Web lists-archives.com

Re: Ethernet is not started at boot

On 07/02/18 14:29, Brian wrote:
> On Tue 06 Feb 2018 at 20:23:13 -0500, Michael Stone wrote:
>> On Wed, Feb 07, 2018 at 01:16:28AM +0000, Brian wrote:
>>> On Tue 06 Feb 2018 at 19:53:16 -0500, Michael Stone wrote:
>>>> On Tue, Feb 06, 2018 at 11:00:12PM +0000, Brian wrote:
>>>>> On Tue 06 Feb 2018 at 17:06:36 -0500, Gene Heskett wrote:
>>>>>> On Tuesday 06 February 2018 14:07:55 Brian wrote:
>>>>>>> 1. auto enp0s25
>>>>>>>                                                  iface enp0s25 inet
>>>>>>> static address
>>>>>>>         netmask
>>>>>>>         gateway
>>>>>>>         network
>>>>>>> netmask and network are not needed. ifupdown will compute them. Note
>>>>>>> there are no examples in interfaces(5) which use these parameters.
>>>>>> You should go and read that man page again.
>>>>> I didn't have to but I did. A line beginning "netmask" is not required".
>>>> If you don't specify a netmask it will default to something that was correct
>>>> in the 80s and is probably wrong on your network. The "netmask line" isn't
>>>> required, but if you don't have a netmask line you should specify the
>>>> netmask in the address line (using "ip/netmask" syntax).
>>> Indeed. But a netmask line (while not necessary) doesn't do any harm.
>> More than that: either a separate netmask entry or a netmask specified as
>> part of the address are almost certainly required for proper operation. It's
>> not technically required, but it is practically required.
> The changelog says:
>   * Calculate netmask internally, so even if a user haven't supplied
>     one or have used CIDR notation, hook scripts will have it properly
>     specified in IF_NETMASK environment variable.
> What is the significance of "...so even if a user haven't supplied ..."?

Calculate how? Using ancient classful addressing rules? Sounds like that
change is a bug, to me. Lack of a mask or prefix length should trigger
an error.


Attachment: signature.asc
Description: OpenPGP digital signature