Will abandoned SOA records eventually be deleted automatically?
Simon at wretched.demon.co.uk
Wed Oct 9 10:52:23 UTC 2002
Cameron Pitcairn wrote:
> On Monday we switched mitchellcenter.info to a new host. To accomplish
> this, we changed the nameservers listed by the registrar (VeriSign). The
> original nameservers were:
> The new nameservers are:
These servers have differing NS entries currently. You might
like to make sure everything is okay.
11:48:43 srw doc-2.2.3 $ dig +short @ns1.infoquesthosting.net
mitchellcenter.info ns ns1.infoquesthosting.net.
11:48:48 srw doc-2.2.3 $ dig +short @ns2.infoquesthosting.net
mitchellcenter.info ns ns2.infoquesthosting.net.
11:49:36 srw doc-2.2.3 $ dig +short @ns1.infoquesthosting.net
ns1.infoquesthosting.net. admin.infoquesthosting.net. 5 900 600
11:49:42 srw doc-2.2.3 $ dig +short @ns2.infoquesthosting.net
ns2.infoquesthosting.net. admin.infoquesthosting.net. 8 900 600
Hmm ns2 is getting ahead of ns1
> By now, the change has propagated and most everyone who tries to go to
> www.mitchellcenter.info goes to the new site, EXCEPT for the woman
> responsible for maintaining the site. She gets sent to the old site. On
> investigation, it turns out that her primary DNS server is -- NS1.NETAXS.COM
> I conjecture (and some playing around with nslookup seems to confirm) that
> since this particular server considers itself authoritative for
> mitchellcenter.info, it never queries a root server when queried about that
> domain, and consequently does not know about the change in nameservers.
> 1. Is my conjecture correct?
Yes, it is easier using "dig".
> 2. If so, will the NETAXS server eventually "figure out" that it is no
> longer authoritative for mitchellcenter.info, or should we make a specific
> request to Netaxs to delete the SOA records? (We want to make sure
> everything is working correctly before actually closing our account with
You'll need to request it.
If companies used seperate servers for recursive resolution, and
authoritative, then this would never happen.
More information about the bind-users