Known Issues in Apache HBase

Learn about the known issues in HBase, the impact or changes to the functionality, and the workaround.

OpDB Data Hub cluster fails to initialize if you are reusing a cloud storage location that was used by an older OpDB Data Hub cluster

Stop HBase using Cloudera Manager before deleting an Operational Database Data Hub cluster.

IntegrationTestReplication fails if replication does not finish before the verify phase begins

During IntegrationTestReplication, if the verify phase starts before the replication phase finishes, the test fails because the target cluster does not contain all of the data. If the HBase services in the target cluster does not have enough memory, long garbage-collection pauses might occur.

Use the -t flag to set the timeout value before starting verification.

Bulk load is not supported when the source is the local HDFS

The bulk load feature (the completebulkload command) is not supported when the source is the local HDFS and the target is an object store, such as S3/ABFS.

Use distcp command to move the HFiles from HDFS to S3 and then run bulk load from S3 to S3.

Snappy compression with /tmp directory mounted with noexec option

Using the HBase client applications such as hbase hfile on the cluster with Snappy compression could result in UnsatisfiedLinkError.

Add -Dorg.xerial.snappy.tempdir=/var/hbase/snappy-tempdir to Client Java Configuration Options in Cloudera Manager that points to a directory where exec option is allowed.

HBASE-28450: BuckeCache.evictBlocksByHfileName does not work after a cache recovery from a file

When the persistent cache is recovered after a region server crashes or restarts, blocks for closed regions or compacted files are not evicted, filling the cache indefinitely, after some time, the cache reaches its capacity, and read performance degrades.

Upgrade to CDH-7.2.18.300 or a newer version.

HBASE-28458: BucketCache.notifyFileCachingCompleted might incorrectly consider a fully cached file

This behavior causes some blocks to be wrongly skipped from getting cached. It confuses region caching ratio metrics, which the cache-aware balancer uses to track how much each region is cached on individual region servers. That affects the efficiency of the cache-aware balancer, which can lead to read performance degradation once regions are moved to region servers with fewer blocks in the cache for the given region.

Upgrade to CDH-7.2.18.300 or a newer version.

HBASE-28804: Bucket cache retrieval from a persistent file is not asynchronous

When running with a cache close to capacity (1.6TB) and around 30M blocks in the cache, the recovery of the persistent cache following crashes or restarts can delay the region server initialization to about four minutes. That causes rolling restarts or upgrades to fail.

Upgrade to the 7.3.1 version and add more nodes to spread the blocks around more nodes, reduce cache usage on individual region servers, or increase the configured block size to reduce the number of blocks.

HBASE-28805: Unable to perform chunked persistence of backing map for persistent bucket cache

The existing HBase cache persistence flushes the whole cache index into the persistent cache file as a single protocol buffer message. When the cache usage is high, with more than 24M blocks in the cache, the persistence thread gets an error and aborts. The thread does not retain the cache again, leading to a huge cache loss during a crash or restart.

Upgrade to the 7.3.1 version, or add more nodes to spread the blocks around more nodes and reduce cache usage on individual region servers, or increase the configured block size to reduce the number of blocks.