svn commit: r185435 - in head: lib/libc/sys lib/libkvm share/man/man4 sys/compat/freebsd32 sys/kern sys/net sys/netinet sys/netinet6 sys/security/mac_bsdextended sys/sys usr.bin/cpuset usr.sbin/jai...

[Julian Elischer]

James Gritton wrote:
> Bjoern A. Zeeb wrote:
>> On Thu, 4 Dec 2008, Dag-Erling Smørgrav wrote:
>>> "Bjoern A. Zeeb" <bz at FreeBSD.org> writes:
>>>>   Jails can have an unrestricted (no duplicate protection, etc.) name
>>>>   in addition to the hostname. The jail name cannot be changed from
>>>>   within a jail and is considered to be used for management purposes
>>>>   or as audit-token in the future.
>>>
>>> Is there a specific reason why we allow duplicates?
>>
>> Yes, so that people have a choice.
>>
>> You would have to take into account "DYING" jails with possibly
>> duplicate names anyway.
>>
>> And it's a purely administrative name (cannot be changed from within
>> the jail) so it's up to the administrator to have 2 "Customer Webserver"
>> jails or a "Customer Webserver Tin" and a "Customer Webserver Box".
>>
>> BTW. in case people wonder how to set it (atm.) from rc.conf, you
>> can use jail_<name>_flags.
> 
> My new (upcoming) jail framework patches also support names, but without 
> allowing duplicates.  This is necessary because jails can be looked up 
> in the kernel by name as well as by ID.  I handle the special case of 
> "dying" jails, allowing a new jail to duplicate the name of one on the 
> the way out.
> 
> As far as people having a choice, I don't see any more reason to offer 
> the choice of duplicate jail names than I do in offering duplicate login 
> names.  It just doesn't make sense in the context of what purpose I 
> expect a name to serve.

hierarchical names make this even less sensible,

> 
> - Jamie