ApsaraDB for HBase:Release notes for minor versions

1.4.6.2:

• Write Ahead Log (WAL) corruption caused by HDFS bugs is fixed.

• Anomalies that occur when you write data into HDFS are fixed. The anomalies may bring the entire RegionServer down and you cannot recover the RegionServer.

• The issue where the system returns the “llegalStateException: No result & no exception” error instead of the NotServingRegionException error during the region splitting process is fixed.

• The issue where the read/write splitting configuration causes RegionServer launch failures is fixed.

• The bug where Phoenix Community Edition continuously creates ZooKeeper connections until the number of connections reaches the upper limit is fixed.

• The configurations to minimize the chances of out of memory (OOM) errors on RegionServers are optimized. The amount of memory resources consumed by wide-scale scans is significantly reduced.

• The kernel is optimized.

1.4.6.3:

• The issue where an index table created by using IndexTool cannot be found if the table name is specified in lower case is fixed.

• The issue where the time zone of the client does not take effect after the time zone is modified is fixed. The default time zone is GTM.

• Default parameters for small-sized ApsaraDB for HBase clusters are optimized.

• The maximum mutation size and bytes to avoid indexing failures on large cells are adjusted.

• Raw scan is used to replace major compaction in order to minimize the chances of index tables being disabled.

• Index writing is forced upon failures to avoid disabling index tables. If the index tables are disabled, the system scans the entire primary table upon queries.

• Memory resource consumption and the number of garbage collections are reduced by disabling scan block cache when upsert-select and create index are executed.

• An NPE error is returned if stale KVs are removed by replication after a column family is deleted.

• Port customization is supported by MiniHBaseCluster.

1.4.9.1:

• Cold data storage and hot data storage in the same ApsaraDB for HBase cluster are supported.

• The Zstandard and Lempel-Ziv-Oberhumer (LZO) compression algorithms are supported.

• The issue where Multiversion Concurrency Control (MVCC) stops responding is fixed.

1.5.0:

• Replication, and synchronization between the primary and secondary clusters are supported.

• A memory leak issue is fixed and reported to the Apache HBase community.

• The issue where the Reader thread is terminated without the RegionServer being aborted when an OOM error occurs is fixed. This issue is reported to the Apache HBase community.

1.5.1:

• Compatibility issues caused by the Zstandard compression algorithm are fixed.

• Some performance issues in cold data storage are fixed.

• Some issues in Phoenix secondary indexes are fixed.

• A memory leak issue is fixed and reported to the Apache HBase community.

• Other bugs are fixed.

1.5.2:

• Automatic locking is supported when only a small amount of disk space remains.

• PREFIX_TREE encoding is not supported when you create a table.

• Some issues are fixed.

1.5.3:

• A switch for dynamically enabling and disabling log subscription is added.

• You are not allowed to use the Hbck command by default.

• Monitoring active connections on the Thrift server is supported.

• Some issues are fixed.

1.5.4: The length of keys and values is limited to prevent OOM errors caused by exceptional requests.

1.5.5: RegionServer failures caused by HLog Exporter are fixed.

1.5.5.1: The logic used to calculate the sizes of reference files is updated.

2.0.2:

• ApsaraDB for HBase Enterprise Edition is released for commercial use.

• ApsaraDB for HBase Basic Edition V1.1 is upgraded to ApsaraDB for HBase V2.0.

• A large number of bugs are fixed and reported to the Apache HBase community.

2.0.3: Some stability issues of AssignmentManager V2 are fixed.

2.0.4:

• Cold data storage and hot data storage in the same ApsaraDB for HBase cluster are supported.

• Some stability issues are fixed.

2.0.5:

• The Phoenix 5.x compatibility issue is fixed.

• Some performance issues in cold data storage are fixed.

2.0.6:

• The Zstandard compression algorithm is supported.

• Some stability issues are fixed and reported to the Apache HBase community.

2.0.7:

• Automatic locking is supported when only a small amount of disk space remains.

• The issue where the Phoenix WAL cannot be cleared is fixed.

• Log subscription and consumption of the RegionServer are supported.

2.0.8:

• A switch for dynamically enabling and disabling log subscription is added.

• You are not allowed to use the Hbck command by default.

2.0.9:

• The DFS client memory leak issue is fixed.

• The issue where SplitTable blocks ModifyTable is fixed.

• The issue where the WAL is corrupted after you enable ASYNC_WAL is fixed.

2.0.10:

• Compaction interruption is supported.

• The numbers of tables and regions supported by a RegionServer can be limited.

2.0.11:

• Request throttling is supported.

• The issue that Disable Compaction does not take effect is fixed.

2.0.12:

• RegionServer failures caused by HLogExporter are fixed.

• By default, you are not allowed to use the Medium Object Storage (MOB) feature.

2.0.13: Memory leaks caused by enabling mslab are fixed.