Known Issues and Limitations in Cloudera Manager 6.1.0

TSB-359 Backup and Disaster Recovery (BDR) HDFS and Hive Replications will fail on clusters running Cloudera Manager 6.1.0

Backup and Disaster Recovery (BDR) HDFS and Hive Replications will fail when replicating from secured (Kerberized) source clusters to destination clusters that have been upgraded to Cloudera Manager 6.1.0.

This also affects new installations of Cloudera Manager 6.1.0 on the destination cluster if an admin restarts the Cloudera Manager service.

Products affected: Cloudera Manager Backup and Disaster Recovery in a secure (Kerberized) environment

Releases affected: Cloudera Manager 6.1.0 (when used as the destination cluster of HDFS and/or Hive replication)

Users affected: Customers using HDFS or Hive Replication

Severity (Low/Medium/High): High

Root Cause and Impact:

In HDFS and Hive Replication, Cloudera Manager first runs a process on the destination cluster to verify if the replication is possible. Due to a bug, the source cluster is treated as an insecure (non-kerberized) cluster. As a result, replication fails.

You will see the exception javax.security.sasl.SaslException: GSS initiate failed [Caused by GSSException: No valid credentials provided (Mechanism level: Fail to create credential. (63) - No service creds)] in the process stderr logs.

Immediate action required: If you use BDR, do not upgrade a destination cluster to Cloudera Manager 6.1.0. Upgrade to Cloudera Manager 6.1.1 or higher when it becomes available.

If you have already upgraded your destination cluster to Cloudera Manager to 6.1.0, use the following workaround:

  1. For an existing HDFS or Hive replication schedule, select Actions > Edit Configuration.
  2. Save the schedule.

Please note that you will need to edit only one schedule even if you have multiple schedules.

Note: This workaround is not persistent. That is, if you restart the Cloudera Manager service, you must repeat the above workaround.

BDR Job Mapper shows the following warning: AuthenticationException: GSSException: No valid credentials provided

A BDR Job Mapper might succeed with the following stack trace in the log message:
2018-12-05 13:57:03,475 WARN [main] org.apache.hadoop.crypto.key.kms.LoadBalancingKMSClientProvider: KMS provider at [https://src-3.example.com:16000/kms/v1/] threw an IOException:
java.io.IOException: org.apache.hadoop.security.authentication.client.AuthenticationException: GSSException: No valid credentials provided (Mechanism level: Fail to create credential. (63) - No service creds)
at org.apache.hadoop.crypto.key.kms.KMSClientProvider.createConnection(KMSClientProvider.java:492)
at org.apache.hadoop.crypto.key.kms.KMSClientProvider.decryptEncryptedKey(KMSClientProvider.java:793) 

The warning appears if you try to replicate from an encrypted source cluster that has multiple KMS instances.

Workaround: You can safely ignore this message because the client succeeds upon fail over.

Affected Versions: CDH 6.1.0

Cloudera Issue: CDH-76053

Diff Format changed for Configuration History page

Color formatting for the diff display omits the red and green colors that indicate what was removed and added.

Affected Versions: Cloudera Manager 6.1.0

Cloudera Issue:OPSAPS-48544

Backup and Disaster Recovery (BDR) performance regression after upgrading to CDH 6.0.0

Hive replication with BDR experiences a performance regression when comparing CDH 6.0.0 and CDH 5.14.4. The slowdown occurs during the import step. For example, the performance regression may only be 10% for 4 million partitions. As the number of partitions goes down though, the performance impact becomes more visible. For example, 100,000 partitions may experience a 20% performance regression.

Affected Versions: Cloudera Manager 6.0.0, 6.01, 6.1.0; CDH 6.0.0, 6.0.1, 6.1.0

Cloudera Issue: OPSAPS-47520

Cloudera Manager allows more than a single space in YARN Admin ACLs

When adding a YARN Admin ACL in Cloudera Manager, you are allowed to enter multiple spaces in the entry. The space is the separator between the user and group lists, and only a single space should be allowed in the entry. All entries that appear after a second single space in a YARN Admin ACL will be ignored.

Affected Versions: Cloudera Manager 6.0.0, 6.0.1, 6.1.0

Cloudera Issue: OPSAPS-47688

Integer data types map to Float in Swagger API client

Integer data types show up as floating point numbers when using the Cloudera Manager API Python client.

Affected Versions: Cloudera Manager 6.0.0, 6.0.1, 6.1.0

Cloudera Issue: OPSAPS-45689

Cloudera Data Science Workbench is not supported with Cloudera Manager 6.x

Cloudera Data Science Workbench (1.4.x and lower) is not currently supported with Cloudera Manager 6.x. If you try to upgrade, you will be asked to remove the CDSW service from your cluster. Cloudera Data Science Workbench will be supported with Cloudera Manager 6 in a future release.

Affected Versions: Cloudera Manager 6.x

Cloudera Issue: DSE-2769

Package Installation of CDH Fails

When you install CDH with packages from a custom repository, ensure that the version of CDH you select for Select the version of CDH matches the version of CDH for the custom repository. Selecting the CDH version and specifying a custom repository are done during the Select Repository stage of installation.

If the versions do not match, installation fails.

Affected Versions: Cloudera Manager 6.x

Fixed Versions: N/A

Apache Issue: N/A

Cloudera Issue: OPSAPS-45703

Backup and Disaster Recovery replication to/from Cloudera Manager 6 clusters require Cloudera Manager 5.14.0 or higher

You can only use BDR to replicate to/from clusters managed by Cloudera Manager 6 with Cloudera Manager 5.14.0 or higher.

Affected versions: Cloudera Manager 6.x

Cloudera Issue: OPSAPS-42207