List:Commits« Previous MessageNext Message »
From:Sven Sandberg Date:March 3 2011 11:24am
Subject:Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779
View as plain text  
On 03/03/2011 10:58 AM, He Zhenxing wrote:
>>>> Actually I think this is a very good idea. Shall we create a fourth
>>>> component? We will then have:
>>>>
>>>> Store
>>>> Send
>>>> Receive
>>>> Reapply
>>>>
>>>> Perhaps with other names, perhaps prefixed by rpl.
>>>>
>>>> Sounds good?
>>>>
>>>> The only problem is commands like SHOW SLAVE STATUS and CHANGE MASTER,
>>>> which are as much part of Receive (IO thread) as of Reapply (SQL
>>>> thread). Where should those be?
>>>>
>>>
>>> Yes, that would be ideal, this was my original goal to achieve when
>>> designing the interfaces. but I think should make this as the next goal
>>> because this would introduce more refactoring of the code.
>>
>> OK. It may be good to spell out this in the WL so that we know we are
>> working in this direction.
>>
>>>
>>> And I think SHOW SLAVE STATUS and CHANGE MASTER should still belong to
>>> the Slave component, which will use the Receive and Reapply component.
>>
>> So we will actually have five components?
>> Store
>> Send
>> Receive
>> Reapply
>> Slave (interface on top of Receive + Reapply)
>>
>> Is this your idea?
>>
>
> Yes, and a Master, which will be on top of the Send component.

OK, so to summarize we have four low-level interfaces:

  Store
  Send
  Receive
  Reapply

Then, we have two interfaces on top of these four:

  Master is on top of Send
  Slave is on top of Receive + Reapply

If I understood right, Slave is needed for the few commands that are 
shared by SQL thread and IO thread. What is Master needed for? Won't all 
master functionality be in the Send component?

/Sven
Thread
bzr commit into mysql-trunk branch (hezx:3710) WL#5779He Zhenxing2 Mar
  • Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779Sven Sandberg2 Mar
    • Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779Sven Sandberg2 Mar
      • Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779He Zhenxing3 Mar
    • Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779He Zhenxing3 Mar
      • Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779Sven Sandberg3 Mar
        • Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779He Zhenxing3 Mar
          • Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779Sven Sandberg3 Mar
            • Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779He Zhenxing3 Mar
              • Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779Sven Sandberg3 Mar
                • Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779He Zhenxing4 Mar
    • Re: bzr commit into mysql-trunk branch (hezx:3710) WL#5779He Zhenxing3 Mar