Fixed issues in Cloudera Manager 6.3.4

TSB-431: Cloudera Manager 6.x issue with the service role Resume
If a selected service role on a node is restarted and fails, and the customer clicks the "Resume" button in Cloudera Manager, the service role on all of the nodes will be restarted concurrently.

Knowledge article: For the latest update on this issue see the corresponding Knowledge article: Cloudera Customer Advisory: Cloudera Manager 6.x issue with service role Resume

Cloudera Bug: OPSAPS-57033: HDFS configuration parameter FS_CHECKPOINT_DIR_LIST causes "Too few entries. The minimum is 1." error
In Cloudera Manager, for HDFS's Secondary Namenode role the parameter FS_CHECKPOINT_DIR cannot be set to null if the role is in-use. However, if the role is removed from the service (for example, if HDFS High Availability mode is enabled, then Secondary Namenode is removed automatically) .This parameter can now be set to null.
Cloudera Bug: OPSAPS-56398: Server needs to accept exported JAVA_HOME for Java 11
Cloudera Manager server can now accept a custom JAVA_HOME location for supported Java versions 8 and 11. In order to provide a JAVA_HOME path:
  1. Log in to the Cloudera Manager server host as the root user.
  2. Run the following command:
    service cloudera-scm-manager stop
  3. Add the following value to the /etc/default/cloudera-scm-server file:
    export JAVA_HOME=/custom/java_home_path/
  4. Run the following command:
    service cloudera-scm-manager start
  5. Ensure that the Cloudera Manager Server is running.
  6. Go to Cloudera Manager > Hosts > Host ConfigurationJava Home Directory and set the custom JAVA_HOME path for hosts that also have Java installed on the custom path.
  7. Restart all the roles for them to start using the new custom JAVA_HOME path.
Cloudera Bug: OPSAPS-56153: Diagnostic bundle uploads fail to authenticate with Proxy
Cloudera Manager diagnostic bundle uploads have fixed the 407 authentication exception issue, when a diagnostic bundle was uploaded using a proxy server with basic authentication.
Cloudera Bug: OPSAPS-56607: Monitor daemon fails to remove principal
Fixed the issue when Cloudera Manager Agent produces the following error when regenerating credentials for a service:
[18/Apr/2020 01:09:25 +0000] 13158 CredentialManager kt_renewer 
WARNING Couldn't kinit as 'solr/' using /var/run/cloudera-scm-agent/process/1546364905-solr-SOLR_SERVER/solr.keytab — kinit: 
Client's credentials have been revoked while getting initial credentials.
Cloudera Bug: OPSAPS-56650: Generate Missing Credentials Fails due to parsing issue
The components in DomainNames (DNs) like cn, dc, ou are valid even with whitespaces which causes the generate missing credentials script ( in Cloudera Manager to fail
Cloudera Bug: OPSAPS-56611: Failed to set a custom Cloudera Manager Server log location
Fixed the issue where relocation of Cloudera Manager Server log directory failed when following the documented steps to set the location of the Cloudera Manager server log.
Cloudera Bug: OPSAPS-55810: Cloudera Manager Host Resource page shows invalid Unit/Values for memory for Ozone roles
The Cloudera Manager > Configuration pages showing invalid "Memory Overcommit warnings" with incorrect units are now fixed.
OPSAPS-57587 Changes to how Cluster Templates handle null values
In Cloudera Manager, the cluster template contains key & value pairs of variables. However, keys without any value(s), get exported into the cluster template's json file. This can happen even if the Cloudera Manager Admin Console exposes the issue by showing up configuration warnings. Importing such templates to new clusters also introduces these warnings. Therefore, an export of such templates now requires the user to replace the 'null' value. Importing the template with such a key will fail with an error message.
Cloudera Bug: OPSAPS-44883: Error when installing MariaDB 10.2
Fixed an issue where a database statement fails with the following error while installing or upgrading Cloudera Manager: - "Key column 'REVISION_ID' doesn't exist in table"
Cloudera Bug: OPSAPS-57607:Cloudera Manager fails to redact information
Fixed an issue where the Cloudera Manager Agent failed to redact with the following error: UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 36
Cloudera Bug: OPSAPS-55143: Disable TLS for agent status server port on RHEL6
Fixed a file descriptor leak in the Cloudera Manager agent when running on Redhat 6 platform with TLS/SSL enabled for Cloudera Manager communications. The fix is to disable TLS/SSL for the Cloudera Manager agent status server port on Redhat 6. Heartbeat protocol and Navigator data (eg audit events) remain encrypted. 6.2.2 :
Fixed a file descriptor leak in the Cloudera Manager agent when running on all platforms with TLS/SSL enabled for Cloudera Manager communications. The fix is to disable TLS/SSL for the Cloudera Manager agent status server port. Heartbeat protocol and Navigator data (eg audit events) remain encrypted.
Cloudera Bug: OPSAPS-53158: Command Succeeded but fetching results failed" caused by file descriptor limit
Fixes file descriptor leak by the Cloudera Manager Agent's HTTP status server when Auto-TLS is enabled. The fix applies to all supported operating systems , except Redhat 6-based operating systems. Customers running on Redhat 6 should either upgrade to Redhat 7, or disable Agent TLS (via Administration -> Settings -> Security: Use TLS Encryption for Agents).
Cloudera Bug: OPSAPS-57254: Agent reporting hangs infinitely
Previously, if the agent encountered a problem with the SSL connection while reporting to the HostMonitor, reporting would hang indefinitely causing the entire node to be marked as in Bad Health. Now, a problem with the SSL connection will no longer block the agent from reporting.
Cloudera Bug: OPSAPS-55786: Excluding directory from disk usage report has no effect
Directories removed from HDFS will not appear in the Current Directory Usage report.
Cloudera Bug: OPSAPS-57249: Reports Manager unable to index 60G fsimage
Reports Manager is now able to process large fsimages.
Cloudera Bug: OPSAPS-57534: Reports Manager historical usage report loses data when a directory is removed from watched directories
Historical Disk Usage reports will not lose the already generated data of the directory after it is set to unwatched.
Cloudera Bug: OPSAPS-56242: External user authorization roles become mixed up when posting multiple mappings to API
When creating external role mappings via the /externalUserMappings API endpoint, if multiple mappings are given in the request, then later mappings will erroneously inherit roles specified in preceding mappings, thus granting those users or groups additional privileges than expected. This has been fixed. Note that existing role mappings that have already been created will not be fixed by this change. Clouderarecommens that you review your existing role mappings in Administration > Users and Roles to correct any inconsistencies.
Cloudera Bug: OPSAPS-57422: HBase 2.0 JMX GET metrics changed
Metrics removed in HBase2 are no longer polled by Cloudera Manager.
Cloudera Bug: OPSAPS-56024: HBase 2.0 JMX metrics changed - "AssingmentManger" is now renamed to "AssignmentManager"
Cloudera Bug: OPSAPS-56473: Solr shard splitting breaks monitoring
Solr metrics were missing after a shard had been split in the cluster.
Cloudera Bug: OPSAPS-56586: Change default wal provider from AsyncFSWal to Filesystem
Added a new HBase configuration property: hbase.wal.regiongrouping.delegate.provider
Cloudera Bug: OPSAPS-56737: Get the thread naming patch merged
The names of Cloudera Manager Agent threads are now visible in operating system tools and follow a new scheme to improve supportability.
Cloudera Bug: OPSAPS-54869: agent - filesystem monitoring proper filtering
Fixed spurious errors about missing mounts logged by the agent when user runtime directories in /run/user are removed normally.
Cloudera Bug: OPSAPS-56714: Possible misinterpretation of Impala query endTime
Impala queries held open after they are finished (e.g. in Hue) now appear on the Impala query monitoring page of Cloudera Manager upon closure, without being logged as "outside acceptance window".
Cloudera Bug: OPSAPS-55356: Enhance CM to have Telemetry Publisher to show the chart with the amount of data exported to WXM.
OPSAPS-55356 Enhance CM to have Telemetry Publisher to show the chart with the amount of data exported to WXM. OPSAPS-55357 Add a configuration to Telemetry Publisher allowing the users to limit the amount of data exported in a given hour. Added the configuration "telemetry.databus.exportedDataThresholdPerHour" to Telemetry Publisher. There will not be UI provided for it. The default behaviour will be as it is today. This configuration can be provided by the customers if they want to control the amount data exported by Telemetry Pubisher in one hour. The Telemetry publisher verifies the amount of data exported in last one hour and if it crosses the spefified threshold amount in bytes, it will skip exporting of the data till next hour.
Cloudera Bug: OPSAPS-56762: Speed up replication diagnostic bundle collection
The performance of collecting Replication diagnostic bundles has been improved.
Cloudera Bug: OPSAPS-57245: schedule for BDR log retention limit should be triggered automatically, but it's not
Adding command_schedules during cluster creation
Cloudera Bug: OPSAPS-56341: Hive BDR fails between Isilon clusters after CM+CDH upgrade to 6x in target
Skipping using web HDFS client configurations for Isilon.
Cloudera Bug: OPSAPS-48770: Update invalidate metadata script to use backticks.
Even if table name starts with a number, the Impala invalidate metadata command succeeds after this fix.
Cloudera Bug: OPSAPS-43909: Exclusion Filter should also apply to Delete Policy
Exclusion filters apply to delete policy as well.
Cloudera Bug: OPSAPS-57291: HDFS replication for Isilon can't run remoteHdfsListing
Fixed for Isilon
Cloudera Bug: OPSAPS-57913: When decommissioning HDFS Datanodes, Cloudera Manager may prematurely report success when decommissioning is still in process in certain cases.
Cloudera Manager now correctly decommissions the Datanodes when HDFS high availability is enabled.

Cloudera Manager should now correctly wait the HDFS Datanodes to finish decommissioning. The reliability when HDFS high availability is enabled is also improved.