It matters only in 2 cases. In the netbios context, a
duplicate name is
as bad as a duplicate IP address. In the DHCP informs
Dynamic DNS, I
think that only the last update will win the duplicate name
issue. In
the long run, if a name doesn't leak out of the box, then
the name is
private.. think RFC1918 private IP space. If there is
multiple use of
an IP, it is ok as long as the address is not leaked into
the public space.
With use of DDNS and the large predominance of the use of
SRV records to
promote services, I can see the name leaking out fairly
easily albeit in
error. How many people really turn off services they don't
need, like
file sharing, or print sharing?
I can see it easy on the LAN for situations to arise where
someone
broadcasts for printer service resolution, and his neighbor
says sure..
a printer is located at 'joes printer' localhost.localdomain.
I do have approximately 80,000 devices that are maintained
by people
whose qualifications to do so span the spectrum. That
said, I have not
really seen any of the above in the last 10 years.
Tim
Carl Karsten wrote:
> Thanks. Not quite as definitive as I had hoped for,
but good enough.
>
> Next Q: what are the implications of multiple machines
with the same hostname?
> In other words: does it matter if they are all left set
to localhost.localdomain?
>
> Carl K
>
> Tim Peiffer wrote:
>
>> A read of the manual page for dhcp-option(5) and
dhclient-script(5) is
>> useful. An excerpt from the man page says:
>> option host-name string;
>>
>> This option specifies the name of the
client. The name may
>> or may
>> not be qualified with the local domain
name (it is preferable
>> to use
>> the domain-name option to specify the
domain name). See
>> RFC 1035
>> for character set restrictions. This
option is only
>> honored by
>> dhclient-script(8) if the hostname for
the client machine
>> is not
>> set.
>>
>> The use of the host-name specified is optional and
depends upon how the
>> client is configured. Now the next logical
question, is how does this
>> work in Winderz?
>>
>>
>> Carl Karsten wrote:
>>
>>> host-name is optional, right?
>>>
>>> I am sure I know the answer, but can't find
support for it.
>>>
>>> This is for a olpc bugreport that goes like
this:
>>>
>>> cjb: doesn't DHCP set the hostname, though?
>>> CarlFK: it can, but from what I have seen dhcp
doesn't always happen
>>> CarlFK: and it isn't guaranteed
>>>
>>> cjb: if dhcp doesn't happen, we're not on a
network and don't care that we don't
>>> have a unique hostname.
>>>
>>> Carl K
>>>
>>>
>>
>>
|