[Icecast] Using Icecast over IPv6

[Thomas.Rucker at tieto.com]

Hi,

>-----Original Message-----
>From: icecast-bounces at xiph.org [mailto:icecast-bounces at xiph.org] On
>Behalf Of Geoff Shang
>Sent: Monday, 06 June, 2011 17:39
>Subject: Re: [Icecast] Using Icecast over IPv6

>On Mon, 6 Jun 2011, Thomas.Rucker at tieto.com wrote:
>
>> To your problem there are two approaches:
>> - bind explicitly to :: and to 0.0.0.0 (so an ipv4 and a separate ipv6
>bind)
>> - change your sysctl.conf to contain: "net.ipv6.bindv6only = 0"
>
>hmmm.  OK but this implies (does it not) that
>/proc/sys/net/ipv6/bindv6only should now contain something other than 0.
>It doesn't (i.e. it does contain 0).

Huh? Then we might be onto something different.
It's been ages since we sat down and fixed icecast to bind to ipv6 properly.
So my recollection is a bit fuzzy and distribution/kernel changes might
have changed our behaviour in an undesirable way.

What certainly works is adding this:
<bind-address>::</bind-address>
to your <listen-socket>. It will make icecast listen to ipv4 AND ipv6 IF
bindv6only is set to '0'.


>> If you opt for the latter don't forget to reload sysctl settings or
>> change that option using the 'sysctl' command. No reboot necessary!
>
>
>If I opt for the latter, will it break when I upgrade to the next Debian
>release?
>
>> This is because the latter setting changed in debian squeeze a while
>> ago. It now defaults to a setting that prevents listening for ipv6 and
>> ipv4 on the same bind. (This also used to break java)
>
>Is there a good reason to disable this?

I'd recommend to read (at least part) of the related debian issue:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=560044

But if it's already '0' then this is likely not the culprit.

Cheers

Thomas