Description of Problem: The hbck repair tool sometimes generates local indexes that are inconsistent with table data when overlapping regions are encountered.

Workaround: If you know the database schema, fix this issue by dropping and recreating all local indexes of the table after the hbck tool completes operation. Alternatively, rebuild the local indexes using the following ALTER query:

ALTER INDEX IF EXISTS index_name ON data_table_name REBUILD

Description of Problem: When Ambari is being proxied by Apache Knox the QuickLinks are not rewritten to go back through the gateway.

Workaround: If all access to Ambari is through Knox, the new Ambari quicklink profile may be used to hide and/or change URLs to go through Knox permanently. Future release will make these reflect the gateway appropriately.

Description of Problem: Since Spark 2.0.0, `DROP PARTITION BY RANGE` does not support relative logical operators. In other words, only '=' is supported while `<', '>', '<=', '>=' are not.

Error Message:

scala> sql().show
org.apache.spark.sql.catalyst.parser.ParseException:
mismatched input }(line 1, pos 31)

== SQL ==
alter table t drop partition (b<1)
-------------------------------^^^

Workaround: To drop partition, use the exact match with '='.

scala> sql().show

Description of Problem: A Hive query submitted to the %jdbc interpreter returns a proxy validation error.

Associated error messages:

Workaround:

Description of Problem: When used with Hive, the %jdbc interpreter might require Hadoop common jar files that need to be added manually.

Workaround:

Description of Problem: Rebuild(ALTER INDEX IDX ON TABLE REBUILD) of indexes created on the table having row_timestamp column will result in no data visible to the User for that Index.

Workaround: Drop the index and recreate the same index. There will not be any extra overhead of recreating index when compared with rebuild Index.

Description of Problem: Zeppelin does not show all WARN messages thrown by spark-shell. The log level that comes as output at the Zeppelin notebook level cannot be changed .

Workaround: Currently, there is no known workaround.

Description of Problem: When upgrading from HDP-2.5.x to HDP-2.6.0 and using Spark2, the Livy interpreter is configured with a scope of 'global', and should be changed to 'scoped'.

Workaround: After upgrading from HDP 2.5 to HDP 2.6, set the interpreter mode for %livy (Spark 2) to "scoped" using the pulldown menu in the %livy section of the Interpreters page.

Description of problem: "ValueError: No JSON object could be decoded" when restarting Zeppelin, when the disk is 100% full.

Associated error message: Get following in error logs

Traceback (most recent call last):
File , line 312, in <module>
Master().execute()
File , line 280, in execute
method(env)
File , line 182, in start
self.update_kerberos_properties()
File , line 232, in update_kerberos_properties
config_data = self.get_interpreter_settings()
File , line 207, in get_interpreter_settings
config_data = json.loads(config_content)
File , line 339, in loads
 _default_decoder.decode(s)

Workaround: Free up some space in disk, then delete /etc/zeppelin/conf/*.json, then restart zeppelin server

Component Affected: Ranger, all

Description of Problem: SSL is deprecated; its use in production is not recommended. Use TLS.

Workaround: For Ambari: Use ssl.enabled.protocols=TLSv1|TLSv1.1|TLSv1.2 and security.server.disabled.protocols=SSL|SSLv2|SSLv3. For help configuring TLS for other components, contact customer support. Documentation will be provided in a future release.

Description of Problem: Zeppelin fails to start during the upgrade process from HDP 2.5 to HDP 2.6. The error starts with

Exception in thread "main" org.apache.shiro.config.ConfigurationException: Unable to instantiate class org.apache.zeppelin.server.ActiveDirectoryGroupRealm for object named 'activeDirectoryRealm'. Please ensure you've specified the fully qualified class name correctly.

Workaround: This error is due to a change in configuration class for Active Directory.

In HDP 2.5:

org.apache.zeppelin.server.ActiveDirectoryGroupRealm

In HDP 2.6:

org.apache.zeppelin.realm.ActiveDirectoryGroupRealm

To resolve this issue, choose one of the following two alternatives:

Component Affected: Zeppelin/Livy

Description of Problem: This occurs when running applications through Zeppelin/Livy that requires 3rd-party libraries. These libraries are not installed on all nodes in the cluster but they are installed on their edge nodes. Running in yarn-client mode, this all works as the job is submitted and runs on the edge node where the libraries are installed. In yarn-cluster mode,this fails because the libraries are missing.

Workaround: Set the location inspark.jars in spark-defaults.conf. For Livy, you will need to set livy.spark.jars (the HDFS location) in livy interpreters conf. Both are globally applicable. The jars need to be present on the livy machine in both cases. Updating livy conf is preferable since it affects only the Zeppelin users.

Description of Problem: Executing concurrent queries over Spark via Spark1-llap package spawns multiple threads. This may cause multiple queries to fail. However, this will not break the spark thrift server. Spark 1.6 is built using Scala 2.10, which is where this issue manifests (i.e. " synchronize reflection code as scala 2.10 reflection is not threadsafe self"). This issue was subsequently fixed in Scala 2.11 based on this patch https://issues.scala-lang.org/browse/SI-6240.

Associated error messages:

Workaround: Isolate the broken queries and re-run them one by one. This will limit the query to one spawned thread.

Description of Problem: After upgrading from Ambari 2.4.2 to Ambari 2.5.2 and subsequent HDP stack upgrade from 2.5 to 2.6, jdbc(hive) interpreter fails to work correctly in Zeppelin.

Associated Error Message: You might see one of the following errors in the Zeppelin stacktrace after running jdbc(hive):

Workaround:

Since Mahout is deprecated in favor of Spark ML, and every code change carries the risk of creating additional incompatibilities, we will document these difficulties rather than change these established behaviors in Mahout. These issues affect only Mahout.

Description of Problem: RDMBS schema for Hive metastore contains an index HL_TXNID_INDEX defined as

CREATE INDEX HL_TXNID_INDEX ON HIVE_LOCKS USING hash (HL_TXNID);

Hash indexes are not recommended by Postgres. Details can be found in https://www.postgresql.org/docs/9.4/static/indexes-types.html.

Workaround: It's recommended that this index is changed to type BTREE.

Description of problem: Sudden increase in Login Session audit events from Ranger Usersync and Ranger Tagsync.

Workaround: If policy storage DB size increases suddenly, periodically backup and purge 'x_auth_sess' table. Take backup of policy DB store and purge 'x_auth_sess' table from Ranger db schema.

Description of problem: Running Hive with Tez fails to load configured native library. For example, Snappy compression library.

Associated error message:

 java.lang.RuntimeException: java.io.IOException: Unable to get CompressorType for codec (org.apache.hadoop.io.compress.SnappyCodec). This is most likely due to missing native libraries for the codec.

Workaround: Add the configuration parameter mapreduce.admin.user.env to tez-site.xml, specifying the native library path. For example,

<property>
  <name>mapreduce.admin.user.env</name>
  <value>LD_LIBRARY=./tezlib/lib/</value>
</property>

Description of problem: The pie chart does not display the correct distribution as per data. This occurs when there is a "," in data i.e. there is number formatting applied to data.

Associated error message: No error message.

Workaround: A manual config in Zeppelin's JDBC interpreter setting, i.e. to add "phoenix.phoenix.query.numberFormat" with value "#.#".

Description of Problem: Livy Spark interpreter will only print out the last line of code in the output. For example, if you submit the following:

print(10)
print(11)

Livy will only print "11" and ignore the first line.

Workaround: If you want to see the output of a particular line, it must be the last line in the code block in a para.

Description of problem: Spark does not handle the INSERT INTO OVERWRITE operations correctly when an HDFS quota is set on the Trash folder. This can result in Spark recording the result of the operation incorrectly. If Spark is unable to move files into the trash due to a quota limit, they will incorrectly be recorded as part of the result.

Workaround: An available patch permanently deletes files when the quota on the trash is reached.

Description of problem: Delete snapshot fails even if user has Admin privilege due to namespace is not considered in Authorization flow for Hbase Ranger plugin.

Associated error message: ERROR: org.apache.hadoop.hbase.security.AccessDeniedException: Insufficient permissions for user '<username>' (action=admin)

Workaround: For delete snapshot to succeed, user needs to be system-wide admin.

Description of problem or behavior: spark.sql.hive.convertMetastoreParquet and spark.sql.hive.convertMetastoreOrc are the option to use Spark's built-in reader and writer instead of Hive SerDe. If these options are true, Spark ignores the table properties. Since Apache Spark 2.0, Spark ignores Parquet Hive table properties because convertMetastoreParquet is true by default.

Workaround:

Description of problem: Concatenating ORC tables from Spark can cause data loss.

Workaround: Use Hive to concatenate ORC tables instead.

Description of problem: Job data is published to listeners (for example, through Atlas as sqoop.job.data.publish.class) during Hive and HCat imports. Currently this happens before the Hive import completes, so it gets reported even if Hive import fails.

Workaround: Currently, there is no known workaround.

Description of Problem: A rare condition can occur with queries against index tables that may not return all expected records. This only happens when a Region of that index table is being compacted and also scanned at the same time. This issue is difficult to reproduce as this incorrect result does not happen each time.

Description of Problem: Only an hbase superuser can perform HBbase backup-and-restore.

Description of problem: This is an issue with the Spark ORC file native implementation where the timestamp type returns incorrect millisecond values with spark.sql.orc.impl=native.

Workaround: Set spark.sql.orc.impl=hive.

Description of problem: Open JDK 8u242 is not supported as it causes Kerberos failure.

Workaround: Use a different version of Open JDK.

CVE-2018-11768: HDFS FSImage Corruption (potential DoS, file/dir takeover)

In very large clusters, the in-memory format to store the user, group, acl, and extended attributes may exceed the size of the on disk format, causing corruption of fsImage.

For more information on this issue, see the corresponding Knowledge article: TSB 2021-327:CVE-2018-11768: HDFS FSImage Corruption (potential DoS, file/dir takeover)

Impact of LDAP Channel Binding and LDAP signing changes in Microsoft Active Directory

Microsoft has introduced changes in LDAP Signing and LDAP Channel Binding to increase the security for communications between LDAP clients and Active Directory domain controllers. These optional changes will have an impact on how 3rd party products integrate with Active Directory using the LDAP protocol.

Workaround

Disable LDAP Signing and LDAP Channel Binding features in Microsoft Active Directory if they are enabled

For more information on this issue, see the corresponding Knowledge article: TSB-2021 405: Impact of LDAP Channel Binding and LDAP signing changes in Microsoft Active Directory

CVE-2020-9492 Hadoop filesystem bindings (ie: webhdfs) allows credential stealing

WebHDFS clients might send SPNEGO authorization header to remote URL without proper verification. A maliciously crafted request can trigger services to send server credentials to a webhdfs path (ie: webhdfs://…) for capturing the service principal

For more information on this issue, see the corresponding Knowledge article: TSB-2021 406: CVE-2020-9492 Hadoop filesystem bindings (ie: webhdfs) allows credential stealing

KMS Load Balancing Provider Fails to invalidate Cache on Key Delete

For more information on this issue, see the corresponding Knowledge article: TSB 2020-434: KMS Load Balancing Provider Fails to invalidate Cache on Key Delete

Corruption of HBase data stored with MOB feature

For more information on this issue, see the corresponding Knowledge article: TSB 2021-465: Corruption of HBase data stored with MOB feature on upgrade from CDH 5 and HDP 2

CVE-2021-27905: Apache Solr SSRF vulnerability with the Replication handler

The Apache Solr ReplicationHandler (normally registered at "/replication" under a Solr core) has a "masterUrl" (also "leaderUrl" alias) parameter. The “masterUrl” parameter is used to designate another ReplicationHandler on another Solr core to replicate index data into the local core. To help prevent the CVE-2021-27905 SSRF vulnerability, Solr should check these parameters against a similar configuration used for the "shards" parameter.

For more information on this issue, see the corresponding Knowledge article: TSB 2021-497: CVE-2021-27905: Apache Solr SSRF vulnerability with the Replication handler

HBase MOB data loss

HBase tables with the MOB feature enabled may encounter problems which result in data loss.

For more information on this issue, see the corresponding Knowledge article: TSB 2021-512: HBase MOB data loss