DNS Round Robin

[Kevin Darcy]

James Sebastian wrote:

>Kevin Darcy <kcd at daimlerchrysler.com> wrote in message news:<c1gjkm$fki$1 at sf1.isc.org>...
>  
>
>>James Sebastian wrote:
>>
>>    
>>
>;
>  
>
>>>rrset-order {
>>>       class IN type A name "hp110.test.com" order cyclic;
>>>       };
>>>
>>>I tried order cyclic and fixed. But lookup results is not in round
>>>robin fashion.
>>>
>>>      
>>>
>>There is no *true* round-robin ordering in modern versions of BIND. The 
>>closest approximation is "cyclic", which as you have noticed picks the 
>>first record at random and then orders the remaining records in 
>>cyclic-sequential order, relative to that record. 
>>    
>>
>
>I fear even this is not happening. By cyclic- sequential I expect it
>to be sequential after first pick.
>Now I am getting like this.
>
>
>[root at rhas3 named]# nslookup hp110
>Server:  rhsa3.test.com
>Address:  172.22.45.57
>
>Name:    hp110t2.test.com
>Address:  127.0.0.3
>Aliases:  hp110.test.com
>
>[root at rhas3 named]# nslookup hp110
>Server:  rhsa3.test.com
>Address:  172.22.45.57
>
>Name:    hp110t3.test.com
>Address:  127.0.0.4
>Aliases:  hp110.test.com
>
>[root at rhas3 named]# nslookup hp110
>Server:  tcsrhsa3.test.com
>Address:  172.22.45.57
>
>Name:    hp110t3.test.com
>Address:  127.0.0.4
>Aliases:  hp110.test.com
>
>[root at rhas3 named]# nslookup hp110
>Server:  rhsa3.test.com
>Address:  172.22.45.57
>
>Name:    hp110t2.test.com
>Address:  127.0.0.3
>Aliases:  hp110.test.com
>
>
>That is 3,4,4,3 when the address are 2,3,4 & 5 alloted to hp110 and
>rrset-order is kept as cyclic.
>
>Is this not random?? I could not find any cyclic behavious here and as
>you told not cyclic-sequential ofcourse. Documentation also I
>understand it should behave what you told.
>
>Can somebody explain me further on this behaviour.
>
>Now, If any want to keep state and do a true round robin any
>alternative DNS packages are there, which keeps the previous state and
>give true round robin??
>
hp110.test.com should not be a multi-valued CNAME (which violates the 
RFCs). Try setting it up as a multi-valued A record instead, and you'll 
see the cyclic ordering relative to the (randomly-chosen) first record.

                                                                         
                              -Kevin