Consider I have a master on DC1 and slave on DC2. Now, DC1 network is completely down for indefinite time and I want to resume the service as soon as possible on DC2 by promoting the slave to master.

There are two problems:

  • I don't know how much the slave is lag behind the master
  • If original master went back online, how to solve the conflict?

Are there any recommended steps and procedures?

1 Answer 1


Master + Slave? Not Dual Master?

With Dual Master, each server is the Master and Slave to the other. In this case either can receive writes, but you should write to only one. More pertinent, failover and failback is readily available.

With Master + Slave, you have not prepared for the failback because things like log_bin and log_slave_updates are not turned, nor is the Master listening to the Slave.

I'm afraid your only option (now) is to stop the Slave (which you promoted to Master), copy the data, build a Slave to it (presumably on the original Master's server). That is, you lost the Master, and whatever data is on it is useless. It's like initially adding a Slave to a running system.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.