Yoshinori Matsunobu sparked some interest recently when he posted about his MySQL MHA HA solution, and there has been some discussion of it internally at Yahoo compared with the standard we currently have.
Full disclosure: I haven’t read every bit of the documentation or tried it out yet, so I apologize in advance to Yoshinori if I mistakenly represent his hard work.
I see a lot of great ideas in Yoshinori’s release, it seems to focus on two main problems:
- A process to monitor an active master and perform a failover when it fails
- The bit that finds the most “caught-up” slave and distributes …