Re: UNIX and ease of use
Avery Pennarun <firstname.lastname@example.org> writes:
> Idea: We could work out some kind of agreement to get an
> auto-registration script for new Debian users. These people look rather
> approachable, especially if we write the software for them.
> - dyn.ml.org is a flat name space. It's not hierarchical. If too many
> users start trying to get these addresses, the entire structure will
> collapse. Their approach not a good technical solution to the problem.
> - It's not using the actual RFC-compliant dynamic DNS "standard." Whether
> this is a problem or not depends on your point of view. I looked at the
> "real" standard and was a bit disgusted.
If we had the resources, and someone to manage it, we could solve all
these particular problems with our own DNS hosting. It'd certainly be
nice for users, and allow things to work without any Dynamic IP hacks.
I'd love to have my home machines under some-name-i-picked.dns.debian.org.
Is this even feasible? Are there free clients and servers for the
> - if I'm not online and I'm running named (to serve names on my LAN, or as
> a cache) internet DNS lookups take _forever_ to time out. This seriously
> confuses netscape. Both netscape (horribly) and BIND (somewhat) are
> broken. Starting/stopping named when the PPP link goes up/down works
> around the problem, but we can't kill named if we're serving names for a
Ahh. I hadn't thought of the LAN issue. Part of my solution involved
making /etc/host.conf a symlink, and swapping the link when ppp came
up or down. One file pointed to had bind specified, the other didn't.
I killed/restarted the server too. That won't work if you have to
keep providing name service locally, but most users won't.
You might be able to solve the LAN issue by also swapping named's
config on net up/down, but that's probably outside the scope of
anything you'd want to do for the "average" Dynamic-IP user.
> I would volunteer to fix these, but my time is massively constricted at the
> moment. I'd especially like to see a solution to the BIND problem. I may
> be fiddling with BIND internals at work sometime soon so it might
> conceivably get done... don't hold your breath.
I have time issues at the moment too, but it would be nice to have a
discussion of how to standardize our DNS approach:
Rob Browning <email@example.com>
PGP fingerprint = E8 0E 0D 04 F5 21 A0 94 53 2B 97 F5 D6 4E 39 30
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
Trouble? e-mail to firstname.lastname@example.org .