[675] in Moira
Re: Hosttable in Moira
daemon@ATHENA.MIT.EDU (Theodore Ts'o)
Wed Oct 20 13:11:52 1993
Date: Wed, 20 Oct 93 13:11:07 EDT
From: tytso@MIT.EDU (Theodore Ts'o)
To: Mark Rosenstein <mar@MIT.EDU>
Cc: jis@MIT.EDU, moiradev@MIT.EDU, ops@MIT.EDU, network@MIT.EDU
In-Reply-To: Mark Rosenstein's message of Wed, 20 Oct 93 12:23:02 EDT,
Date: Wed, 20 Oct 93 12:23:02 EDT
From: Mark Rosenstein <mar@MIT.EDU>
Sender: mar@MIT.EDU
Content-Length: 1030
I think it's a really bad idea to not sanity check hostnames in the
update_host query. Someone will enter bad data eventually and break
the updates. I`ve seen it happen to every other moira update. That's
why moira sanity checks almost everything now.
What sort of sanity checking are we talking about? I'm not sure we're
all talking about the same sorts of sanity checks. If we are talking
about the following:
must start with a letter, remaining chars letter, numbers, and
hyphen, must not end with a hyphen.
Well, we currently have three hostnames which violates these
conventions: 3-462-1, M&M, and HCS_. None of these will break the
updates. Those rules come from the DNS RFC --- it's a little anal that
way --- but the actual bind implementation doesn't care in the least.
So, it would be perfectly safe to bypass this check. However, I still
think the standard distribute client should do the check, and strongly
discourage a normal user from picking such a hostname.
If we're talking about a sanity check where we make sure that a hostname
doesn't conflict with an already exisiting CNAME or another hostname,
then of course we should be doing *that* check all the time. :-)
- Ted
