IPv4 Exhausted? Complete IPv6 Migration Guide for Linux Ops—from Configuration to Troubleshooting

IPv4 address pools have been depleted, making IPv6 migration essential for Linux operations. The article begins with a clear explanation of IPv6 address format (128‑bit, eight hexadecimal groups) and its massive address space, then introduces the three main IPv6 address types: link‑local (fe80::/10), unique local (fc00::/7), and global unicast (2000::/3). It contrasts IPv6 with IPv4, highlighting the disappearance of NAT, automatic address configuration via SLAAC, multicast‑only traffic, and native IPSec support.

Checking Kernel Support

Modern Linux distributions (post‑2010) enable IPv6 by default, but the guide advises confirming support using three methods:

# Method 1: cat /proc/net/if_inet6
cat /proc/net/if_inet6
# Method 2: lsmod | grep ipv6
lsmod | grep ipv6
# Method 3: sysctl net.ipv6.conf.all.disable_ipv6
sysctl net.ipv6.conf.all.disable_ipv6

If the last command returns net.ipv6.conf.all.disable_ipv6 = 1, IPv6 is disabled and can be re‑enabled temporarily with sysctl -w net.ipv6.conf.all.disable_ipv6=0 or permanently by adding net.ipv6.conf.all.disable_ipv6 = 0 to /etc/sysctl.conf and reloading with sysctl -p. Some systems may also disable IPv6 via the GRUB parameter ipv6.disable=1, which should be removed and GRUB updated.

Configuration on Major Distributions

CentOS/RHEL 7/8/9

For CentOS 7/8, network scripts are used:

# Find interface name
ip addr show
# Edit configuration
vi /etc/sysconfig/network-scripts/ifcfg-eth0
# Add/modify
IPV6INIT=yes
IPV6_AUTOCONF=no
IPV6_DEFROUTE=yes
IPV6_FAILURE_FATAL=no
IPV6ADDR=2408:xxxx:xxxx:xxxx::100/64
IPV6_DEFAULTGW=fe80::1
IPV6_DNS1=2400:3200::1
IPV6_DNS2=2400:3200:baba::1
# Restart NetworkManager
systemctl restart NetworkManager

CentOS 9 replaces network‑scripts with nmcli:

# Show connections
nmcli con show
# Modify IPv6 settings
nmcli con mod 'Wired connection 1' \
  ipv6.method manual \
  ipv6.addresses '2408:xxxx::100/64' \
  ipv6.gateway 'fe80::1' \
  ipv6.dns '2400:3200::1,2400:3200:baba::1'
# Apply changes
nmcli con up 'Wired connection 1'

Ubuntu 18.04/20.04/22.04

Ubuntu uses netplan YAML files:

network:
  ethernets:
    eth0:
      addresses:
        - 2408:xxxx:xxxx:xxxx::100/64
      gateway6: fe80::1
      nameservers:
        addresses:
          - 2400:3200::1
          - 2400:3200:baba::1
  version: 2

netplan apply

YAML indentation must be two spaces; a mis‑indent can break the network.

Debian

Older Debian versions edit /etc/network/interfaces:

iface eth0 inet6 static
  address 2408:xxxx:xxxx:xxxx::100/64
  gateway fe80::1
  dns-nameservers 2400:3200::1 2400:3200:baba::1

systemctl restart networking

Debian 12 also supports netplan with the same syntax as Ubuntu.

Verification and Testing

After configuration, verify the address with ip -6 addr show (or ip -6 addr show eth0) and ensure both a link‑local address (fe80::) and the configured global address appear. Check the routing table with ip -6 route show and confirm a default route like default via fe80::1 dev eth0. Perform connectivity tests using ping6 or ping -6 against the gateway, an IPv6 DNS server (e.g., 2400:3200::1), and a domain name (e.g., ipv6.google.com). Remember to append %eth0 when pinging link‑local addresses.

Common Troubleshooting Scenarios

DNS AAAA records missing

Use nslookup -type=aaaa ipv6.google.com or dig AAAA ipv6.google.com. If no AAAA record is returned, switch to a DNS server that supports IPv6.

No default route

Add it manually: ip -6 route add default via fe80::1 dev eth0.

Firewall blocking IPv6

Check forwarding with sysctl net.ipv6.conf.all.forwarding; enable with sysctl -w net.ipv6.conf.all.forwarding=1. Configure ip6tables rules to allow loopback, established connections, SSH, and ICMPv6, then set default policies to DROP and save the rules.

SSH slow due to reverse DNS

Disable DNS lookup in /etc/ssh/sshd_config by setting UseDNS no and restart SSH.

Services not listening on IPv6

Check listening sockets with ss -tulpn | grep :80. A line showing :::80 indicates dual‑stack listening. Adjust Nginx or MySQL configurations to bind to :: or use listen [::]:80 ipv6only=off;.

Missing ip6tables rules

Typical rules include allowing loopback, established connections, SSH, and ICMPv6 echo requests, then dropping everything else. Save rules with service ip6tables save (CentOS) or ip6tables-save > /etc/ip6tables/rules.v6 (Ubuntu/Debian).

Best Practices for IPv6 Operations

Adopt a dual‑stack approach (IPv4 + IPv6) while the ecosystem transitions. Publish both A and AAAA records in DNS, prefer IPv6 for internal traffic, and monitor both stacks. Suggested address planning reserves ::1‑::100 for network devices, ::101‑::200 for servers, and ::201‑::254 for gateways.

Monitoring checks include:

Presence of a global IPv6 address: ip -6 addr show eth0 | grep -q 'inet6.*global' Default route existence: ip -6 route show | grep -q 'default' Periodic ping6 to a known DNS server

Neighbor discovery table: ip -6 neigh show Recommended tools: traceroute6, tcpdump -i eth0 ip6, mtr -6, and curl -6 for IPv6‑specific testing.

Conclusion

IPv6 adoption is inevitable; mastering its concepts, configuration, validation, and troubleshooting equips Linux administrators to transition smoothly and avoid the pitfalls that many encounter.