Verify that replication works

Confirm data has been replicated from a source cluster to a remote destination cluster.

Install and configure YARN on the source cluster.

If YARN cannot be used in the source cluster, configure YARN on the destination cluster to verify replication.

If neither the source nor the destination clusters can have YARN installed, you can configure the tool to use local mode; however, performance and consistency could be negatively impacted.
Ensure that you have the required permissions:
- You have sudo permissions to run commands as the hbase user, or a user with admin permissions on both clusters.
- You are an hbase user configured for submitting jobs with YARN.
note
To use the hbase user in a secure cluster, use Cloudera Manager to add the hbase user as a YARN whitelisted user. For a new installation, the hbase user is already added to the whitelisted users. In addition, /user/hbase should exist on HDFS and owned as the hbase user, because YARN will create a job staging directory there.

Run the VerifyReplication command:

src-node$ sudo -u hbase hbase org.apache.hadoop.hbase.mapreduce.replication.VerifyReplication peer1 table1
...
        org.apache.hadoop.hbase.mapreduce.replication.VerifyReplication$Verifier$Counters
                BADROWS=2
                CONTENT_DIFFERENT_ROWS=1
                GOODROWS=1
                ONLY_IN_PEER_TABLE_ROWS=1
        File Input Format Counters
                Bytes Read=0
        File Output Format Counters
                Bytes Written=0

The following table describes the VerifyReplication counters:

Table 1. VerifyReplication Counters
Counter	Description
`GOODROWS`	Number of rows. On both clusters, and all values are the same.
`CONTENT_DIFFERENT_ROWS`	The key is the same on both source and destination clusters for a row, but the value differs.
`ONLY_IN_SOURCE_TABLE_ROWS`	Rows that are only present in the source cluster but not in the destination cluster.
`ONLY_IN_PEER_TABLE_ROWS`	Rows that are only present in the destination cluster but not in the source cluster.
`BADROWS`	Total number of rows that differ from the source and destination clusters; the sum of `CONTENT_DIFFERENT_ROWS` + `ONLY_IN_SOURCE_TABLE_ROWS` + `ONLY_IN_PEER_TABLE_ROWS`

By default, VerifyReplication compares the entire content of table1 on the source cluster against table1 on the destination cluster that is configured to use the replication peer peer1.

Use the following options to define the period of time, versions, or column families

Table 2. VerifyReplication Counters
Option	Description
`--starttime=<timestamp>`	Beginning of the time range, in milliseconds. Time range is forever if no end time is defined.
`--endtime=<timestamp>`	End of the time range, in milliseconds.
`--versions=<versions>`	Number of cell versions to verify.
`--families=<cf1,cf2,..>`	Families to copy; separated by commas.

The following example, verifies replication only for rows with a timestamp range of one day:

src-node$ sudo -u hbase hbase org.apache.hadoop.hbase.mapreduce.replication.VerifyReplication --starttime=1472499077000 --endtime=1472585477000 --families=c1  peer1 table1