Known Issues in MapReduce and YARN
This topic describes known issues, unsupported features and limitations for using MapReduce and YARN in this release of Cloudera Runtime.
Known Issues
- OPSAPS-56577: If a Kerberos principal other than "yarn" is configured for the YARN service, then Cloudera Manager will erroneously skip adding the custom principal to the YARN keytab, causing YARN to fail to start due to a Kerberos authentication failure. This also affects Ambari to Cloudera Manager migrations, if Ambari is configured with a princpal other than "yarn" for the YARN service. A similar issue affects Hive, when using Hive LLAP.
- Workaround: You must specify "yarn" as the Kerberos principal for YARN, and specify "hive" as the Kerberos principal for Hive. When performing an Ambari to Cloudera Manager migration, set the principals for both services to those values before performing the migration.
- Fair Scheduler to Capacity Scheduler migration - fs2cs tool
-
fs2cs tool does not convert all Fair Scheduler queue configurations to Capacity Scheduler queue configurations.
- CDPD-12123: HDFS replication performance is either slowed down or not on the expected lines with CDH/CM 7.1.1.
- Post-upgrade, queue which the user is running workloads cannot grow beyond the configured capacity till its maximum capacity.
- DOCS-5966: Third party applications do not launch if MapReduce framework path is not included in the client configuration
- MapReduce application framework is loaded from HDFS instead of being present on the NodeManagers. By default, the mapreduce.application.framework.path property is set to the appropriate value, but third party applications with their own configurations will not launch.
- JobHistory URL mismatch after server relocation
- After moving the JobHistory Server to a new host, the URLs listed for the JobHistory Server on the ResourceManager web UI still point to the old JobHistory Server. This affects existing jobs only. New jobs started after the move are not affected.
- CDH-49165: History link in ResourceManager web UI broken for killed Spark applications
- When a Spark application is killed, the history link in the ResourceManager web UI does not work.
- CDH-6808: Routable IP address required by ResourceManager
- ResourceManager requires routable
host:port
addresses foryarn.resourcemanager.scheduler.address
, and does not support using the wildcard 0.0.0.0 address.
- OPSAPS-52066: Stacks under Logs Directory for Hadoop daemons are not accessible from Knox Gateway.
- Stacks under the Logs directory for Hadoop daemons, such as NameNode, DataNode, ResourceManager, NodeManager, and JobHistoryServer are not accessible from Knox Gateway.
- COMPX-1445: Queue Manager operations are failing when Queue Manager is installed separately from YARN
- If Queue Manager is not selected during YARN installation, Queue Manager operations are failing. Queue Manager says 0 queues are configured and several failures are present. That is because ZooKeeper configuration store is not enabled.
- COMPX-1451: Queue Manager does not support multiple Resource
- When YARN High Availability is enabled there are multiple Resource Managers. Queue Manager receives multiple ResourceManager URLs for a High Availability cluster. It picks the active ResourceManager URL only when Queue Manager page is loaded. Queue Manager cannot handle it gracefully when the currently active ResourceManager goes down while the user is still using the Queue Manager UI.
- COMPX-3134: Yarn applications can get stuck due to a NullPointerException in Capacity Scheduler
- If you enable Asynchronous scheduling (yarn.scheduler.capacity.schedule-asynchronously.enable=true) in capacity scheduler, there is an edge-case where NullPointerException can cause the scheduler thread to exit and the applications get stuck without allocated resources. This can be recognized by NullPointerException thrown by the capacity scheduler.
- YARN cannot start if Kerberos principal name is changed
- If the Kerberos principal name is changed in Cloudera Manager after launch, YARN will not be able to start. In such case the keytabs can be correctly generated but YARN cannot access ZooKeeper with the new Kerberos principal name and old ACLs.
- COMPX-8687: Missing access check for getAppAttemps
- When the Job ACL feature is enabled using Cloudera Manager (
mapreduce.cluster.acls.enabled
property is not generated to all configuration files, including theyarn-site.xml
configuration file. As a result the ResourceManager process will use the default value of this property. The default property ofmapreduce.cluster.acls.enabled
isfalse
.
property), the
Unsupported Features
-
The following YARN features are currently not supported in Cloudera Data Platform:
- GPU support for Docker
- Hadoop Pipes
- Fair Scheduler
- Application Timeline Server (ATS 2 and ATS 1.5)
- Container Resizing
- Distributed or Centralized Allocation of Opportunistic Containers
- Distributed Scheduling
- Native Services
- Pluggable Scheduler Configuration
- Queue Priority Support
- Reservation REST APIs
- Resource Estimator Service
- Resource Profiles
- (non-Zookeeper) ResourceManager State Store
- Shared Cache
- YARN Federation
- Rolling Log Aggregation
- Docker on YARN (DockerContainerExecutor) on Data Hub clusters
- Moving jobs between queues
- Dynamic Resource Pools
Technical Service Bulletins
- TSB 2021-539: Capacity Scheduler queue pending metrics can become negative in certain production workload scenarios causing blocked queues
- The pending metrics of Capacity Scheduler queues can become negative in certain production
workload scenarios.
Once this metric becomes negative, the scheduler is unable to schedule any further resource requests on the specific queue. As a result, new applications are stuck in the
ACCEPTED
state unless YARN ResourceManager is restarted or failed-over. - Knowledge article
- For the latest update on this issue see the corresponding Knowledge article: TSB 2021-539: Capacity Scheduler queue pending metrics can become negative in certain production workload scenarios causing blocked queues