Re: nscd for freebsd

Hi, list

Some time ago we are discuss about renaming 'cached' to 'nscd':
http://docs.freebsd.org/cgi/getmsg.cgi?fetch=338214+0+archive/2006/freebsd-current/20061119.freebsd-current

Could somebody talk me current situation with this (especially I'm 
interesting plans for MFC).

Michael Bushkov wrote:
> On Monday 13 November 2006 14:58, Dmitriy Kirhlarov wrote:
>> I'm interested in name service caching daemon for freebsd (it's useful
>> for external userbase, stored in ldap and for monitoring hosts, with
>> resolving/polling of many hosts). It's much more useful, then local
>> cached named and local ldap replica.
>>
>> I found "cached" daemon in HEAD.
>>
>> Can somebody answer me -- when this daemon will be MFC-ed to RELENG_6?
>> Also why this strange name?
>> It's impossible to find this name in google, cause this daemon used to
>> be called nscd.
>> Is there any reason for inventing new name?
>>
> Actually, cached is called "cached" because initially it was intended only to 
> cache the already obtained results of nsswitch queries. This is not how nscd 
> behaves. Nscd makes all queries (to LDAP, NIS, etc) by itself and caches the 
> results. cached only cached the results of requests, that have been actually 
> made by other applications. The name "cached" was used to highlight this 
> difference.
> 
> Later, however, the nscd-like functionality was added to cached, but the name 
> was left unchanged. Personally I don't see any reasons why "cached" can't be 
> renamed to nscd right now - so I guess, it would happen in the nearest 
> future.
> 
> cached is similar to nscd in many ways (including configuration file). To 
> enable nscd-like behavior for particular nsswitch database (so that cached 
> will make all queries by itself), "perform-actual-lookups" directive should 
> be specified in cached.conf. Please see cached(8) manual page for more 
> details about this. The choice of using or not using "perform-actual-lookups" 
> mode depends on many factors and should be made by system administrator. The 
> default value for this option is "no".
> 
> Can't say anything about when cached would be MFC'ed. All I can say is that 
> it's possible. It will require a number of changes in libc, but they all are 
> fine-grained and, as far as I remember, the patch, that I've used to add 
> caching support to -CURRENT was almost clearly applicable to RELENG_6.
> 

WBR.
Dmitriy