Re: dns managment thru ganymede

[Jonathan Abbey <jonabbey@arlut.utexas.edu>]

Vikram (vkhare@winstar.com) wrote to Gil Kloepfer (gil@arlut.utexas.edu)
with a question about DNS and Ganymede.  This is an important issue, and
I wanted to say a few words about it and share it with the ganymede
mailing list (in addition to sending this to Vikram himself. ;-)

| > Gil,
| > 
| >     I pulled your e-mail off the ARL:UT web pages for Ganymede & I hope
| > you don't mind if I ask you a question.  I was wondering what the status
| > of DNS management was.  Would you recommend using Ganymede for managing
| > 7 or 8 zones (a few hundred hosts on 10.x.x.x space) or should we wait
| > till the 1.0 release?  If so, when do you think you guys would have that
| > come out?

Gil replied as follows:

| [...]
| That said, I am probably not the correct person to ask about this.
| However, I do know that we're currently discussing what the best
| way to manage DNS is going to be with Ganymede.  Unfortunately, there
| are lots of "special cases" here at ARL:UT with respect to how
| DNS is configured, but we do ultimately expect to handle almost
| all DNS management through Ganymede.  I don't know the current
| status of Ganymede's DNS management, but if all you're going to be
| doing is managing 7 or 8 zones of 10.x.x.x addresses, you can probably
| do that fairly easily.  The complications will likely occur if you
| are not using netmasks of 255.255.255.0, which is what GASH was
| pretty much limited to.

In Ganymede 0.98, only the GASH and GASHARL schema kits have any support
for DNS, and they are somewhat idiosyncratic, in that they assume that 
you are managing a single DNS domain across a set of discontinuous 'Class C'
networks (in our case, we've got a whole bunch of 255.255.255.0 subnets
pulled from one of UT's Class B's, so they're not actually Class C's, but
we pretend they are).

In addition, the GASH schema kits are designed to import data from GASH,
which uses a file called hosts_info which is rather complex and finicky.
There's currently no support for importing data from BIND zone files in
any way, so bootstrapping up into using Ganymede for DNS will likely be
a big pain for most people.

We're working on reworking the GASH schema so that we can support a collection
of networks with various subnet masks and designated IP ranges within those
networks, which we believe will allow us to support things like NAT and DHCP,
as well as allocating small subnetworks for our ISDN home users, and so forth.
I'm working on a document today describing a model for doing all this.  Once
I get this done and we kick it around a little bit here at ARL, I'll send it
to this list for comments.

Supporting multiple DNS domains needs to be there as well, and I'm not yet
sure I know how that really should be done.  One of the things that Ganymede
does not do well is to have a real hierarchy of objects in the tree.. that is,
the Ganymede client can't show objects in the tree that also have objects
underneath them (as in a DNS hierarchy, an IP subnet hierarchy, etc).  Doing
all of this right is complex, and might involve significant modifications to
the client and the server.

There are bound to be simpler ways of doing all that, but I need to put more
thought into that.

Most of this is not likely going to happen by Ganymede 1.0.  Depending on how
the extended subnet/IP-range document comes out, I may try to get that
implemented by 1.0, as it solves some problems for us that would prevent us
from doing some tricks we've done by hand-editing our GASH files in ways that
aren't strictly compatible with the GASH logic.  The multiple DNS domain
stuff will likely come after 1.0.

In addition, we need to get our NIS/DNS build logic cleaned up and tested
for 1.0 so people could plausibly use the GASH schema stuff in production.

With regards to the DNS zone-file import stuff, that's another big topic
that's going to need to be figured out by somebody..  GASH and the GASH schema
stuff we're doing for Ganymede are designed around a system-and-interface
abstraction, where Ganymede has an explicit awareness of a system as a
discrete object with attributes, and I'm not yet sure that that can be easily
synthesized from pre-existing zone files.  It may be awhile before someone
comes up with a good way to bootstrap naked zone files into Ganymede, at least
with a GASH-like schema.  It's entirely possible that someone could choose to
author a DNS schema for Ganymede that uses DNS records as the basic entities,
which could then easily be imported from BIND zone files, but that would
look very different from how we've been doing things so far with GASH and
Ganymede, and would probably be less approachable for the average group
administrator or secretary who just wants to get their new system an IP
address.

| >     Btw, I read all of the LISA papers and docs for this.  It looks
| > great!

Thanks very much.  I have fond hopes for it all, hopefully it will be very
useful for people.

| > ps, I'm running Ganymede under hp/ux 10.20 and so far NIS stuff works
| > fine.  If you
| > need some feedback for the beta releases under hp/ux let me know.

I would *love* to get feedback.  Feedback, feedback, feedback, gimme feedback!

Which schema kit are you using?  Did you have any difficulty getting data
imported?  Have you modified the schema or plug-in classes for customization
purposes?  Are you using a builder script to actually propagate your NIS data
into your NIS servers?  Have you created any custom roles?  Does everything
seem to make sense?  Where and how are you running the client?

;-)

-- Jon

-------------------------------------------------------------------------------
Jonathan Abbey				              jonabbey@arlut.utexas.edu
Applied Research Laboratories                 The University of Texas at Austin
-------------------------------------------------------------------------------

----------------------------------------------------------------------------
To make changes to your subscription to the ganymede mailing list, send
mail to majordomo@arlut.utexas.edu.

To unsubcribe, include the line

unsubscribe ganymede

in the body of your mail message

Visit the Ganymede web page at http://www.arlut.utexas.edu/gash2

----------------------------------------------------------------------------