How to force the HMS schema version to use?

Question:

 

We are using a shared Hive Metastore (HMS) database and see errors when trying to create datasets through the Okera Portal. How can we fix this problem?

 

Answer:

Seeing errors talking to an existing HMS database often is related to the version of its schema. You can find out by connecting to the HMS database and print the content of the VERSION table. For example:

mysql> use hms_db;
mysql> select * from VERSION;
+--------+-----------------+------------------------------+
| VER_ID | SCHEMA_VERSION  | VERSION_COMMENT              |
+--------+-----------------+------------------------------+
| 1      | 2.3.0           | Hive release version 2.3.0   |
+--------+-----------------+------------------------------+
1 row in set (0.10 sec)

When you are using a HMS version of 2.0 or later, you must tell our services of that fact by setting an additional environment variable called ENABLE_HMS_2_SCHEMA. This can be done within the /etc/okera/env.sh file  by, for example, set the option at the end of env.sh file:

$ vi /etc/okera/env.sh
#!/bin/bash
...
export OKERA_ENABLE_HMS_2_SCHEMA="True"

After that you need to restart the Deployment Manager and recreate the affected ODAS clusters.

Note: If you need to edit this setting for a running cluster that you do not want to restart, please contact the Okera support to be guided through the process of editing the cluster configuration in place.