[v6ops] Re: Dynamic addresses

[Erik Auerswald <[e-mail]>]

Hi,

On Tue, Aug 13, 2024 at 06:54:39AM +0200, Marco Moock wrote:
> Am Tue, 13 Aug 2024 06:16:22 +0530
> schrieb Daryll Swer <[e-mail]>:
> 
> > Even if you change the values to N, you still have broken
> > connectivity for N time. I am of the opinion, that no, SLAAC is
> > indeed working as specified, and there are no reasons to mess with
> > the values. What needs to be changed is: for ISPs to give the user
> > options for either A. Dynamic ia_pd at the user's discretion with the
> > downsides for SLAAC operational behaviour Or B. Static ia_pd that
> > does not change (unless drastic changes occurs upstream, which also
> > applies to v4 anyway).
> 
> Can someone please explain why dynamic prefixes create a problem when
> used properly?

One example problem:

Long running TCP connections, e.g., SSH, break when the source address
can no longer be used, e.g., reaches a valid lifetime of 0.

I see this with changing IPv6 prefixes since IPv6 was added to my Internet
connection.  I did not see this before with only IPv4 (i.e., the external
IPv4 address was more stable than the delegated global IPv6 prefix).

Br,
Erik