11. Known Issues

Insert after alter table on Acid table causes ArrayIndexOutOfBoundsException

Issue: Schema evolution is not supported for Acid tables. Alter Table commands on Acid table may make the table unreadable. There is not w/a after that.

Ambari ranger-env settings for admin and usersync log dir don't seem to do anything.

Issue: The log directory specified for the ranger-admin and ranger-usersync component configurations are not being used in the current release.

Alter table transactional=false on ACID table misleading user. create transactional table would be better

Issue: Once an ACID table is created (with transactional=true) and data is written to it, changing the property to transactional=false may cause wrong data to be returned by queries over this table.

Meta WALs are not split or cleared

Issue: In case a meta region gets moved from one region server to another region server, the write ahead logs for the meta table might be left in the HDFS directory for the regionserver if that particular regionserver fails. This might leave the WAL files in that directory until they are removed manually, and also may result in the previous regionservers to be listed as "dead regionservers" in the master UI indefinitely. However, this is harmless and does not cause any data loss other other operational issues in HBase.

UserSync - group sync is marked as INTERNAL even if the group is synced from LDAP

Issue: When the groups are synched from LDAP/AD, the ranger admin portal is showing the group as INTERNAL group instead of EXTERNAL group. It does not affect any functionality of Ranger.

After local index changes, the explain plan could show scan over the physical data table name instead of the local index. The query plan contains index id, for example -32766, represents using local indexes.

After local index changes, the explain plan could show scan over the physical data table name instead of the local index. The query plan contains index id, for example -32766, represents using local indexes.

Phoenix-queryServer ha is failing with RejectedExecutionException

Issue: It is possible the clients using the Phoenix QueryServer might unexpectedly see application-level exceptions when the QueryServer is in the process of shutting down.

Workaround: Users should simply resubmit their query against an operation Phoenix QueryServer instance.

Using empty master secret for Knox causes NullPointerException

Issue: If an empty string is used as the Knox master secret the Knox gateway server will fail to start and a NullPointerException will be found in the gateway.log file.

Error: The error found in the gateway.log file will appear as shown below. More information can be produced by modifying the gateway-log4j.properties file.

2015-12-10 17:03:30,389 INFO  hadoop.gateway (GatewayServer.java:startGateway(217)) - Starting gateway...
2015-12-10 17:03:30,579 FATAL hadoop.gateway (GatewayServer.java:startGateway(234)) - Failed to start gateway: java.lang.NullPointerException

Workaround: User can recover from this situation by replacing the empty string master secret with another value using the command below. This command should be executed as the 'knox' user.

bin/knoxcli.sh create-master --force

If this does not solve the problem it may be necessary to delete the contents of the data/security directory before executing this command.

Configuring KafkaSpout to Connect to HBase or Hive

Before connecting to HBase or Hive, we recommend that you add the following exclusions to your POM file, for the curator framework:

<exclusion>
  <groupId>org.apache.curator</groupId>
  <artifactId>curator-client</artifactId>
</exclusion>

<exclusion>
  <groupId>org.apache.curator</groupId>
  <artifactId>curator-recipes</artifactId>
</exclusion>

<exclusion>
  <groupId>org.apache.curator</groupId>
  <artifactId>apache-curator</artifactId>
</exclusion>

Issue: The Phoenix Thin Client Driver, org.apache.phoenix.queryserver.client.Driver, is not automatically registered in the JDBC DriverManager.

Workaround: