On Friday, July 26th, 2024 at PM 10:42, David Wright <deb...@lionunicorn.co.uk> wrote:
> There is a bug report #960809, which seems related, and > might be worth adding your experience to, if you think so. > > https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=960809 > Thank you for letting me know about this. I also noticed this issue earlier, but it seems to have been resolved, or at least I can not reproduce it. Not, all the main IPv6 address of each subnet are correctly mark as `mngtmpaddr`. Is it still reasonable to add my experience to existing bug report, or should I submit a new one instead? But I still worried this is an isolated case. Will try to reproduce it on another environment once I got time. > > I did wonder whether any of the randomness wrt reboots might be > time-related, as skim reading the RFC, it seems to allow for storing > a history of addresses used, and periodic generation of new ones > rather than a fresh one every reboot. > I didn't pay much attention to this since it only worked as expected maybe once or twice out of dozens of attempts. Based on my experience, the `addr_gen_mode` sysctl parameter might be responsible for this. It can stabilize the auto-configured IPv6, but it only affect the management IPv6; the temporary one changes with every reboot. Strangely, I do have a few Debian server that work regardless of the configuration order, but freshly installed systems never do.
