You might try using DRBD. It's a block-level replication technology that will synchronously (or asynchronously, if replicating over a distance) create a raid-1 relationship between two or more nodes.
DRBD can be used not only with newly allocated storage, it can also replicate existing volumes - provided you can either extend that volume to make room a small amount of metadata at the end, or have another volume attached that you can use as dedicated metadata storage. Dedicated metadata is easier to remove if you only intend to use this temporarily.
The process would generally involve establishing a DRBD device on each node that stores your tables, shutting down the database on one side, then immediately starting it on the other - with identical data.