dynamic update problems

[Kevin Darcy]

I'm really not sure why you're trying to use BIND this way, as opposed to
using an SQL-based database (e.g. MySQL or a commerical RDBMS) or an
LDAP-based directory.

If you *insist* on using BIND like this, though,

a) you may be able to get away with just specifying "fetch-glue no" in your
"options". Looks like the server choked on trying to fill in the address of
"host.xxx.yyy.edu" in the Additional section of a response. "fetch-glue no"
stops it from trying to fill in that data. But, it may still want to
resolve that name as part of the Dynamic Update process, so...

b) failing that, you'd need to arrange some other way for the server to be
able to resolve the "host.xxx.yyy.edu" name, even if all of your clients
are local and using 127.0.0.1. Since your nameserver has no connection to
the regular DNS namespace, there's no way that it can simply "know" that
"host.xxx.yyy.edu" resolves to its own IP address. To house the A record
for "host.xxx.yyy.edu", you could define "host.xxx.yyy.edu", "xxx.yyy.edu",
"yyy.edu" or just "edu" as a zone in your server, but my recommendation
would be to just simplify everything by throwing *all* of your names into a
single root zone -- change your zone's name to just "." -- and define the
"host.xxx.yyy.edu" A record there.


- Kevin
Suresh Thesayi wrote:

> Hello,
>
>    I am quite a newbie to DNS/BIND and I have a problem that I could not
> figure out. Probably I have configured it wrong. I am using BIND as a
> database server for an application. The primary purpose is to run and
> maintain a database of records (Telephone addresses and name - telephone
> translation). So, I don't want BIND to do anything at all for me with
> respect to Name resolution and all. All I want it to do is for it to
> answer database queries and do database updates. It does the former
> correctely but failes to update the records dynamically. I have made it
> to listen on a different port and also changed the port number the
> clients should send queries to. These are the messages I get while
> sending update requests for my records. In the client side I get the
> following error
>
> res_update: send error for ds.extdistrib5.itx.cnrg
> failed update: Connection timed out
>
> In the server side I get the following error message:
> do additional "host.xxx.yyy.edu" (from "extdistrib5.itx.cnrg")
> found it
> sysquery(host.xxx.yyy.edu, 1, 1, 0, 0)
> qnew(0xa04c0)
> findns: np 0x92c14 'host'
> findns: np 0x92bf8 'xxx'
> findns: np 0x92bdc 'yyy'
> findns: np 0x92bc0 'edu'
> np_parent(0x0) couldn't find root entry
> findns: No root nameservers for class IN?
> sysquery: findns error (SERVFAIL) on host.xxx.yyy.edu?
> ns_freeqry(0xa14c0)
>
> I guess the problem is that the forwarding has failed or mis configured.
> What I want is for it not to forward at all and also for simplicity I
> will be using only the primary server (JUST 1 STANDALONE).
>
> CAN ANYONE HELP ME?? Please suggest what I should put in my config and
> zone files to make this possible. Especially my zone files. What I
> currently have is (Example for zone ext2distrib5.itx.cnrg)
>
> Zone file db.extdistrib5:
> $ORIGIN itx.cnrg.
> extdistrib5     IN      SOA     host.xxx.yyy.edu.  userid.xxx.yyy.edu. (
>
>                 4       ; Serial
>                 10800   ; Refresh after 3 hours
>                 3600    ; Retry after 1 hour
>                 604800  ; Expire after 1 week
>                 86400 ) ; Minimum TTL of 1 day
>                 IN      NS      host.xxx.yyy.edu.
> $ORIGIN extdistrib5.itx.cnrg.
> ds      0       IN      TXT     "10012"
> ds      0       IN      TXT     "10015"
>
> Corresponding entry in my named.conf file is as follows
> named.conf:
> zone "extdistrib5.itx.cnrg" in {
>    type master;
>    file "../itx_ds/db.extdistrib5";
>    allow-update {any; };
> };
>
> I have not included named.root file as hint file. Probably thats why it
> could not find any root servers, I guess. But I don't want to. Can you
> suggest what I should do.
>
> Thanks in advance
> Suresh