In order to have a small redudant system, this is the way to go.
I've made a mistake and allowed colisions to happen
The increment-increment and offset increment will problably solve this
On 27 Oct 2011, at 00:38, Rick James <rjames@stripped> wrote:
> This is a big reason not to allow writing to both Masters.
>
> On 10/26/11 12:20 PM, Ricardo Freitas wrote:
>> Hello, my fellow developers and admins
>>
>>
>>
>> I have a question regarding a master-master replication system.
>>
>>
>>
>> I implemented one with the help of some of you guys - however I had some
>> problems regarding the agent so I have a master-master and not a mmm system
>> - and now I come across the awfull 1062 error - the 'Error 'Duplicate entry
>> '123020901' for key 'PRIMARY'' on query.'
>>
>>
>>
>> The issue here is I sometimes insert on master 1 (let's call it like this)
>> and sometimes on master 2. I thought this wouldn't be an issue as both
>> database would talk to one another however I think this might be the origin
>> of the problem.
>>
>>
>>
>> I already changed the system to always force the inserts query to be on
>> Master 1 however I would like to know how to avoid this. I read something
>> about configuring a skip-slave-error but I don't think this would be that
>> much of a solution as sometimes things happens and it's important to have
>> some kind of notion or warning.
>>
>>
>>
>> I read about the auto-increment-increment = 2 or even the increment offset.
>>
>>
>>
>> If I configure Master 2 to auto-increment-increment = 2 and master 1 to
>> auto-increment-increment = 1, would this help?
>>
>>
>>
>> Master 1 - example
>>
>> Row 1
>>
>> Row 2
>>
>> Row 3
>>
>>
>>
>> Master 2 - example
>>
>>
>>
>> Row 2
>>
>> Row 4
>>
>> Row 6
>>
>>
>>
>> As far as I'm concerned, eventually sometimes collisions would occur. I
>> thought that PK's weren't replicated as the database would just receive the
>> insert query and viola.
>>
>>
>>
>> How can I avoid this in the future?
>>
>>
>>
>> I read about UUID as well but as I've inherit this system I have a lot of
>> sites already relaying on the pk (for query purposes) so I can't just
>> changed this.
>>
>>
>>
>> Thanks a lot!
>>
>>
>>
>> Ricardo
>>
>>
>>
>>
>>
>
> --
> Rick James - MySQL Geek
>
>
> --
> MySQL Replication Mailing List
> For list archives: http://lists.mysql.com/replication
> To unsubscribe: http://lists.mysql.com/replication?unsub=1
>
