or upper layer protocol) following. Only the hop-by-hop ff02::1:ff/104 solicited-node address, the 24 low-order bits are equal to the interfaces IP 24 low-order bits
Authentication Header (51) NH 50
header must be examined by every node on the path and, if present, it ESP Header (50) NH 60 ff02::1:2 all link-local DHCP relay agents and servers
must be the first header following the main IPv6 header. Every EH must only Destination Options (60) NH 6 ff0X::fb Multicast Domain Name Service v6 (all scopes)
occur once, only the destination options EH may occur twice - before a TCP Header (6)
routing EH and before the upper layer header. ff0X::101 Network Time Protocol (all scopes)
Neighbor Discovery (ND): Neighbor Solicitation (NS) and Neighbor Advertisement (NA) IPv6 and DNS (RFC 3596)
Neighbor Solicitation (ICMPv6 type 135) messages are sent to determine the link-layer address of a neighbor The IPv6 equivalent to the IPv4 A Resource Record is the AAAA RR. No big difference there. The A6 RR with additional
(multicasts) or to verify that a neighbor is still reachable (unicasts). fields for prefix length and prefix name defined in RFC 2874 was declared experimental in favour of AAAA RRs. See RFC
3363 and 3364 for more information and discussion.
2001:db8::1 → ff02::1:ff00:2 (destination IP is the destinations solicited-node multicast address)
ICMPv6 type 135, target 2001:db8::2, option 1 (source link-layer addr) 00:03:ba:24:a9:6c Reverse delegation is written in nibble format with the new domain IP6.ARPA. So 2001:db8::2 becomes
INITIATOR

TARGET
2.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.8.b.d.0.1.0.0.2.ip6.arpa.. The shorter
2001:db8::2 → 2001:db8::1, ICMPv6 type 136, Flags: S
target 2001:db8::2, option 2 (target link-layer) 00:03:ba:2e:02:c1 bitstring/bitlabel format (RFC 2672 and 2673) was later discarded but may be still used by older resolvers (BIND > 8.3
and < 9.2.3). The tool ipv6calc helps with the conversion!
In the example above node 2001:db8::1 wants to reach 2001:db8::2 but does not know the link-layer address of The host command will look for both A and AAAA records, using dig you have to explicitly ask for AAAA records (dig
2001:db8::2. So it sends a NS packet to the solicited-node multicast address of 2001:db8::2 host.example.com aaaa). Reverse lookups as usual can be done using host without further switches (host
(ff02::1:ff00:0/104 followed by the last 24 bits of the interface ID) along with its own link-layer address and 2001:db8::1) or with dig using the -x switch (dig -x 2001:db8::1).
receives a NA (ICMPv6 type 136) packet with the targets link-layer address.
Linux IPv6 Interface Configuration examples (Steps might slightly differ between distributions)
Duplicate Address Detection (DAD): To perform DAD the NS message is sent with the unspecified source IP :: and to
the solicited-node multicast address of the IP which should be configured. If there is already a node using this desired Manual configuration: You can temporarily configure an IPv6 address with the ifconfig or ip command:
# ifconfig eth0 inet6 add 2001:db8::2/64 or
IP it will answer with a NA packet sent to the all-node multicast address ff02::1.
# ip addr add 2001:db8::2/64 dev eth0
Neighbor Discovery (ND): Router Solicitation (RS) and Router Advertisement (RA) Add a default route
# route -A inet6 add default 2001:db8::1 or
Router Solicitation (RS) packets are sent in order to receive a Router Advertisement (RA) message independently from # ip -6 route add default via 2001:db8::1
the periodically sent RAs. This is typical during stateless address autoconfiguration after successful DAD. The source IP To check the configuration use ifconfig eth0 or ip -6 addr show eth0 respectively route -A inet6 or ip
used for the RS message can be :: or the link-local IP for this interface.
-6 route show. For making the changes permanent you'll have to edit the config files specific for your distribution.
:: or fe80::203:baff:fe24:a96c → ff02::2, ICMPv6 type 133 (RS)
option 1 (source link-layer) 00:03:ba:24:a9:6c (only when source IP is not ::) Automatic configuration using SLAAC: Just having IPv6 enabled and IPv4 configured on the interface should normally

ROUTER
NODE

fe80::21e:79ff:fe1e:f000 → ff02::1, ICMPv6 type 134 (RA), lifetime 1800s


do the trick.
option 2, option 3 (prefix information) 2001:db8::/64, pltime = 7d, vltime = 30d
SLAAC with privacy extensions (RFC 4941): To deal with security and privacy concerns regarding EUI-64 interface IDs
After receiving the RS message a router sends a RA message to the all-nodes multicast address. The RA message enable and prefer temporary addresses over other public addresses with:
# sudo sysctl net.ipv6.conf.all.use_tempaddr = 2 all = for all (current) interfaces
contains, amongst others, information about the router lifetime (time in seconds the router expects to be a default default = for new interfaces
router), all available prefixes and their preferred (pltime) and valid (vltime) lifetimes. When pltime reaches zero the # sudo sysctl net.ipv6.conf.all.use_tempaddr = 2
eth0, wlan0, enp3s0 usw = specific interface
address becomes deprecated and should not be used for new connections. When the vltime reaches zero the address To make these settings boot proof put them into /etc/sysctl.conf.
becomes invalid. Change valid and preferred lifetime of temporary addresses by editing net.ipv6.conf.all.temp_valid_lft and
net.ipv6.conf.all.temp_prefered_lft. Defaults are 604800 (7d) and 86400 (1d) seconds.
Stateless Address Autoconfiguration (RFC 4862) and Stateful Autoconfiguration DHCPv6 (RFC 3315)
Stateless Address Autoconfiguration (SLAAC) comes in handy when it's not important which exact address a node *NIX IPv6 Console Tools
uses as long as it's properly routable. SLAAC uses mechanisms of Neighbor Discovery. ping6 IPv6 version of ping. Solaris ping supports IPv6 out of the box.
Steps taken during SLAAC presuming there were no DAD errors along the way: forming a link-local address → DAD for traceroute6 IPv6 versions of traceroute and tracepath. Also try mtr -6.
the link-local address → activating the link-local address and sending RS message(s) to ff02::2 → forming a global tracepath6
address for each received prefix within an RA message with set "autonomous address-configuration flag" → DAD for
each tentative global address → addresses become valid and preferred (for pltime > 0). See RFC 6106 for DNS ip -6 Configure or view interfaces, routes, ND, list neighbors, multicasts.... on linux
configuration options advertising via RAs. ipv6calc Powerful tool for all sorts of conversions and information gathering. See
http://www.deepspace6.net/projects/ipv6calc.html
DHCPv6 can assign IPs and additional information like DNS/NTP Servers. A client sends a SOLICIT message (type 1) to
the All_DHCP_Relay_Agents_and_Servers multicast IP ff02::1:2. Servers answer with a ADVERTISE message (2). The tcpdump ip6 Packet sniffing tools with IPv6 options. Also works with options like icmp6.
client chooses a server, sends a REQUEST message (3) and receives a REPLY message (7) with configuration options. snoop inet6
DAD has to be performed for every address received! Alternatively, and in coexistence with SLAAC, DHCPv6 can only
provide clients with additional information like DNS and NTP servers. The client sends a INFORMATION-REQUEST IPv6 RFCs (available at http://tools.ietf.org/html/rfc<RFC number>)
message (11) and receives the options in a REPLY message (7). See RFC 3315 for detailed description of DHCPv6 RFC 8200 IPv6 Specifications RFC 4193 Unique Local IPv6 Unicast Addresses
messages and options. RFC 4291 IPv6 Addressing Architectures RFC 2375 IPv6 Multicast Address Assignments
Connect to IPv6 IPs on the Command Line or in a Browser RFC 4861 IPv6 Neighbor Discovery RFC 3849 IPv6 Address Prefix For Documentation

CLI # ssh '2001:db8:dead:f00d:203:baff:fe24:a9c6' RFC 4862 IPv6 Stateless Address Configuration RFC 4941 Privacy Extensions for SLAAC in IPv6
RFC 8201 Path MTU Discovery for IPv6 RFC 6147 DNS64 – DNS Extensions for NAT64
# lynx http://[2001:db8:dead:f00d:203:baff:fe24:a9c6]
RFC 3596 DNS Extensions to Support IP Version 6 RFC 6146 Stateful NAT64
# wget ftp://[2001:db8:dead:f00d:203:baff:fe24:a9c6]
RFC 4443 ICMPv6 for IPv6 RFC 6434 IPv6 Node Requirements
Browser http://[2001:db8:dead:f00d:203:baff:fe24:a9c6]
RFC 3587 IPv6 Global Unicast Address Format RFC 6540 IPv6 Support Required for All IP-Capable Nodes

You might also like