We recently encountered an incident where a failover from the primary cluster to a disaster recovery cluster was being tested. MirrorMaker 2 was configured as an active->passive DR solution.

When the user wanted to failback to the primary, they reversed the direction of MM2 by simply swapping the source and target clusters in the MM2 configuration. This did not work as anticipated and left the user with an unexpected task of rebuilding the primary cluster and redeploying MM2 to replicate back from DR from scratch.

This documentation task would add some warnings around trying to recover by reversing MM2 and detail the expected recovery steps and complexity for a failback / recovery scenario.