Permalink
Commits on Jan 12, 2017
  1. JCBC-999: Redispatch config messages if failed.

    Motivation
    ----------
    If messages going down the config service in core-io, because the
    sockets are opened dynamically and other components depend on this
    fail fast mode (while not ideal), such messages need to be retried
    aggressively to overcome issues where a node is down for example.
    
    Modifications
    -------------
    Simple retry logic has been added to the config service messages,
    retrying after 100ms and as a result providing a better experience,
    even if a node is down or not reachable.
    
    In the future, if needed, we can make this configurable but for
    now lets just stick with sane defaults.
    
    Result
    ------
    Bucket and Cluster-level config messages have a much higher chance
    of succeeding, even under individual node failures.
    
    Change-Id: If27355d444fc411dd933a65e74c07816cb5dd376
    Reviewed-on: http://review.couchbase.org/71897
    Tested-by: Michael Nitschinger <michael@nitschinger.at>
    Reviewed-by: Sergey Avseyev <sergey.avseyev@gmail.com>
    Reviewed-by: Subhashni Balakrishnan <b.subhashni@gmail.com>
    daschl committed Jan 12, 2017
  2. JCBC-1037: Fail FTS Query with IndexDoesNotExistException

    Motivation
    ----------
    Right now if an index does not exist we don't fail with the right
    error message for FTS.
    
    Modifications
    -------------
    This changeset makes the response explicitly check if the index
    does not exist and if so raises an IndexDoesNotExistException
    as part of the error chain with a FTS response.
    
    By the way, this also fixes the longer standing issue that the
    FTS queries should be skipped if the index is not found, leading
    to failing tests.
    
    Result
    ------
    Proper awareness if an FTS index doesn't exist, and also fixing
    failing tests and instead ignoring them.
    
    Change-Id: I721677bfbb489ebd92ee9b91ab7325fc2f71cf59
    Reviewed-on: http://review.couchbase.org/71856
    Tested-by: Michael Nitschinger <michael@nitschinger.at>
    Reviewed-by: Subhashni Balakrishnan <b.subhashni@gmail.com>
    daschl committed Jan 11, 2017
Commits on Jan 5, 2017
  1. Start 2.4.1 Development

    Change-Id: I3b57ec404f02fbaafe9fc3e3db71b51d87a613b2
    Reviewed-on: http://review.couchbase.org/71598
    Reviewed-by: Michael Nitschinger <michael@nitschinger.at>
    Tested-by: Michael Nitschinger <michael@nitschinger.at>
    daschl committed Jan 5, 2017
  2. Prepare 2.4.0 Release

    Change-Id: If48c7fad09c8043af6b5eb8077234a13f0344d17
    Reviewed-on: http://review.couchbase.org/71590
    Reviewed-by: Michael Nitschinger <michael@nitschinger.at>
    Tested-by: Michael Nitschinger <michael@nitschinger.at>
    daschl committed Jan 5, 2017
Commits on Jan 3, 2017
  1. Ignore Guava collection tests by default.

    When they are enabled by default and used with the maven command
    line runner it seems like the test suite never ends. We need to
    look into this separately but for now we can unblock the testing.
    
    Change-Id: I257e3dc6181e0e1f472f91c5b12eaff99c04a79f
    Reviewed-on: http://review.couchbase.org/71470
    Reviewed-by: Sergey Avseyev <sergey.avseyev@gmail.com>
    Reviewed-by: Subhashni Balakrishnan <b.subhashni@gmail.com>
    Tested-by: Michael Nitschinger <michael@nitschinger.at>
    daschl committed Jan 2, 2017
Commits on Jan 2, 2017
  1. Prevent NPE on toString()

    Change-Id: If9c048963233597f4f184d2ae01317dcd207eb2c
    Reviewed-on: http://review.couchbase.org/71468
    Reviewed-by: Sergey Avseyev <sergey.avseyev@gmail.com>
    Tested-by: Michael Nitschinger <michael@nitschinger.at>
    daschl committed Jan 2, 2017
  2. JCBC-1012: Add support for BigInteger and BigDecimal

    Motivation
    ----------
    It has been reported several times (on the forums and through customer
    interactions) that BigDecimal and BigInteger support through the JsonObject
    and JsonArray types would be appreciated.
    
    Modifications
    -------------
    This changeset enables this functionality and adds test cases. Both JsonObject
    and JsonArray now have more getters and setters for these types, but there
    is one gotcha. By default jackson will read the BigDecimal as a double, which
    is the type returned. For very long numbers there might be precision loss.
    
    To combat this, the system property com.couchbase.json.decimalForFloat has
    been introduced which, when set to true, will always read doubles as BigDecimal
    on the jackson side and therefore trade speed for accuracy. Separate tests
    have been added to assert this on a unit-test level.
    
    Result
    ------
    It is now possible to use BigInteger and BigDecimal types the same way as
    one would do with other already supported types.
    
    Change-Id: I357a9533692d705510018228fcadeec760f811f5
    Reviewed-on: http://review.couchbase.org/71435
    Tested-by: Michael Nitschinger <michael@nitschinger.at>
    Reviewed-by: Subhashni Balakrishnan <b.subhashni@gmail.com>
    Reviewed-by: Sergey Avseyev <sergey.avseyev@gmail.com>
    daschl committed Dec 30, 2016