Bind locking up

[mediamacros]

We are having an odd problem with Bind on our server. Everythign runs along fine and then after a day or two you can't access sites. Log into the control pannel and restart hte bind service and whamo, it works. The problem is, I can't find a reason why its locking. Is there any type of common isses that might cause this?

[QT]

There are a few things that would cause BIND to lock up. It could range from a DoS attack to just using a lot of memory.

Check /var/log/messages for errors from named and check how much memory and processor time it is using (top or ps).

[mediamacros]

I have this in the DNS logs over and over again. Anyone have the English translation for what this is?

The DNS server encountered a packet addressed to itself -- IP address my.ip.here
The DNS server should never be sending a packet to itself. This situation usually indicates a configuration error.

Check the following areas for possible self-send configuration errors:
1) Forwarders list. (DNS servers should not forward to themselves).
2) Master lists of secondary zones.
3) Notify lists of primary zones.
4) Delegations of subzones. Must not contain NS record for this DNS server unless subzone is also on this server.

Example of self-delegation:
-> This DNS server dns1.foo.com is the primary for the zone foo.com.
-> The foo.com zone contains a delegation of bar.foo.com to dns1.foo.com,
(bar.foo.com NS dns1.foo.com)
-> BUT the bar.foo.com zone is NOT on this server.

Note, you should make this delegation check (with nslookup or DNS manager) both on this DNS server and on the server(s) you delegated the subzone to. It is possible that the delegation was done correctly, but that the primary DNS for the subzone, has any incorrect NS record pointing back at this server. If this incorrect NS record is cached at this server, then the self-send could result. If found, the subzone DNS server admin should remove the offending NS record.

[mediamacros]

Also this is in the application log...

c:\winnt\system32\dns\etc/domain.com.hosts: WARNING SOA expire value is less than 7 days (86400)

What does this mean? I thought webappliance (Ensim) was suposed to properly configure this itself. :(

[QT]

Originally posted by mediamacros
I have this in the DNS logs over and over again. Anyone have the English translation for what this is?

The DNS server encountered a packet addressed to itself -- IP address my.ip.here
The DNS server should never be sending a packet to itself. This situation usually indicates a configuration error.

Check the following areas for possible self-send configuration errors:
1) Forwarders list. (DNS servers should not forward to themselves).
2) Master lists of secondary zones.
3) Notify lists of primary zones.
4) Delegations of subzones. Must not contain NS record for this DNS server unless subzone is also on this server.

Example of self-delegation:
-> This DNS server dns1.foo.com is the primary for the zone foo.com.
-> The foo.com zone contains a delegation of bar.foo.com to dns1.foo.com,
(bar.foo.com NS dns1.foo.com)
-> BUT the bar.foo.com zone is NOT on this server.

Note, you should make this delegation check (with nslookup or DNS manager) both on this DNS server and on the server(s) you delegated the subzone to. It is possible that the delegation was done correctly, but that the primary DNS for the subzone, has any incorrect NS record pointing back at this server. If this incorrect NS record is cached at this server, then the self-send could result. If found, the subzone DNS server admin should remove the offending NS record.

If you have a subdomain (foo.domain.com) delegated to another nameserver, you need to make sure there are NS records in domain.com for foo.

foo IN NS ns1.foo.domain.com.
foo IN NS ns2.foo.domain.com.

[mediamacros]

I only have one domain with a sub domain and they are on the same box. Here is my setup...


domain.com is in the Server Beach DNS as a slave
domain.com is on my server and I have A records for www, mail, ftp, "blank", and subdomain for that domain

subdomain.domain.com is a different site but IIS handles that part and it appears to work fine. I can access the site with no problems, etc.

Shoudl it have its own zone instead? COuld htis be causing my Bind lockups?

[QT]

It looks like your NS records are incorrect.
If you're using GeoDNS for slave, then you need to have these added:

domain.com. IN NS ns1.geodns.net.
domain.com. IN NS ns2.geodns.net.

In addition to your own nameserver.

[mediamacros]

I have that as well. :) Thanks. I will check for any correctable errors, clear the server logs and watch it like a hawk to see if it happens again. Thanks.