When you upgrade Cloudera Manager, configurations for some services may become stale. This page describes those configurations and whether a restart can be delayed.
Sentry and Solr stale configuration
On a KeyValue Store Indexer service, Sentry was enabled if the Solr dependency was using Sentry, even if the KeyValue Store Indexer was set to none in its Sentry dependency configuration. This is now corrected for CDH 5.14 or higher clusters.
After upgrading Cloudera Manager, clusters on CDH 5.14 or higher will be marked as stale if you have Sentry enabled for Solr but not enabled for KeyValue Store Indexer. If you are affected by this bug, restart the stale services to pick up this fix.
Cloudera Issue: OPSAPS-43695
Kafka Broker and MirrorMaker stale configuration
Kafka broker and MirrorMaker processes now listen on only the loopback interface for JMX connections. The fix causes Kafka brokers and MirrorMaker to be marked as stale after upgrading to Cloudera Manager 6.0.0 or later.
Perform a rolling restart of Kafka.
Cloudera Issue: OPSAPS-46633
HDFS DataNode stale configuration
After upgrading to Cloudera Manager 5.10 or higher, the following configuration for HDFS DataNodes will be marked as stale: dfs.datanode.balance.max.concurrent.moves.
The staleness is caused by the following new feature: HDFS balancer can now be configured to specify which hosts are included and excluded or which hosts are used as sources for transferring replicas. Additional properties for tuning the performance of the balancer can now also be configured starting with CDH 5.10.0.
You can safely ignore this warning and defer restarting.
Cloudera Issue: OPSAPS-36642
YARN MapReduce Job History stale configuration
When upgrading to Cloudera Manager 5.10, YARN will be marked as having stale configuration due to mapreduce.jobhistory.loadedjob.tasks.max. Unless you change this parameter and want the non-default value to take effect (only takes effect in CDH 5.9+), you can simply ignore this staleness and defer restarting YARN.
Cloudera Issue: OPSAPS-32132
Change in Hue Load Balancer version causes Hue Load Balancer and Server to be marked with stale configuration
- The Enable TLS/SSL for Hue property is set to true.
- The Hue load balancer is enabled
Workaround: If your cluster uses Apache httpd 2.4 as the Hue load balancer, restart the Hue service promptly. If your cluster uses an earlier version of httpd, there is no urgency to restart the Hue service. (Apache httpd 2.4 is installed automatically by some recent versions of Linux, or may have been explicitly installed.)
Cloudera Issue: OPSAPS-40700, OPSAPS-41850
Hue Load Balancer SSL Handshake error
The Hue load balancer previously set the ProxyPreserveHost directive to On, when it should have been set to Off. This causes problems making SSL connections when using Apache httpd 2.4 or higher. The error caused problems when verifying the CN, which older versions of Apache httpd did not encounter because they did not properly verify the CN.
When upgrading Cloudera Manager, the Hue load balancer may be marked as having a stale configuration. If you are experiencing issues connecting to Hue with SSL, restart the Hue service to update the configuration.
Cloudera Issue: OPSAPS-40700
Maintenance State Minimal Block Replication staleness after upgrade
Upgrading to Cloudera Manager 5.12 or later may show Maintenance State Minimal Block Replication as a stale configuration under HDFS, suggesting a restart. It is safe to ignore this warning and delay restart.
Cloudera Issue: OPSAPS-39102
YARN ACL configuration property staleness after upgrade
- ACL for viewing a job - mapreduce.job.acl-view-job
- ACL for modifying a job - mapreduce.job.acl-modify-job
- Enable MapReduce ACLs - mapreduce.cluster.acls.enabled
It it safe to defer restart if you are not using YARN job view/modify ACLs.
Cloudera Issue: OPSAPS-33586