HDP-2.3.6 Release Notes
Also available as:
PDF

HBase

In addition to any Apache patches ported over from the 2.4.x core, HDP 2.3.6 provides HBase 1.1.2 and the following Apache patches:

  • HBASE-13318: RpcServer.getListenerAddress should handle when the accept channel is closed.

  • HBASE-13639: SyncTable - rsync for HBase tables.

  • HBASE-14347: Add a switch to DynamicClassLoader to disable it.

  • HBASE-14604: Improve MoveCostFunction in StochasticLoadBalancer.

  • HBASE-14606: TestSecureLoadIncrementalHFiles tests timed out in trunk build on Apache.

  • HBASE-14655: Narrow the scope of doAs() calls to region observer notifications for compaction.

  • HBASE-14674: Rpc handler / task monitoring seems to be broken after 0.98.

  • HBASE-14840: Sink cluster reports data replication request as success though the data is not replicated.

  • HBASE-14941: locate_region shell command.

  • HBASE-14976: Add RPC call queues to the web UI.

  • HBASE-15039: HMaster and RegionServers should try to refresh token keys from zk when facing InvalidToken.

  • HBASE-15066: Small improvements to Canary tool.

  • HBASE-15125: HBaseFsck's adoptHdfsOrphan function creates region with wrong end key boundary.

  • HBASE-15128: Disable region splits and merges switch in master.

  • HBASE-15145: Default hbase-env.sh breaks add peer command that writes to /hbase-secure/replication znode for kerberized clusters.

  • HBASE-15147: Shell should use Admin.listTableNames() instead of Admin.listTables().

  • HBASE-15252: Data loss when replaying wal if HDFS timeout.

  • HBASE-15295: MutateTableAccess.multiMutate() does not get high priority causing a deadlock.

  • HBASE-15592: Print Procedure WAL content.

  • HBASE-15613: TestNamespaceCommand times out.

  • HBASE-15622: Superusers does not consider the keytab credentials.

  • HBASE-15720: Print row locks at the debug dump page.

  • HBASE-15811: Batch Get after batch Put does not fetch all Cells We were not waiting on all executors in a batch to complete. The test for no-more-executors was damaged by the 0.99/0.98.4 fix "HBASE-11403 Fix race conditions around Object#notify".

  • HBASE-15837: MemStore size accounting is wrong if postBatchMutate() throws exception.

HDP 2.3.6 provides HBase 1.1.2 and ports the following Apache patches from the 2.4.x core:

  • HBASE-14107: Administrative Task: Provide an API to List all procedures.

  • HBASE-14108: Administrative Task: provide an API to abort a procedure.

  • HBASE-14283: Reverse scan doesn’t work with HFile inline index/bloom blocks.

  • HBASE-14307: Incorrect use of positional read API in HFileBlock.

  • HBASE-14432: Enforce ACL on procedure admin tasks.

  • HBASE-14468: Compaction improvements: FIFO compaction policy.

  • HBASE-14471: Thrift - HTTP Error 413 full HEAD if using Kerberos authentication.

  • HBASE-14487: Shell command to list all procedures.

  • HBASE-14488: Shell command to abort a procedure.

  • HBASE-14575: Relax region read lock for compactions.

  • HBASE-14761: Deletes with and without visibility expression do not delete the matching mutation.

  • HBASE-14812: Fix ResultBoundedCompletionService deadlock.

  • HBASE-14866: VerifyReplication should use peer configuration in peer connection.

  • HBASE-14883: TestSplitTransactionOnCluster#testFailedSplit flakey.

  • HBASE-14928: Start row should be set for query through HBase REST gateway involving globbing option.

  • HBASE-14963: Remove Guava dependency from HBase client code.

  • HBASE-14987: Compaction marker whose region name doesn't match current region's needs to be handled.

  • HBASE-15014: Fix filterCellByStore in WALsplitter is awful for performance.

  • HBASE-15022: replication_admin.rb throws undefined method 'getZooKeeperClusterKey' for ZKUtil.

  • HBASE-15035: bulkloading HFiles with tags that require splits do not preserve tags.

  • HBASE-15220: Change two logs in SimpleRegionNormalizer to INFO level.

  • HBASE-15221: Reload the cache on re-tried puts in HTableMultiplexer and adds a close() method to HTableMultiplexer.

  • HBASE-15232: Handle region location cache management in AsyncProcess for multi()'s.

  • HBASE-15621: Suppress HBase SnapshotHFile cleaner error messages when a snapshot is going on.

HDP 2.3.4.7 provided HBase 1.1.2 with no additional Apache patches.

HDP 2.3.4 provided HBase 1.1.2 and the following Apache patches:

  • HBASE-13103: [ergonomics] add region size balancing as a feature of master.

  • HBASE-13250: chown of ExportSnapshot does not cover all path and files.

  • HBASE-14207: Region was hijacked and remained in transition when RS failed to open a region and later regionplan changed to new RS on retry.

  • HBASE-14268: Improve KeyLocker.

  • HBASE-14280: Bulk Upload from HA cluster to remote HA HBase cluster fails.

  • HBASE-14309: Allow load balancer to operate when there is region in transition by adding force flag.

  • HBASE-14314: Metrics for block cache should take region replicas into account.

  • HBASE-14342: Recursive call in RegionMergeTransactionImpl.getJournal().

  • HBASE-14359: HTable#close will hang forever if unchecked error/exception thrown in AsyncProcess#sendMultiAction.

  • HBASE-14361: ReplicationSink should create Connection instances lazily.

  • HBASE-14367: Add normalization support to shell.

  • HBASE-14445: ExportSnapshot does not honor -chmod option.

  • HBASE-14463: Severe performance downgrade when parallel reading a single key from BucketCache.

  • HBASE-14475: Region split requests are always audited with HBase user rather than request user.

  • HBASE-14497: Reverse Scan threw StackOverflow caused by readPt checking.

  • HBASE-14536: Balancer and SSH interfering with each other leading to unavailability.

  • HBASE-14565: Make ZK connection timeout configurable in MiniZooKeeperCluster.

  • HBASE-14578: URISyntaxException during snapshot restore for table with user defined namespace.

  • HBASE-14581: Znode cleanup throws auth exception in secure mode.

  • HBASE-14591: Region with reference HFile may split after a forced split in IncreasingToUpperBoundRegionSplitPolicy.

  • HBASE-14605: Split fails due to 'No valid credentials' error when SecureBulkLoadEndpoint#start tries to access HDFS.

  • HBASE-14621: ReplicationLogCleaner stuck on RS crash.

  • HBASE-14624: BucketCache.freeBlock is too expensive.

  • HBASE-14631: Region merge request should be audited with request user through proper scope of doAs() calls to region observer notifications.

  • HBASE-14632: Region server aborts due to unguarded dereference of Reader.

  • HBASE-14680: Two configs for snapshot timeout and better defaults.

  • HBASE-14706: RegionLocationFinder should return multiple servernames by top host.

  • HBASE-14759: Avoid using Math.abs when selecting SyncRunner in FSHLog.

  • HBASE-14788: Splitting a region does not support the hbase.rs.evictblocksonclose config when closing source region.

  • HBASE-14799: Commons-collections object deserialization remote command execution vulnerability.

  • HBASE-14809: Grant / revoke Namespace admin permission to group.

  • HBASE-14885: NullPointerException in HMaster#normalizeRegions() due to missing TableDescriptor.

  • HBASE-14893: Race between mutation on region and region closing operation leads to NotServingRegionException.

HDP 2.3.2 provided HBase 1.1.2 and the following Apache patches:

  • HBASE-14258 Make region_mover.rb script case insensitive with regard to hostname

  • HBASE-14258 Make region_mover.rb script case insensitive with regard to hostname

  • HBASE-14269 FuzzyRowFilter omits certain rows when multiple fuzzy key exist

  • HBASE-14302 TableSnapshotInputFormat should not create back references when restoring snapshot

  • HBASE-14313 After a Connection sees ConnectionClosingException it never recovers

  • HBASE-14449 Rewrite deadlock prevention for concurrent connection close

  • HBASE-14474 DeadLock in RpcClientImpl.Connection.close()

HDP 2.3.0 provided HBase 1.1.1 and the following Apache patches:

  • HBASE-11658 Piped commands to HBase shell should return non-zero if shell command failed

  • HBASE-11940 Add utility scripts for snapshotting / restoring all tables in cluster