ISC-DHCP and BIND 9 DNS: DDNS update fails for /27 subnet P.S.
Mirsad Goran Todorovac
mirsad.todorovac at alu.unizg.hr
Sun Dec 12 14:45:14 UTC 2021
Hello Crist,
I have implemented the recommended changes. It works forward and reverse
for the test record, from out domain or others, or for almost all of the
test records.
There are still some spurious failures, such as this one:
200 IN CNAME 200.186.198.193.dhcp.slava.alu.hr.
201 IN CNAME 201.186.198.193.dhcp.slava.alu.hr.
nslookup 193.198.186.200 works and .201 doesn't, despite the symmetric
definition:
root at domac:/etc/bind/zones# nslookup 193.198.186.200
200.186.198.193.in-addr.arpa canonical name =
200.192/27.186.198.193.in-addr.arpa.
200.192/27.186.198.193.in-addr.arpa canonical name =
200.186.198.193.dhcp.slava.alu.hr.
200.186.198.193.dhcp.slava.alu.hr name = test-record1.slava.alu.hr.
Authoritative answers can be found from:
186.198.193.dhcp.slava.alu.hr nameserver = domac.alu.hr.
domac.alu.hr internet address = 161.53.235.3
root at domac:/etc/bind/zones# nslookup 193.198.186.201
** server can't find 201.186.198.193.in-addr.arpa: NXDOMAIN
root at domac:/etc/bind/zones#
I can't get to the bottom of this, I don't know enough BIND9 internals.
It will take real-life production load tomorrow to see how this will
behave with DHCP DDNS updates. :-)
You said ABSOLUTELY NO WARRANTY but I am an open source fan and I can
live with that ;-)
Until tomorrow, then ...
Kind regards,
Mirsad Todorovac
On 12/12/2021 10:33 AM, Mirsad Goran Todorovac wrote:
>
> Hi Crist,
>
> Now the resolution from the problematic record started working again
> without any change in zones or BIND9 options, also without the server
> process restart ... :-/
>
> root at domac:~# nslookup -query=any 195.192/27.186.198.193.in-addr.arpa.
> Server: 161.53.235.3
> Address: 161.53.235.3#53
>
> 195.192/27.186.198.193.in-addr.arpa name = test-record.slava.alu.hr.
>
> root at domac:~# nslookup -query=ptr 193.198.186.195
> Server: 161.53.235.3
> Address: 161.53.235.3#53
>
> Non-authoritative answer:
> 195.186.198.193.in-addr.arpa canonical name =
> 195.192/27.186.198.193.in-addr.arpa.
> 195.192/27.186.198.193.in-addr.arpa name = test-record.slava.alu.hr.
>
> Authoritative answers can be found from:
> 192/27.186.198.193.in-addr.arpa nameserver = domac.alu.hr.
> domac.alu.hr internet address = 161.53.235.3
>
> root at domac:~#
>
> I guess this was something with timeouts. Suppose this will work
> satisfactory on desktops that usually keep the same IP address
> assigned by DHCP across the lease renewals, but not for laptops,
> Android and iPhone devices that connect and disconnect, and change
> network ...
>
> Why I want smartphones to have reverse PTRs is to see in logs if
> something becomes virus infected or even spambot, and not have to
> browse DHCP leases in forensic analysis, which my fellow administrator
> probably would not know how to do ...
>
> Kind regards,
> Mirsad Todorovac
>
> On 12/12/2021 10:19 AM, Mirsad Goran Todorovac wrote:
>>
>> Hi Crist,
>>
>> Thank you for your explanation. It was much appreciated.
>> However, as I previously asserted, it is impossible to know how the
>> system will behave without testing it with real life production load
>> on Monday :-)
>>
>> On 12/11/2021 11:18 PM, Crist Clark wrote:
>>> Looks like you're trying to use the setup in that serverfault link.
>>> That example only works on an internal network.
>> I thought the 186.198.193. part was enough to make the zone unique.
>> But your assertion is correct: I would collide if any other
>> administrators on other subnets in range 193.198.186.0/24 decide to
>> make reverse DHCP DDNS update in the future. Thanks for the thought!
>>> The point of the example I gave is that you are going to build your
>>> own reverse zone inside of a zone you control on the Internet. Now
>>> that you've given some examples, I can perhaps make it more obvious
>>> what I'm suggesting. Your DNS zone would look something like,
>>>
>>>> $ORIGIN 192-27.186.198.193.in-addr.arpa.
>>>>
>>>> @ IN NS domac.alu.hr <http://domac.alu.hr/>.
>>>> @ IN NS bjesomar.srce.hr <http://bjesomar.srce.hr/>.
>>>>
>>>> 195 IN PTR test-record.dhcp.slava.alu.hr
>>>> <http://test-record.slava.alu.hr/>.
>>>>
>>>> $GENERATE 200-222 $ CNAME $.186.198.193.dhcp.slava.alu.hr
>>>> <http://186.198.193.dhcp.slava.alu.hr>.
>>>>
>>>
>>> And your DHCP configuration,
>>>>
>>>> ddns-domainname "slava.alu.hr <http://slava.alu.hr/>";
>>>> ddns-rev-domainname "dhcp.slaval.alu.hr <http://dhcp.slaval.alu.hr>";
>>>> zone slava.alu.hr <http://slava.alu.hr/>. {
>>>> primary 127.0.0.1;
>>>> key DDNS_UPDATE;
>>>> }
>>>>
>>> NOT TESTED. NO GUARANTEES. NOT SUITABLE FOR ANY GIVEN PURPOSE. YOUR
>>> MILEAGE MAY VARY. PLEASE CONSULT YOUR PERSONAL PHYSICIAN BEFORE
>>> STARTING ISC PRODUCTS.
>> Noted. :-) I am not afraid of experimenting. But failures of the
>> experimental setup are perceived as my incompetence, and success
>> taken for granted the very next day ;-)
>>> One other odd thing, sometimes you refer to a
>>> "192/27.186.198.193.in-addr.arpa" zone and sometimes
>>> "192-27.186.198.193.in-addr.arpa." Those are different names and are
>>> not interchangeable. Both are totally fine for use in DNS, but a lot
>>> of administrators don't like the '/' in zone names since they often
>>> use the zone name in file names. Slashes present a problem in
>>> file names on *nix flavored OSes. I used the dash, '-', version in
>>> my example.
>>
>> This setup is mandated from the upper level sysadmins and I cannot
>> control it, I can only beg them to use a hyphen as in RFC 2317
>> chapter 4 last paragraph, but I cannot guarantee that they will
>> change it. It is their arbitrary decision. :-/
>>
>> Frankly, /27 is more readable, but if it creates havoc in Linux
>> resolver, then what the heck ...
>>
>> Thank you very much again for your advice. I will post back here on
>> the results with your recommended zone setup.
>>
>> Kind regards,
>> Mirsad Todorovac
>>
>>> Hi again,
>>>
>>> I had some luck in making this setup work. So far, so good ...
>>> However, there's no telling how the DHCP DDNS will function with
>>> the new 186.198.193.dhcp. zone before Monday morning when the
>>> subsidiary computers power up.
>>>
>>> However, I have an odd behavior which I cannot explain: without
>>> any change to zone a reverse resolution stopped working. The
>>> setup just doesn't seem stable enough to work with DHCP-updated
>>> dynamic DNS in our organization, with a lot of smartphones and
>>> wireless devices frequently signing on and off.
>>>
>>> The zone is:
>>>
>>> $ORIGIN 192/27.186.198.193.in-addr.arpa.
>>>
>>> @ IN NS domac.alu.hr <http://domac.alu.hr>.
>>> ;@ IN NS bjesomar.srce.hr <http://bjesomar.srce.hr>.
>>>
>>> 195 IN PTR test-record.slava.alu.hr
>>> <http://test-record.slava.alu.hr>.
>>>
>>> 200 IN CNAME 200.186.198.193.dhcp.
>>> 201 IN CNAME 201.186.198.193.dhcp.
>>>
>>> ; MT 20211211:
>>> ; Here's the magic:
>>>
>>> $GENERATE 202-222 $ CNAME $.186.198.193.dhcp.
>>>
>>> The command output shows that resolution succeeds, but nslookup
>>> can't finish it for some unknown spurious reason.
>>>
>>> root at domac:~# nslookup -query=any
>>> 195.192/27.186.198.193.in-addr.arpa.
>>> Server: 161.53.235.3
>>> Address: 161.53.235.3#53
>>>
>>> 195.192/27.186.198.193.in-addr.arpa name =
>>> test-record.slava.alu.hr <http://test-record.slava.alu.hr>.
>>>
>>> root at domac:~# nslookup -query=ptr 193.198.186.195
>>> Server: 161.53.235.3
>>> Address: 161.53.235.3#53
>>>
>>> ** server can't find 195.186.198.193.in-addr.arpa: NXDOMAIN
>>>
>>> root at domac:~#
>>>
>>> This kind of setup that sometimes works and sometimes doesn't
>>> will make me look incompetent.
>>> I know that BIND 9 is great open source server with lots of
>>> bells and whistles. But right now I can't study all those and I
>>> just want to survive, providing a solution fast enough for our
>>> uplevel sysadmins.
>>>
>>> The /etc/bind/named.conf.local part looks like:
>>>
>>> zone "192/27.186.198.193.in-addr.arpa" in {
>>> type master;
>>> file "/etc/bind/zones/192-27.186.198.193.in-addr.arpa.db";
>>> };
>>>
>>> zone "186.198.193.dhcp" in {
>>> type master;
>>> file "/var/cache/bind/186.198.193.dhcp.db";
>>> allow-update { key DDNS_UPDATE; };
>>> };
>>>
>>> What possibly could be killing the name resolution between
>>> resolving 195.192/27.186.198.193.in-addr.arpa to
>>> test-record.slava.alu.hr <http://test-record.slava.alu.hr>. and
>>> resolving 193.198.186.195 that apparently fails?
>>>
>>> Is there a way to see more interim debugging output?
>>>
>>> Thank you very much.
>>>
>>> Kind regards,
>>> Mirsad Todorovac
>>>
>>>
>>>
>>> On 12/11/2021 10:25 AM, Mirsad Goran Todorovac wrote:
>>>>
>>>> Hi Crist,
>>>>
>>>> Thank you for your reply and the information provided.
>>>>
>>>> I have roughly implemented this workaround. I was hoping there
>>>> was a way to instruct BIND to masquerade a delegated domain
>>>> with data from another (dynamically updated from ISC DHCP) zone.
>>>>
>>>> More accurately, my (from upper level) mandated delegation is
>>>> the literal 192/27.186.198.193.in-addr.arpa, using
>>>> 192-27.186.198.193.in-addr.arpa says "ignoring records outside
>>>> of the origin" or something like that.
>>>>
>>>> I have used the following records in the zone:
>>>>
>>>> $ORIGIN 192/27.186.198.193.in-addr.arpa.
>>>>
>>>> @ IN NS domac.alu.hr <http://domac.alu.hr>.
>>>> @ IN NS bjesomar.srce.hr <http://bjesomar.srce.hr>.
>>>>
>>>> 195 IN PTR test-record.slava.alu.hr
>>>> <http://test-record.slava.alu.hr>.
>>>>
>>>> $GENERATE 200-222 $ CNAME $.186.198.193.dhcp.
>>>>
>>>> /etc/dhcp/dhcpd.conf has this:
>>>>
>>>> ddns-domainname "slava.alu.hr <http://slava.alu.hr>";
>>>> ddns-rev-domainname "dhcp";
>>>> zone slava.alu.hr <http://slava.alu.hr>. {
>>>> primary 127.0.0.1;
>>>> key DDNS_UPDATE;
>>>> }
>>>> zone 186.198.193.dhcp. {
>>>> primary 127.0.0.1;
>>>> key DDNS_UPDATE;
>>>> }
>>>>
>>>> However, don't I have to convince people managing
>>>> bjesomar.srce.hr <http://bjesomar.srce.hr> to be a slave server
>>>> for the "186.198.193.dhcp" zone? Or the dynamically updated
>>>> reverse PTR record will have effect only in my local domain
>>>> (which I had even before the entire setup), won't it?
>>>>
>>>> Also, I get spurious REFUSED or NXDOMAIN errors, some pass with
>>>> time, so there must be some TTL or timeout.
>>>>
>>>> Kind regards,
>>>>
>>>> Mirsad
>>>>
>>>> On 12/11/2021 6:04 AM, Crist Clark wrote:
>>>>> No idea if this is the best way. It is a way.
>>>>>
>>>>> Do you control any other zone? Let’s say you own
>>>>> “example.com.” You can tell ISC DHCP to build the reverse zone
>>>>> at an arbitrary base name instead of in-addr.arpa.
>>>>>
>>>>> Configure DHCP to put the reverse records at say,
>>>>> “rev.example.com.” So you’ll get records at,
>>>>>
>>>>> 193.186.198.193.rev.example.com
>>>>> <http://193.186.198.193.rev.example.com>
>>>>> 194.186.198.193.rev.example.com
>>>>> <http://194.186.198.193.rev.example.com>
>>>>> …
>>>>>
>>>>> And in your RFC 2317-style delegation, you then enumerate
>>>>> another CNAME layer,
>>>>>
>>>>> $ORIGIN 192-27.186.198.193.in-addr.arpa.
>>>>> 193 IN CNAME 193.186.198.193.rev.example.com
>>>>> <http://193.186.198.193.rev.example.com>.
>>>>> 194 IN CNAME 194.186.198.193.rev.example.com
>>>>> <http://194.186.198.193.rev.example.com>.
>>>>> …
>>>>>
>>>>> On Fri, Dec 10, 2021 at 2:51 PM Mirsad Goran Todorovac
>>>>> <mirsad.todorovac at alu.unizg.hr> wrote:
>>>>>
>>>>> Hello,
>>>>>
>>>>> I have a problem with DHCP DDNS update to BIND 9 reverse
>>>>> PTR zone subnet that is owned by several organizations, so
>>>>> I can't get a direct DHCP DDNS update access with a key or
>>>>> with hostname.
>>>>>
>>>>> I have been delegated domain name
>>>>> |192-27.186.198.193.in-addr.arpa from the upper level
>>>>> admins, and that appears to be immutable.|
>>>>>
>>>>> |However, my subnet is 193.198.186.192/27
>>>>> <http://193.198.186.192/27>, and DHCP only knows how to
>>>>> perform DDNS update to 186.198.193.in-addr.arpa. (See
>>>>> here:
>>>>> https://serverfault.com/questions/806875/how-to-tell-isc-dhcp-correct-zone-for-reverse-zone-ddns-update
>>>>> and here:
>>>>> https://lists.isc.org/mailman/htdig/dhcp-users/2006-August/001422.html
>>>>> ).
>>>>> |
>>>>>
>>>>> |(This setup is because we have DHCP addresses that are
>>>>> not over NAT, but /24 subnet is shared with other
>>>>> organizations, even under another Minstry.)|
>>>>>
>>>>> |I want to have the effect of delegating the same database
>>>>> to upper level under their zone name, while updating the
>>>>> same database under my DHCP-understood zone name.|
>>>>>
>>>>> |I tried this /etc/bind/named.conf.local:|
>>>>>
>>>>> |zone "192-27.186.198.193.in-addr.arpa" in { type master;
>>>>> file "/var/cache/bind/192-27.186.198.193.in-addr.arpa.db";
>>>>> }; zone "186.198.193.in-addr.arpa" in { type master; file
>>>>> "/var/cache/bind/192-27.186.198.193.in-addr.arpa.db";
>>>>> allow-update { key DDNS_UPDATE; }; }; |
>>>>>
>>>>> (Two zones with the same file.)
>>>>>
>>>>> What I got was:
>>>>>
>>>>> |root at domac:/etc/bind# named-checkconf
>>>>> /etc/bind/named.conf.local:49: writeable file
>>>>> '/var/cache/bind/192-27.186.198.193.in-addr.arpa.db':
>>>>> already in use: /etc/bind/named.conf.local:44
>>>>> root at domac:/etc/bind# Can you please tell me is there a
>>>>> way to achieve the effect of the above (illegal) setup? I
>>>>> can't change DHCP nor I know an option to tell it to
>>>>> accept update to |||192-27.186.198.193.in-addr.arpa| (it is a syntax error).
>>>>> The DHCP dhcpd.conf subnet configuration is: |||subnet 193.198.186.192 netmask 255.255.255.224 { range
>>>>> 193.198.186.200 193.198.186.222; # MT 20211210 option
>>>>> subnet-mask 255.255.255.224; option domain-name-servers
>>>>> 161.53.235.3, 161.53.2.70; option domain-name
>>>>> "slava.alu.hr <http://slava.alu.hr>"; ddns-domainname
>>>>> "slava.alu.hr <http://slava.alu.hr>"; zone slava.alu.hr
>>>>> <http://slava.alu.hr>. { primary 127.0.0.1; key
>>>>> DDNS_UPDATE; } zone 186.198.193.in-addr.arpa. { primary
>>>>> 127.0.0.1; key DDNS_UPDATE; } option broadcast-address
>>>>> 193.198.186.223; option routers 193.198.186.193;
>>>>> default-lease-time 43200; max-lease-time 86400; } | Thank
>>>>> you very much for your time reading this mail and help.
>>>>> Kind regards, -- Mirsad Goran Todorovac Academy of Fine
>>>>> Arts | Faculty of Graphic Arts University of Zagreb |
>>>>>
>>>>> _______________________________________________
>>>>> Please visit
>>>>> https://lists.isc.org/mailman/listinfo/bind-users to
>>>>> unsubscribe from this list
>>>>>
>>>>> ISC funds the development of this software with paid
>>>>> support subscriptions. Contact us at
>>>>> https://www.isc.org/contact/ for more information.
>>>>>
>>>>>
>>>>> bind-users mailing list
>>>>> bind-users at lists.isc.org
>>>>> https://lists.isc.org/mailman/listinfo/bind-users
>>>>>
>>>>
>>>> _______________________________________________
>>>> Please visithttps://lists.isc.org/mailman/listinfo/bind-users to unsubscribe from this list
>>>>
>>>> ISC funds the development of this software with paid support subscriptions. Contact us athttps://www.isc.org/contact/ for more information.
>>>>
>>>>
>>>> bind-users mailing list
>>>> bind-users at lists.isc.org
>>>> https://lists.isc.org/mailman/listinfo/bind-users
>>> _______________________________________________
>>> Please visit https://lists.isc.org/mailman/listinfo/bind-users
>>> to unsubscribe from this list
>>>
>>> ISC funds the development of this software with paid support
>>> subscriptions. Contact us at https://www.isc.org/contact/ for
>>> more information.
>>>
>>>
>>> bind-users mailing list
>>> bind-users at lists.isc.org
>>> https://lists.isc.org/mailman/listinfo/bind-users
>>>
>
> _______________________________________________
> Please visithttps://lists.isc.org/mailman/listinfo/bind-users to unsubscribe from this list
>
> ISC funds the development of this software with paid support subscriptions. Contact us athttps://www.isc.org/contact/ for more information.
>
>
> bind-users mailing list
> bind-users at lists.isc.org
> https://lists.isc.org/mailman/listinfo/bind-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/bind-users/attachments/20211212/76c2990a/attachment-0001.htm>
More information about the bind-users
mailing list