Replicating data to Impala clusters
Impala metadata replication is performed as a part of Hive replication. Impala replication is only supported between two CDH clusters. The Impala and Hive services must be running on both clusters.
Replicating Impala Metadata
Impala metadata replication is enabled by default but the legacy Impala C/C++ UDF's (user-defined functions) are not replicated as expected. As a workaround to ensure the replicated Impala functions work on the target, you can edit the location of the UDF functions after the replication run is complete. To accomplish this task, you can edit the “path of the UDF function” to map it to the new cluster address or you can use a script.
Invalidating Impala metadata
For Impala clusters that do not use LDAP authentication, configure during Hive external table replication policy creation so that the
replication job automatically invalidates Impala metadata after replication completes. If
the clusters use Sentry, the Impala user should have permissions to run INVALIDATE
METADATA.
The configuration causes the Hive/Impala replication job to run the Impala
INVALIDATE METADATA statement per table on the destination cluster after
completing the replication. The statement purges the metadata of the replicated tables and
views within the destination cluster's Impala upon completion of replication, allowing other
Impala clients at the destination to query these tables successfully with accurate results.
However, this operation is potentially unsafe if DDL operations are being performed on any
of the replicated tables or views while the replication is running. In general, directly
modifying replicated data/metadata on the destination is not recommended. Ignoring this can
lead to unexpected or incorrect behavior of applications and queries using these tables or
views.
Alternatively, you can run the INVALIDATE METADATA statement manually for
replicated tables.
