Problem with dhcpd and ddns, deleting A record, leaving stale TXT/DHCID and PTR behind

Fri Mar 3 14:28:11 UTC 2017

The 4.3.5 dhcp server with bind 9.9.4 seems to work fine, but exhibits the
following behavior (ignore timezone missmatch)

Initial add works fine. Release works fine, but throws error on subsequent
renew. It still adds the record anyway shortly after

Mar  2 20:41:56 dhcp001 dhcpd: Removed forward map from jesstest100.domain.
to 10.1.71.150
Mar  2 20:41:56 dhcp001 dhcpd: Removed reverse map on
150.71.1.10.in-addr.arpa.
Mar  2 20:42:08 dhcp001 dhcpd: DHCPDISCOVER from 00:50:56:94:90:61 via eth0
Mar  2 20:42:09 dhcp001 dhcpd: DHCPOFFER on 10.1.71.150 to
00:50:56:94:90:61 (jesstest100) via eth0
Mar  2 20:42:09 dhcp001 dhcpd: DHCPREQUEST for 10.1.71.150 (10.1.71.135)
from 00:50:56:94:90:61 (jesstest100) via eth0
Mar  2 20:42:09 dhcp001 dhcpd: DHCPACK on 10.1.71.150 to 00:50:56:94:90:61
(jesstest100) via eth0
Mar  2 20:42:09 dhcp001 dhcpd: DHCPREQUEST for 10.1.71.150 (10.1.71.135)
from 00:50:56:94:90:61 (jesstest100) via 10.1.71.131
Mar  2 20:42:09 dhcp001 dhcpd: DHCPACK on 10.1.71.150 to 00:50:56:94:90:61
(jesstest100) via 10.1.71.131
Mar  2 20:42:09 dhcp001 dhcpd: Added new forward map from
jesstest100.domain. to 10.1.71.150
Mar  2 20:42:09 dhcp001 dhcpd: Added reverse map from
150.71.1.10.in-addr.arpa. to jesstest100.domain.

Mar  2 15:41:56 dns001 named[2209]: client 10.1.71.135#47886/key ddns:
updating zone 'domain.com/IN': deleting an RR at jesstest100.domain.com A
Mar  2 15:41:56 dns001 named[2209]: zone domain.com/IN: sending notifies
(serial 2017031863)
Mar  2 15:41:56 dns001 named[2209]: client 10.1.71.135#47886/key ddns:
signer "ddns" approved
Mar  2 15:41:56 dns001 named[2209]: client 10.1.71.135#47886/key ddns:
updating zone 'domain.com/IN': deleting an RR at jesstest100.domain.com
DHCID
Mar  2 15:42:01 dns001 named[2209]: zone domain.com/IN: sending notifies
(serial 2017031864)
Mar  2 15:42:09 dns001 named[2209]: client 10.1.71.135#47886/key ddns:
signer "ddns" approved
Mar  2 15:42:09 dns001 named[2209]: client 10.1.71.135#47886/key ddns:
updating zone 'domain.com/IN': adding an RR at 'jesstest100.domain.com' A
Mar  2 15:42:09 dns001 named[2209]: client 10.1.71.135#47886/key ddns:
updating zone 'domain.com/IN': adding an RR at 'jesstest100.domain.com'
DHCID
Mar  2 15:42:09 dns001 named[2209]: client 10.1.71.135#47886/key ddns:
updating zone 'domain.com/IN': update unsuccessful: jesstest100.domain.com:
'name not in use' prerequisite not satisfied (YXDOMAIN)
Mar  2 15:42:09 dns001 named[2209]: zone domain.com/IN: sending notifies
(serial 2017031865)
Mar  2 15:42:09 dns001 named[2209]: client 10.1.71.135#47886/key ddns:
signer "ddns" approved
Mar  2 15:42:09 dns001 named[2209]: client 10.1.71.135#47886/key ddns:
updating zone 'domain.com/IN': deleting rrset at 'jesstest100.domain.com' A
Mar  2 15:42:09 dns001 named[2209]: client 10.1.71.135#47886/key ddns:
updating zone 'domain.com/IN': adding an RR at 'jesstest100.domain.com' A

On Thu, Mar 2, 2017 at 10:03 AM, Jess Johnson <jester at 125gp.com> wrote:

> I built a test DNS server using 9.9.4 (whatever comes by default with
> centos7) and the dhcp server can correctly delete the TXT/DHCID record.
>
> My guess is dhcpd 4.2.6 and 4.3.5  has a bug talking to the older version
> 9.8.2.
>
> Jess
>
> On Wed, Mar 1, 2017 at 5:22 PM, Patrick Trapp <ptrapp at nex-tech.com> wrote:
>
>> The message came through successfully yesterday. I wish I had something
>> to offer.
>>
>>
>>
>> *From:* dhcp-users [mailto:dhcp-users-bounces at lists.isc.org] *On Behalf
>> Of *Jess Johnson
>> *Sent:* Wednesday, March 1, 2017 5:17 PM
>> *To:* dhcp-users at lists.isc.org
>> *Subject:* Problem with dhcpd and ddns, deleting A record, leaving stale
>> TXT/DHCID and PTR behind
>>
>>
>>
>> Apologies if multiples of these get through. I've been trying to send to
>> the list for a couple of days and not having any luck.
>>
>>
>>
>> Summary - dhcpd removes A record, but leaves DHCID and PTR behind,
>> causing problems in future.
>>
>>
>>
>> dhcpd 4.3.5 (custom install with no options to configure)
>>
>> bind-9.8.2-0.47.rc1.el6_8.3.x86_64
>>
>>
>>
>> ddns-updates on;
>>
>> ddns-update-style standard;
>>
>> ddns-rev-domainname "in-addr.arpa.";
>>
>> update-optimization off;
>>
>> #update-conflict-detection false;
>>
>>
>>
>> update-static-leases on;
>>
>> ignore client-updates;
>>
>> one-lease-per-client on;
>>
>>
>>
>> We are using host definitions as such
>>
>>
>>
>> host esxi-xxxx-a5-30-10 {
>>
>>   hardware ethernet 2C:59:E5:49:56:38;
>>
>>   ddns-hostname "esxi-xxxx-a5-30-10";
>>
>>   ddns-domainname "domain.com.";
>>
>>   option host-name "esxi-xxxx-a5-30-10";
>>
>>   option domain-name "domain.com";
>>
>> }
>>
>>
>>
>> The host boots on untagged network (no DDNS), PXE boots ESXi with tagged
>> interface on different VLAN with DDNS enabled. At this point it registers
>> A, PTR and DHCID just fine.
>>
>> Autodeploy then applies a host profile which cause the interface to
>> down/up on the same network segment. When it goes down, dhcp removes the A
>> record, but not the DHCID or PTR. When it comes back up on the same IP
>> address (MAC address stayed the same), I get
>>
>>
>>
>> Feb 28 19:16:22 dhcp001 dhcpd: DHCPOFFER on 10.1.66.37 to
>> 2c:59:e5:49:15:98 via 10.1.66.2
>>
>> Feb 28 19:16:22 dhcp001 dhcpd: DHCPREQUEST for 10.1.66.37 (10.1.71.135)
>> from 2c:59:e5:49:15:98 via 10.1.66.2
>>
>> Feb 28 19:16:22 dhcp001 dhcpd: DHCPACK on 10.1.66.37 to 2c:59:e5:49:15:98
>> via 10.1.66.2
>>
>> Feb 28 19:16:22 dhcp001 dhcpd: Added new forward map from
>> esxi-xxxx-a5-30-14.domain.com. to 10.1.66.37
>>
>> Feb 28 19:18:58 dhcp001 dhcpd: DHCPRELEASE of 10.1.66.37 from
>> 2c:59:e5:49:15:98 via eth0 (found)
>>
>> Feb 28 19:18:58 dhcp001 dhcpd: Removed forward map from
>> esxi-xxxx-a5-30-14.domain.com. to 10.1.66.37
>>
>> Feb 28 19:19:01 dhcp001 dhcpd: DHCPOFFER on 10.1.66.37 to
>> 2c:59:e5:49:15:98 via 10.1.66.3
>>
>> Feb 28 19:19:01 dhcp001 dhcpd: DHCPREQUEST for 10.1.66.37 (10.1.71.135)
>> from 2c:59:e5:49:15:98 via 10.1.66.3
>>
>> Feb 28 19:19:01 dhcp001 dhcpd: DHCPACK on 10.1.66.37 to 2c:59:e5:49:15:98
>> via 10.1.66.3
>>
>> Feb 28 19:19:04 dhcp001 dhcpd: DHCPREQUEST for 10.1.66.37 from
>> 2c:59:e5:49:15:98 via eth0
>>
>> Feb 28 19:31:52 dhcp001 dhcpd: Forward map from esxi-xxxx-a5-30-14.domain
>> .com. to 10.1.66.37 FAILED: timed out
>>
>> and
>>
>> updating zone 'domain.com/IN': update unsuccessful:
>> esxi-xxxx-a5-30-10.domaincom: 'name not in use' prerequisite not satisfied
>> (YXDOMAIN)
>>
>>
>>
>> # host -a esxi-xxxx-a5-30-10.domain.com
>>
>> Trying "esxi-xxxx-a5-30-10.domain.com"
>>
>> ;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 42622
>>
>> ;; flags: qr aa rd ra; QUERY: 1, ANSWER: 1, AUTHORITY: 3, ADDITIONAL: 3
>>
>>
>>
>> ;; QUESTION SECTION:
>>
>> ;esxi-xxxx-a5-30-10.domain.com. IN ANY
>>
>>
>>
>> ;; ANSWER SECTION:
>>
>> esxi-xxxx-a5-30-10.domain.com. 3600 IN DHCID
>> AAEBABsJWn8ZMp8B0wG1h6pv4FVrVZobDzBUVM8QVYrQPak=
>>
>>
>>
>> ;; AUTHORITY SECTION:
>>
>> domain.com.  86400   IN      NS      dns002.domain.com.
>>
>> domain.com.  86400   IN      NS      dns003.domain.com.
>>
>> domain.com.  86400   IN      NS      dns001.domain.com.
>>
>>
>>
>>
>>
>> It's my understanding it should be deleting the DHCID record because
>> there is no A record? The esxi server still has a valid lease on 10.1.66.34
>>
>>
>>
>> What am I missing?
>>
>> _______________________________________________
>> dhcp-users mailing list
>> dhcp-users at lists.isc.org
>> https://lists.isc.org/mailman/listinfo/dhcp-users
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/dhcp-users/attachments/20170303/ba6be17c/attachment.html>