10 Feb
2013
10 Feb
'13
11:59 a.m.
I've noticed that my internal Ubuntu servers have odd default routes for and fe80::
address:
::/0 fe80::204:edff:febc:b011 UGDAe 1024 0 1 eth0
It seems that this is picked up from my router somehow and I've found that there is a
need to use the following:
echo 0 > /proc/sys/net/ipv6/conf/default/accept_ra
Having looked at the configuration on my Bitfolk server I see the same sort of setup,
although more extensive:
pre-up echo "/sbin/modprobe ipv6" &&
/sbin/modprobe ipv6 || true
post-up echo "/proc/sys/net/ipv6/conf/default/forwarding=0" && echo 0
> /proc/sys/net/ipv6/conf/default/forwarding || true
post-up echo "/proc/sys/net/ipv6/conf/all/forwarding=0" && echo 0
> /proc/sys/net/ipv6/conf/all/forwarding || true
post-up echo "/proc/sys/net/ipv6/conf/default/accept_ra=0" && echo 0
> /proc/sys/net/ipv6/conf/default/accept_ra || true
post-up echo "/proc/sys/net/ipv6/conf/all/accept_ra=0" && echo 0
> /proc/sys/net/ipv6/conf/all/accept_ra || true
post-up echo "/proc/sys/net/ipv6/conf/$IFACE/accept_ra=0" && echo 0
> /proc/sys/net/ipv6/conf/$IFACE/accept_ra || true
post-up echo "/proc/sys/net/ipv6/conf/default/autoconf=0" && echo 0
> /proc/sys/net/ipv6/conf/default/autoconf || true
post-up echo "/proc/sys/net/ipv6/conf/all/autoconf=0" && echo 0
> /proc/sys/net/ipv6/conf/all/autoconf || true
post-up echo "/proc/sys/net/ipv6/conf/$IFACE/autoconf=0" && echo 0
> /proc/sys/net/ipv6/conf/$IFACE/autoconf || true
I'm curious about two things:
1. why post-up and not pre-up
2. why doesn't this work on my own Ubuntu servers?
I've tried the single line in pre-up and post-up and the full set of commands above.
On one occasion I did loose the default route to the fe80:: address, but generally I
don't (with no configuration change!). This is on two servers, one 10.04 and the other
12.04. The Bitfolk one is still on 8.04 at the moment (although due an imminent upgrade to
12.04). As a side issue, my 10.04 server never brings up the IPv6 addressing on boot and I
always have to stop and restart it, and then services, to get things working properly (not
a major issue as it too is due an imminent upgrade to 12.04).
I thought I'd ask here since my reference was the Bitfolk configuration :)
--
Paul Tansom | Aptanet Ltd. | http://www.aptanet.com/ | 023 9238 0001
======================================================================
Registered in England | Company No: 4905028 | Registered Office:
Crawford House, Hambledon Road, Denmead, Waterlooville, Hants, PO7 6NU
10 Feb
10 Feb
3:10 p.m.
Hi Paul,
On Sun, Feb 10, 2013 at 11:59:30AM +0000, Paul Tansom wrote:
I've noticed that my internal Ubuntu servers have
odd default routes for and fe80:: address:
::/0 fe80::204:edff:febc:b011 UGDAe 1024 0 1 eth0
It is okay to have default route through a link-local address. You
would also get one on BitFolk if you didn't do any static routing
and just let RA set it up for you. As long as your default route is
reachable and works, does it matter which address on it is used?
It seems that this is picked up from my router somehow
and I've found that there is a need to use the following:
echo 0 > /proc/sys/net/ipv6/conf/default/accept_ra
A need to use that in order to achieve what outcome? Since that will
disable RA, you shouldn't pick up a default route by RA, yes. Was
disabling RA what you wanted to do?
Having looked at the configuration on my Bitfolk
server I see the same sort of setup, although more extensive:
pre-up echo "/sbin/modprobe ipv6" &&
/sbin/modprobe ipv6 || true
post-up echo "/proc/sys/net/ipv6/conf/default/forwarding=0" && echo 0
> /proc/sys/net/ipv6/conf/default/forwarding || true
post-up echo "/proc/sys/net/ipv6/conf/all/forwarding=0" && echo 0
> /proc/sys/net/ipv6/conf/all/forwarding || true
post-up echo "/proc/sys/net/ipv6/conf/default/accept_ra=0" && echo 0
> /proc/sys/net/ipv6/conf/default/accept_ra || true
post-up echo "/proc/sys/net/ipv6/conf/all/accept_ra=0" && echo 0
> /proc/sys/net/ipv6/conf/all/accept_ra || true
post-up echo "/proc/sys/net/ipv6/conf/$IFACE/accept_ra=0" && echo 0
> /proc/sys/net/ipv6/conf/$IFACE/accept_ra || true
post-up echo "/proc/sys/net/ipv6/conf/default/autoconf=0" && echo 0
> /proc/sys/net/ipv6/conf/default/autoconf || true
post-up echo "/proc/sys/net/ipv6/conf/all/autoconf=0" && echo 0
> /proc/sys/net/ipv6/conf/all/autoconf || true
post-up echo "/proc/sys/net/ipv6/conf/$IFACE/autoconf=0" && echo 0
> /proc/sys/net/ipv6/conf/$IFACE/autoconf || true
I'm curious about two things:
1. why post-up and not pre-up
If you do it early, the changing interface state can cause the
settings to change again anyway.
I've tried the single line in pre-up and post-up
and the full set of commands above. On one occasion I did loose the default route to the
fe80:: address, but generally I don't (with no configuration change!). This is on two
servers, one 10.04 and the other 12.04.
So you're saying you have all the above lines in your
/etc/network/interfaces but still end up with a default route to an
fe80 address set up by RA?
Cheers,
Andy
--
http://bitfolk.com/ -- No-nonsense VPS hosting
11 Feb
11 Feb
3:13 p.m.
** Andy Smith <andy(a)bitfolk.com> [2013-02-10 15:10]:
My understanding was that this stopped the interface picking up any addressing or routes
from the router for IPv6
Yup.
** end quote [Andy Smith]
--
Paul Tansom | Aptanet Ltd. | http://www.aptanet.com/ | 023 9238 0001
======================================================================
Registered in England | Company No: 4905028 | Registered Office:
Crawford House, Hambledon Road, Denmead, Waterlooville, Hants, PO7 6NU
On Sun, Feb 10, 2013 at 11:59:30AM +0000, Paul Tansom
wrote:
I initially started investigating this when I was failing to download updates through
aptitude or apt-get, before that I hadn't really noticed. When I 'ping6' the
IP address above I get 'connect: Invalid argument', although 'ping6 -I
eth0' works fine. All my routing is static on the servers is static so I had to work
out where this new route had come from!
I've noticed that my internal Ubuntu servers
have odd default routes for and fe80:: address:
::/0 fe80::204:edff:febc:b011 UGDAe 1024 0 1 eth0
It is okay to have default route through a link-local address. You
would also get one on BitFolk if you didn't do any static routing
and just let RA set it up for you. As long as your default route is
reachable and works, does it matter which address on it is used? It seems that
this is picked up from my router somehow and I've found that there is a need to use
the following:
echo 0 > /proc/sys/net/ipv6/conf/default/accept_ra
A need to use that in order to achieve what outcome? Since that will
disable RA, you shouldn't pick up a default route by RA, yes. Was
disabling RA what you wanted to do? Having looked
at the configuration on my Bitfolk server I see the same sort of setup, although more
extensive:
pre-up echo "/sbin/modprobe ipv6" &&
/sbin/modprobe ipv6 || true
post-up echo "/proc/sys/net/ipv6/conf/default/forwarding=0" && echo 0
> /proc/sys/net/ipv6/conf/default/forwarding || true
post-up echo "/proc/sys/net/ipv6/conf/all/forwarding=0" && echo 0
> /proc/sys/net/ipv6/conf/all/forwarding || true
post-up echo "/proc/sys/net/ipv6/conf/default/accept_ra=0" && echo 0
> /proc/sys/net/ipv6/conf/default/accept_ra || true
post-up echo "/proc/sys/net/ipv6/conf/all/accept_ra=0" && echo 0
> /proc/sys/net/ipv6/conf/all/accept_ra || true
post-up echo "/proc/sys/net/ipv6/conf/$IFACE/accept_ra=0" && echo 0
> /proc/sys/net/ipv6/conf/$IFACE/accept_ra || true
post-up echo "/proc/sys/net/ipv6/conf/default/autoconf=0" && echo 0
> /proc/sys/net/ipv6/conf/default/autoconf || true
post-up echo "/proc/sys/net/ipv6/conf/all/autoconf=0" && echo 0
> /proc/sys/net/ipv6/conf/all/autoconf || true
post-up echo "/proc/sys/net/ipv6/conf/$IFACE/autoconf=0" && echo 0
> /proc/sys/net/ipv6/conf/$IFACE/autoconf || true
I'm curious about two things:
1. why post-up and not pre-up
If you do it early, the changing interface state can cause the
settings to change again anyway.
I've tried the single line in pre-up and
post-up and the full set of commands above. On one occasion I did loose the default route
to the fe80:: address, but generally I don't (with no configuration change!). This is
on two servers, one 10.04 and the other 12.04.
So you're saying you have all the above lines in your
/etc/network/interfaces but still end up with a default route to an
fe80 address set up by RA?
5:25 p.m.
Hi Paul,
On Mon, Feb 11, 2013 at 03:13:15PM +0000, Paul Tansom wrote:
My understanding was that this stopped the interface picking up any addressing or routes
from the router for IPv6
Basically yes. Though you probably want to explicitly also disable
the "all" and "eth0" interfaces as well.
Okay, what is the output of:
$ grep . /proc/sys/net/ipv6/conf/*/{forwarding,accept_ra,autoconf}
please?
A link-local address of fe80::204:edff:febc:b011 implies a MAC
address of 00:04:ed:bc:b0:11, btw, and the OUI for 00:04:ED is
"Billion Electric Co., Ltd.". So hopefulyl that helps you track down
the device which is sending the RA.
Cheers,
Andy
¹ Stateless Address Auto Configuration.
https://en.wikipedia.org/wiki/IPv6#Stateless_address_autoconfiguration_.28S…
usually through Router Advertisement (RA) packets sent by the
default router.
--
http://bitfolk.com/ -- No-nonsense VPS hosting
** Andy Smith <andy(a)bitfolk.com> [2013-02-10
15:10]:
Right; when IPv6 is configured (through SLAAC¹ or statically), it can
start to expose problems if the configuration is not correct.
On Sun, Feb 10, 2013 at 11:59:30AM +0000, Paul
Tansom wrote:
I initially started investigating this when I was failing to download updates through
aptitude or apt-get, before that I hadn't really noticed. I've noticed that my internal Ubuntu servers
have odd default routes for and fe80:: address:
::/0 fe80::204:edff:febc:b011 UGDAe 1024 0 1 eth0
It is okay to have default route through a link-local address. You
would also get one on BitFolk if you didn't do any static routing
and just let RA set it up for you. As long as your default route is
reachable and works, does it matter which address on it is used? When I 'ping6' the IP address above I get
'connect: Invalid argument', although 'ping6 -I eth0' works fine.
That is normal behaviour; you can't communicate with link-local
addresses except from an address on that link, and that's what '-I
eth0' is forcing.
All my routing is static on the servers is static so I
had to work out where this new route had come from!
Yep, most operating systems that support IPv6 will automatically
listen to RA which does mean that any misconfigured device on the
network that starts to spew RA packets can potentially break
everything..
If you do want to statically configure IPv6 then that's one thing,
but you may also want to track down this device that is sending out
RA packets, especially if they are wrong.
It seems
that this is picked up from my router somehow and I've found that there is a need to
use the following:
echo 0 > /proc/sys/net/ipv6/conf/default/accept_ra
A need to use that in order to achieve what outcome? Since that will
disable RA, you shouldn't pick up a default route by RA, yes. Was
disabling RA what you wanted to do? > > post-up echo
"/proc/sys/net/ipv6/conf/default/forwarding=0" && echo 0 >
/proc/sys/net/ipv6/conf/default/forwarding || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/forwarding=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/forwarding || true
> > post-up echo "/proc/sys/net/ipv6/conf/default/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/default/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/$IFACE/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/$IFACE/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/default/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/default/autoconf || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/autoconf || true
> > post-up echo "/proc/sys/net/ipv6/conf/$IFACE/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/$IFACE/autoconf || true
So you're
saying you have all the above lines in your
/etc/network/interfaces but still end up with a default route to an
fe80 address set up by RA?
Yup.
12 Feb
12 Feb
12:29 p.m.
** Andy Smith <andy(a)bitfolk.com> [2013-02-11 17:25]:
My understanding was that this stopped the interface picking up any addressing or routes
from the router for IPv6
Basically yes. Though you probably want to explicitly also disable
the "all" and "eth0" interfaces as well.
I'll take a look at that. I'm heading back into the device to get back the route
cause at the moment!
Yup, although I've only just noticed that there's a variable in there for $IFACE!
That copy and paste was a late night 'what the heck I'll try this' type
scenario, which is fine as the server is both local and non-critical :)
For my two servers I have:
paul@humpback:~$ grep . /proc/sys/net/ipv6/conf/*/{forwarding,accept_ra,autoconf}
/proc/sys/net/ipv6/conf/all/forwarding:0
/proc/sys/net/ipv6/conf/default/forwarding:0
/proc/sys/net/ipv6/conf/eth0/forwarding:0
/proc/sys/net/ipv6/conf/lo/forwarding:0
/proc/sys/net/ipv6/conf/all/accept_ra:1
/proc/sys/net/ipv6/conf/default/accept_ra:0
/proc/sys/net/ipv6/conf/eth0/accept_ra:1
/proc/sys/net/ipv6/conf/lo/accept_ra:1
/proc/sys/net/ipv6/conf/all/autoconf:1
/proc/sys/net/ipv6/conf/default/autoconf:1
/proc/sys/net/ipv6/conf/eth0/autoconf:0
/proc/sys/net/ipv6/conf/lo/autoconf:1
paul@minke:~$ grep . /proc/sys/net/ipv6/conf/*/{forwarding,accept_ra,autoconf}
/proc/sys/net/ipv6/conf/all/forwarding:0
/proc/sys/net/ipv6/conf/default/forwarding:0
/proc/sys/net/ipv6/conf/eth0/forwarding:0
/proc/sys/net/ipv6/conf/lo/forwarding:0
/proc/sys/net/ipv6/conf/all/accept_ra:1
/proc/sys/net/ipv6/conf/default/accept_ra:0
/proc/sys/net/ipv6/conf/eth0/accept_ra:1
/proc/sys/net/ipv6/conf/lo/accept_ra:1
/proc/sys/net/ipv6/conf/all/autoconf:1
/proc/sys/net/ipv6/conf/default/autoconf:1
/proc/sys/net/ipv6/conf/eth0/autoconf:1
/proc/sys/net/ipv6/conf/lo/autoconf:1
minke is the one that doesn't bring up the IPv6 on eth0 automatically. The only
difference I've noted here is the eth0/autoconf:1 line.
Hi Paul,
On Mon, Feb 11, 2013 at 03:13:15PM +0000, Paul Tansom wrote:
I know what the device is, I just haven't managed to switch it off yet :(
** Andy Smith <andy(a)bitfolk.com>
[2013-02-10 15:10]:
Right; when IPv6 is configured (through SLAAC¹ or statically), it can
start to expose problems if the configuration is not correct.
On Sun, Feb 10, 2013 at 11:59:30AM +0000, Paul
Tansom wrote:
I initially started investigating this when I was failing to download updates through
aptitude or apt-get, before that I hadn't really noticed. I've noticed that my internal Ubuntu servers
have odd default routes for and fe80:: address:
::/0 fe80::204:edff:febc:b011 UGDAe 1024 0 1 eth0
It is okay to have default route through a link-local address. You
would also get one on BitFolk if you didn't do any static routing
and just let RA set it up for you. As long as your default route is
reachable and works, does it matter which address on it is used? When I 'ping6' the IP address above I get
'connect: Invalid argument', although 'ping6 -I eth0' works fine.
That is normal behaviour; you can't communicate with link-local
addresses except from an address on that link, and that's what '-I
eth0' is forcing.
All my routing is static on the servers is static
so I had to work out where this new route had come from!
Yep, most operating systems that support IPv6 will automatically
listen to RA which does mean that any misconfigured device on the
network that starts to spew RA packets can potentially break
everything..
If you do want to statically configure IPv6 then that's one thing,
but you may also want to track down this device that is sending out
RA packets, especially if they are wrong. It seems that
this is picked up from my router somehow and I've found that there is a need to use
the following:
echo 0 > /proc/sys/net/ipv6/conf/default/accept_ra
A need to use that in order to achieve what outcome? Since that will
disable RA, you shouldn't pick up a default route by RA, yes. Was
disabling RA what you wanted to do? > >
post-up echo "/proc/sys/net/ipv6/conf/default/forwarding=0" && echo 0
> /proc/sys/net/ipv6/conf/default/forwarding || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/forwarding=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/forwarding || true
> > post-up echo "/proc/sys/net/ipv6/conf/default/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/default/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/$IFACE/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/$IFACE/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/default/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/default/autoconf || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/autoconf || true
> > post-up echo "/proc/sys/net/ipv6/conf/$IFACE/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/$IFACE/autoconf || true
> > So you're saying you have all the above lines in your
> > /etc/network/interfaces but still end up with a default route to an
> > fe80 address set up by RA? Yup.
Okay, what is the output of:
$ grep . /proc/sys/net/ipv6/conf/*/{forwarding,accept_ra,autoconf}
please? A link-local address of fe80::204:edff:febc:b011
implies a MAC
address of 00:04:ed:bc:b0:11, btw, and the OUI for 00:04:ED is
"Billion Electric Co., Ltd.". So hopefulyl that helps you track down
the device which is sending the RA.
Yup, that's the one, the only device it could be. I have router announcements off in
the web GUI.....
... and it would appear it takes a full reboot of the sever rather than just an
ifdown/ifup to sort things out! When I am home again I'll try it on the other server
which I'd rather not reboot remotely as it handles mail and I wouldn't be popular
with my wife if I rebooted it now!! With any luck I'm sorted at last. One of those
jobs that you keep popping back to and going over the same ground again because you
can't quite remember where you've got to in the debug process! Why do I never give
my own network quite the care and attention I do for customers?!
Cheers,
Andy
¹ Stateless Address Auto Configuration.
https://en.wikipedia.org/wiki/IPv6#Stateless_address_autoconfiguration_.28S…
usually through Router Advertisement (RA) packets sent by the
default router.
** end quote [Andy Smith]
--
Paul Tansom | Aptanet Ltd. | http://www.aptanet.com/ | 023 9238 0001
======================================================================
Registered in England | Company No: 4905028 | Registered Office:
Crawford House, Hambledon Road, Denmead, Waterlooville, Hants, PO7 6NU
3:02 p.m.
Hi Paul,
On Tue, Feb 12, 2013 at 12:29:10PM +0000, Paul Tansom wrote:
Yup, although I've only just noticed that there's a variable in there for $IFACE!
That copy and paste was a late night 'what the heck I'll try this' type
scenario, which is fine as the server is both local and non-critical :)
If this is in an /etc/network/interfaces file (or script called from
there), then $IFACE is literally correct as it is substituted for
the interface name, e.g. "eth0", that is being worked on.
It won't work if it is called from the command line of course
(unless you set IFACE first).
Note that accept_ra is set to 1 for both "eth0" and "all". I'm
not
sure if that's enough to turn autoconfiguration on, but it does
indicate that the above commands weren't run.
> > post-up echo
"/proc/sys/net/ipv6/conf/default/forwarding=0" && echo 0 >
/proc/sys/net/ipv6/conf/default/forwarding || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/forwarding=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/forwarding || true
> > post-up echo "/proc/sys/net/ipv6/conf/default/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/default/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/$IFACE/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/$IFACE/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/default/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/default/autoconf || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/autoconf || true
> > post-up echo "/proc/sys/net/ipv6/conf/$IFACE/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/$IFACE/autoconf || true
> > So you're saying you have all the above lines in your
> > /etc/network/interfaces but still end up with a default route to an
> > fe80 address set up by RA? Okay, what is
the output of:
$ grep . /proc/sys/net/ipv6/conf/*/{forwarding,accept_ra,autoconf}
please?
For my two servers I have:
paul@humpback:~$ grep . /proc/sys/net/ipv6/conf/*/{forwarding,accept_ra,autoconf}
/proc/sys/net/ipv6/conf/all/forwarding:0
/proc/sys/net/ipv6/conf/default/forwarding:0
/proc/sys/net/ipv6/conf/eth0/forwarding:0
/proc/sys/net/ipv6/conf/lo/forwarding:0
/proc/sys/net/ipv6/conf/all/accept_ra:1
/proc/sys/net/ipv6/conf/default/accept_ra:0
/proc/sys/net/ipv6/conf/eth0/accept_ra:1
/proc/sys/net/ipv6/conf/lo/accept_ra:1
/proc/sys/net/ipv6/conf/all/autoconf:1
/proc/sys/net/ipv6/conf/default/autoconf:1
/proc/sys/net/ipv6/conf/eth0/autoconf:0
/proc/sys/net/ipv6/conf/lo/autoconf:1 paul@minke:~$ grep .
/proc/sys/net/ipv6/conf/*/{forwarding,accept_ra,autoconf}
/proc/sys/net/ipv6/conf/all/forwarding:0
/proc/sys/net/ipv6/conf/default/forwarding:0
/proc/sys/net/ipv6/conf/eth0/forwarding:0
/proc/sys/net/ipv6/conf/lo/forwarding:0
/proc/sys/net/ipv6/conf/all/accept_ra:1
/proc/sys/net/ipv6/conf/default/accept_ra:0
/proc/sys/net/ipv6/conf/eth0/accept_ra:1
/proc/sys/net/ipv6/conf/lo/accept_ra:1
/proc/sys/net/ipv6/conf/all/autoconf:1
/proc/sys/net/ipv6/conf/default/autoconf:1
/proc/sys/net/ipv6/conf/eth0/autoconf:1
/proc/sys/net/ipv6/conf/lo/autoconf:1
minke is the one that doesn't bring up the IPv6 on eth0 automatically. The only
difference I've noted here is the eth0/autoconf:1 line.
When you say "doesn't bring up the IPv6 on eth0 automatically" do
you mean that as a good thing, i.e. it is doing the static
configuration that you expect?
Note again that some accept_ra and autoconf are set to 1 there so
that still hasn't been disabled like the above commands would do,
although has maybe been disabled "enough".
Anyway, on whichever server is misbehaving, if you went through and set
the above all to 0 where they are 1 I expect you would find that
after a minute or two the global scope v6 address and default route
would disappear.
... and it would appear it takes a full reboot of the
sever rather than just an ifdown/ifup to sort things out!
What new settings were you trying to put into effect with ifdown/ifup? Making it run the
post-up commands again?
Cheers,
Andy
--
http://bitfolk.com/ -- No-nonsense VPS hosting
15 Feb
15 Feb
11:59 a.m.
** Andy Smith <andy(a)bitfolk.com> [2013-02-12 15:03]:
Note that accept_ra is set to 1 for both "eth0" and "all". I'm
not
sure if that's enough to turn autoconfiguration on, but it does
indicate that the above commands weren't run.
Hmm, interesting.
Not so much a good thing, no. The same config is bringing IPv6 up on the other box
(admittedly a later Ubuntu version). This will be irrelevant once I've upgraded this
box too (I hope!), but is a curiousity.
That was the plan. Becoming a moot point now, but still a ciriousity that I may come back
to so I understand it - I do so hate using one solution and not understanding another
potential solution properly!
On Tue, Feb 12, 2013 at 12:29:10PM +0000, Paul Tansom
wrote:
Yup, although I've only just noticed that there's a variable in there for $IFACE!
That copy and paste was a late night 'what the heck I'll try this' type
scenario, which is fine as the server is both local and non-critical :)
If this is in an /etc/network/interfaces file (or script called from
there), then $IFACE is literally correct as it is substituted for
the interface name, e.g. "eth0", that is being worked on.
It won't work if it is called from the command line of course
(unless you set IFACE first).
Yes, run from /etc/network/interfaces
> > post-up echo
"/proc/sys/net/ipv6/conf/default/forwarding=0" && echo 0 >
/proc/sys/net/ipv6/conf/default/forwarding || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/forwarding=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/forwarding || true
> > post-up echo "/proc/sys/net/ipv6/conf/default/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/default/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/$IFACE/accept_ra=0"
&& echo 0 > /proc/sys/net/ipv6/conf/$IFACE/accept_ra || true
> > post-up echo "/proc/sys/net/ipv6/conf/default/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/default/autoconf || true
> > post-up echo "/proc/sys/net/ipv6/conf/all/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/all/autoconf || true
> > post-up echo "/proc/sys/net/ipv6/conf/$IFACE/autoconf=0"
&& echo 0 > /proc/sys/net/ipv6/conf/$IFACE/autoconf || true
> > So you're saying you have all the above lines in your
> > /etc/network/interfaces but still end up with a default route to an
> > fe80 address set up by RA? Okay, what is the output of:
$ grep . /proc/sys/net/ipv6/conf/*/{forwarding,accept_ra,autoconf}
please?
For my two servers I have:
paul@humpback:~$ grep . /proc/sys/net/ipv6/conf/*/{forwarding,accept_ra,autoconf}
/proc/sys/net/ipv6/conf/all/forwarding:0
/proc/sys/net/ipv6/conf/default/forwarding:0
/proc/sys/net/ipv6/conf/eth0/forwarding:0
/proc/sys/net/ipv6/conf/lo/forwarding:0
/proc/sys/net/ipv6/conf/all/accept_ra:1
/proc/sys/net/ipv6/conf/default/accept_ra:0
/proc/sys/net/ipv6/conf/eth0/accept_ra:1
/proc/sys/net/ipv6/conf/lo/accept_ra:1
/proc/sys/net/ipv6/conf/all/autoconf:1
/proc/sys/net/ipv6/conf/default/autoconf:1
/proc/sys/net/ipv6/conf/eth0/autoconf:0
/proc/sys/net/ipv6/conf/lo/autoconf:1 paul@minke:~$
grep . /proc/sys/net/ipv6/conf/*/{forwarding,accept_ra,autoconf}
/proc/sys/net/ipv6/conf/all/forwarding:0
/proc/sys/net/ipv6/conf/default/forwarding:0
/proc/sys/net/ipv6/conf/eth0/forwarding:0
/proc/sys/net/ipv6/conf/lo/forwarding:0
/proc/sys/net/ipv6/conf/all/accept_ra:1
/proc/sys/net/ipv6/conf/default/accept_ra:0
/proc/sys/net/ipv6/conf/eth0/accept_ra:1
/proc/sys/net/ipv6/conf/lo/accept_ra:1
/proc/sys/net/ipv6/conf/all/autoconf:1
/proc/sys/net/ipv6/conf/default/autoconf:1
/proc/sys/net/ipv6/conf/eth0/autoconf:1
/proc/sys/net/ipv6/conf/lo/autoconf:1
minke is the one that doesn't bring up the IPv6 on eth0 automatically. The only
difference I've noted here is the eth0/autoconf:1 line.
When you say "doesn't bring up the IPv6 on eth0 automatically" do
you mean that as a good thing, i.e. it is doing the static
configuration that you expect? Note again that some accept_ra and autoconf are set to
1 there so
that still hasn't been disabled like the above commands would do,
although has maybe been disabled "enough".
Anyway, on whichever server is misbehaving, if you went through and set
the above all to 0 where they are 1 I expect you would find that
after a minute or two the global scope v6 address and default route
would disappear.
Both servers are misbehaving, or rather were, both are now playing nicely, but this is
down to the router config change that for some reason has now picked up (it may be the
lack of a reboot, but I'm pretty sure both that is an instant change and that the
router has been rebooted).
... and it
would appear it takes a full reboot of the sever rather than just an ifdown/ifup to sort
things out!
What new settings were you trying to put into effect with ifdown/ifup? Making it run the
post-up commands again?
Cheers,
Andy
--
http://bitfolk.com/ -- No-nonsense VPS hosting
_______________________________________________
users mailing list
users(a)lists.bitfolk.com
https://lists.bitfolk.com/mailman/listinfo/users
** end quote [Andy Smith]
--
Paul Tansom | Aptanet Ltd. | http://www.aptanet.com/ | 023 9238 0001
======================================================================
Registered in England | Company No: 4905028 | Registered Office:
Crawford House, Hambledon Road, Denmead, Waterlooville, Hants, PO7 6NU
4087
days inactive
4092
days old
6 comments
2 participants
participants (2)
-
Andy Smith -
Paul Tansom