Ingests random data into a table configured for replication, and then verifies that the original table and the replicated table have equivalent data using a Merkle tree.
Configure merkle-env.sh
Be sure to set the username/password to connect to Accumulo and the Accumulo instance name and ZooKeepers. The example defaults to using the same instance as the source and destination.
The randomness of the generated data can be controlled via some parameters in this file. Note that if deletes are introduced, it is very likely to cause an incorrect verification since the tombstone'ing of that delete will differ on the source and remote due to the order of minor compactions that occur.
Run configure-replication.sh
This script sets up the two instances for replication, creates the tables that will be replicated and adds some splits to them.
This is destructive to any existing replication configuration so use it with care on a real instance.
Run ingest-data.sh
Ingests the configured amount of random data into the source table.
Run ‘accumulo-cluster stop’ && ‘accumulo-cluster start’ on the source instance
A tabletserver in the source instance is likely to still be referencing a WAL for a presently online tablet which will prevent that file from being replicated. Stopping the local instance will ensure that all WALs are candidate for replication.
Run verify-data.sh
This will compute the leaves merkle tree for the source and destination tables and then compute the root hash for both. The root hash is presented to the user.
If the root hashes are equal, the test passed; otherwise, the test fails.