CentOS 6 – Disabling IPv6 Addressing
I have just moved a host from a network that supports static IPv4 and IPv6. The IPv4 addr is set in ifcfg-eth0, and the IPv6 via RA (I set the MAC so I get an IPv6 addr that I like).
I just moved the host to a network that supports static IPv4, but only dymanic IPv6, so at this time (until I get static IPv6), I need to disable the global IPv6 addressing. So in the ifcfg-eth0 file I set:
IPV6INIT=”no”
But I am still getting a global IPv6 (and of course local scope).
What else do I need to do to disable the listening for RA announcements and setting an IPv6 global address? I do not want to reboot the box. I
can restart the network as needed.
I seem to recall, once upon atime an option in /etc/sysconfig/network
thanks
17 thoughts on - CentOS 6 – Disabling IPv6 Addressing
‘All’ I need is for the system not to have a global IPv6 address. Then it will not try to connect to other global IPv6 systems which will reject the connection, as the IPv6 rDNS cannot be set, given it is a dynamic IPv6 assigned address from the ISP.
I tried:
# cat /etc/sysconfig/network NETWORKING=yes HOSTNAME=z9m9z.htt-consult.com NETWORKING_IPV6=no IPV6INIT=no
and ‘service network restart’ but still showing IPv6 addressing.
I ended up simply disabling IPv6 for postfix. If it does not send with IPv6, no MTAs will fail to accept an IPv6 connection with no DNS or rDNS
for that address. And really it SHOULD be that only postfix would be starting an outbound connection. Well there is spamassasin and clamav, but I don’t believe they verify the source IP address asking for updates.
I would try adding the below line to /etc/sysconfig/network.
IPV6_AUTOCONF=no
Ryan
Try:
sysctl -w net.ipv6.conf.all.accept_ra=1
to persist between boots, be sure to add this to your /etc/sysctl.conf file.
This should prevent the box from listening to any RA announcements.
Chris
Sorry – that should be
sysctl -w net.ipv6.conf.all.accept_ra=0
to disable that, not 1.
Chris
+1
IPv6 = solution looking for a problem.
Disabled on all our systems!
—–Original Message—
I have to disagree on that. NATs is the problem and I am one of the causes of that problem as one of the principals behind RFC 1918.
What has happened is that HTTP has become the transport for the Internet. Very bad in a number of ways.
But for another time. Perhaps. Right now I have to deal with a new ISP
that was on the road to static IPv6 when somehow the lead engineer kind of stopped responding to emails and I won’t find out the details until IETF later this month.
Added this, did a ‘service network restart’ and still seeing the IPv6 addr.
No change after running this and trying both:
system network restart
ifdown eth0; ifup eth0
Still having an IPv6 addr.
The box has been up for 140 days. Would like to keep it running…
This box is really Redsleeve 6, which is the port of CentOS 6 to arm.
The kernel I am using is the F19 kernel. All of this MIGHT be contributing to things not working as they would on a ‘normal’ CentOS
box. I am awaiting the start of the CentOS7-arm work ;)
How about, in your /etc/sysconfig/network file adding or editing the line for IPV6 to be:
NETWORKING_IPV6=no
and then try a ‘service network restart’ and see what you get.
Chris
One of the first things I tried. It is still in there and doing no difference.
What I have is:
# cat network NETWORKING=yes HOSTNAME=z9m9z.htt-consult.com NETWORKING_IPV6=no IPV6INIT=no IPV6_AUTOCONF=no
and:
cat network-scripts/ifcfg-eth0
DEVICE=”eth0″
BOOTPROTO=none ONBOOT=”yes”
TYPE=”Ethernet”
NAME=”System eth0″
MACADDR:67:15:00:00:03
MTU00
DNS1P.253.254.2
DNS22.168.224.2
GATEWAY=”50.253.254.14″
IPADDR=”50.253.254.3″
NETMASK=”255.255.255.240″
HOSTNAME=”z9m9z.htt-consult.com”
IPV6INIT=”no”
I have used all the magic glue to say “no ipv6” and it just chugs along.
On the contrary. NAT and HTTP are the reasons most households are connected. But now we have http 2.0 to provide some pretense of security.
it’s: service network restart
Try also setting these in sysctl:
net.ipv6.conf.all.disable_ipv6=1
net.ipv6.conf.default.disable_ipv6=1
net.ipv6.conf.eth0.disable_ipv6=1
and then run: service network restart
Peter
Typo. My dsyelxia at work again.
error: “net.ipv6.conf.eth0.disable_ipv6” is an unknown key
And no more IPv6. Now to document this so I can reverse it when the time comes!
thanks
Hmm, I’ve used the information in this link in the past with good results:
Don’t know how this would with with Redsleeve, but with both CentOS 6
and RHEL 6 it works fine. I was able to disable IPv6 on-the-fly without a reboot using the “sysctl -w” method.
Your Mileage May Vary!
s/eth0/your_interface_name/
…or just leave it out, it will probably work with one, or both of the other two.
Peter