Details about Round robin behaviour ??

[Kevin Darcy]

I stand corrected.

It's important for folks to realize -- as I apparently forget from time 
to time -- that when "rrset-order" is mentioned in developer docs, that 
it applies to the *default* record-ordering behavior of named, not just 
when it is explicitly-defined.

                                                                         
                                    - Kevin
Mark Andrews wrote:

>>Outside of sortlist interactions, I'm not aware of any difference 
>>between 9.2.3+ and 9.3.x in this area.
>>
>>                                                                         
>>                                                   - Kevin
>>    
>>
>
>>From 9.3.1rc1 CHANGES
>
>1732.   [bug]           'rrset-order name "*"' wasn't being applied to ".".
>                        [RT #12467]
>
>1582.   [bug]           rrset-order failed to work on RRsets with more
>                        than 32 elements. [RT #10381]
>
>1555.   [func]          'rrset-order cyclic' no longer has a random starting
>                        point. [RT #7572]
>
>1224.   [bug]           'rrset-order' and 'sortlist' should be additive
>                        not exclusive.   
>
>1223.   [func]          'rrset-order' partially works 'cyclic' and 'random'
>                        are supported.
>
>
>  
>
>>mathieu.millet at ratp.fr wrote:
>>
>>    
>>
>>>>-----Message d'origine-----
>>>>        
>>>>
>>>>From : Kevin Darcy [mailto:kcd at daimlerchrysler.com] 
>>>      
>>>
>>>>Sent : jeudi 3 mars 2005 00:19
>>>>To : bind-users at isc.org
>>>>Object : Re: Details about Round robin behaviour ??
>>>>
>>>>mathieu.millet at ratp.fr wrote:
>>>>
>>>>   
>>>>
>>>>        
>>>>
>>>>>	Hi everyone,
>>>>>
>>>>>I'm interesting in getting details about the behaviour of round robin 
>>>>>feature in Bind.
>>>>>
>>>>>More specially, can somebody tell me precisely :
>>>>>	- The default behaviour of Bind when you have multiple identical 
>>>>>RRsets with Bind 9.2.3+ and Bind 9.3.x ? Are all RRsets returned ? If 
>>>>>yes, in which order ?
>>>>>
>>>>>     
>>>>>
>>>>>          
>>>>>
>>>>The default behavior of BIND is to filter out duplicate RRs. This is 
>>>>mandated by RFC 2181, Section 5. That's the answer to your first 
>>>>question, and renders the other 2 questions meaningless.
>>>>   
>>>>
>>>>        
>>>>
>>>OK. Wrong term. So my question is : what is the default behaviour of BIND
>>>when you have multiple RRs with the same label, class and type (but not
>>>data) with Bind 9.2.3+ and Bind 9.3.x (Remember I have asked questions about
>>>Round Robin) ? ...
>>>
>>> 
>>>
>>>      
>>>
>>>>>	- What is the support status of the option rrset-order (and
>>>>>orderclass) with Bind 9.2.3+ and Bind 9.3.x ?
>>>>>	- What does "rrset-order in more complete." mean (specified in Bind 
>>>>>9.3 Changes) ?
>>>>>
>>>>>I have already looked at Bind ARM and DNS&BIND from O'Reilly and I 
>>>>>can't find a conclusive answer to these questions (Maybe I didn't 
>>>>>understand the answers in these books ;o).
>>>>>
>>>>>     
>>>>>
>>>>>          
>>>>>
>>>>Instead of asking these speculative questions, why don't you just 
>>>>configure what you want to configure and see if those 
>>>>versions of BIND 
>>>>behave they way you want them to? Sometimes an ounce of real-world 
>>>>testing and verification is worth a ton of code description and/or 
>>>>explanation...
>>>>   
>>>>
>>>>        
>>>>
>>>Sorry, but I'm getting confused with the behaviour of several Bind versions
>>>(I have tested) and I would like to have an answer from a Bind expert. Also,
>>>since I didn't find these informations in documentation and in archives, it
>>>would be a good idea to leave a trace for such questions.
>>>
>>> 
>>>
>>>      
>>>
>>
>>    
>>
>--
>Mark Andrews, ISC
>1 Seymour St., Dundas Valley, NSW 2117, Australia
>PHONE: +61 2 9871 4742                 INTERNET: Mark_Andrews at isc.org
>
>
>  
>