AsyncHBase 1.6.0 has been released

[tsuna <ts...@gmail.com>]

Hi all,
After a fairly long hiatus, I'm happy to announce that AsyncHBase
1.6.0 is now officially out.  This is mostly a bug fix release, and
all users are encouraged to upgrade.

AsyncHBase continues to hold on to its promise of keeping its API
backward compatible, while keeping compatibility under the hood with
all production releases of HBase of the past few years, while
providing better performance and vastly better concurrency than
HBase’s built-in client.

This release was tested against HBase 0.89, 0.90, 0.92, 0.94, 0.96,
and 0.98.  It hasn’t been tested yet against the developer preview
releases of the 0.99 train.  Any feedback there would be interesting.


Tag: https://github.com/OpenTSDB/asynchbase/tree/v1.6.0
The release is also available in Maven central (it’s percolating to
its final destination as I write this email).


* Version 1.6.0 (2014-10-07) [a56249b]

This is mostly a bugfix release, but the introduction of a few new APIs made
it become 1.6.0 instead of 1.5.1.

New public APIs:
  - Introduction of RegionMovedException.
  - Introduction of a lot of new filters (BinaryComparator,
    BinaryPrefixComparator, BitComparator, CompareFilter,
    DependentColumnFilter, FamilyFilter, FilterComparator,
    QualifierFilter, RegexStringComparator, RowFilter,
    SubstringComparator, TimestampsFilter, ValueFilter).
Noteworthy bug fixes:
  - When a region was unavailable, some RPCs could be spuriously retried more
    than once, which could lead to double-counting when the RPC was an atomic
    increment (#81).
  - Fix an error handling bug that arises when an RPC response fails to be
    de-serialized properly (something that shouldn't happen in the first
    place).
  - Fix a compatibility issue with the latest versions of CDH5 that were
    causing scanners to raise an InvalidResponseException.
  - Properly handle RegionMovedException.
  - Make our probes to recover from moving regions more lightweight by
    crafting an unlikely key that is very unlikely to be present in the table,
    since unfortunately exists() still materializes the row in memory in the
    RegionServer, which is problematic for tables with big rows (#82).
  - Fixed an off-by-one in the serialization of batched RPCs that could lead
    to an uncaught IndexOutOfBoundsException.


$ git diff --stat v1.4.1..v1.5.0 | tail -n 1
 54 files changed, 2199 insertions(+), 43 deletions(-)

$ git shortlog v1.5.0..v1.6.0
Aditya Kishore (1):
      MultiAction failure does not clone the HBaseException correctly

Benoit Sigoure (9):
      Start v1.6.0
      Update third-party dependencies.
      Fail the RPC when we fail to de-serialize its response.
      Catch up with the protobuf changes in upstream HBase 0.98.
      Fix increment coalescing integration tests for HBase 0.98
      Fix off-by-one error in error handling of batched edits.
      Craft a very unlikely key when sending probe "exists" RPCs.
      Complain when a DeleteRequest is created with a zero-length
qualifier list.
      Release v1.6.0

Dan Burkert (1):
      Add comparison filters and a timestamp filter.

Jesse Chang (1):
      Retry RPC at RegionMovedException.

Philip K. Warren (1):
      Fix compatibility issue with CDH5.

dvdreddy (2):
      Add tests to reproduce AlreadyNSREd bug #81
      Do not schedule RPCs known to be NSRE'd for retry when they are
first issued.

Thanks to the people above who contributed to this release.

-- 
Benoit "tsuna" Sigoure