Skip to content

Latest commit

 

History

History
120 lines (101 loc) · 3.54 KB

RELEASE_NOTES.md

File metadata and controls

120 lines (101 loc) · 3.54 KB
Raw

5.1.3

  • Bug fix the batch increment method

5.1.2

  • Sort rollups before creating metrics for them, so different jvms create them in the same order
  • precreate metrics (with zero counts) for each rollup

5.1.1

  • Move writes meter to writeAsync(Batch batch), which is always called, as opposed to writeAsync(WriteBuilder, Op), which clients may choose not to call.

5.1.0

  • Add per-rollup metrics to DataCubeIo
  • Due to possible high cardinality, DataCubeIo constructors now require a boolean controlling whether those metrics are maintained
  • New constructor for HBaseIdService
  • Remove DataCubeIo constructor missing metrics scope

5.0.0

  • Bump to java 1.8.
  • Move from gauava to java optionals
  • Format every file according to intellij auto format.
  • Remove commented out code
  • Remove the "hacklog"
  • Remove unnecessary public and static declarations from enums and interfaces.
  • Remove unused parameters and classes
  • Add a batch increment operation, give the db harness a shutdown method.
  • New constructor for HBaseIdService

4.1.1

  • Use a single thread pool for async id service lookups
  • Make BoxedByteArray serializable

4.1.0

  • Execute id service lookups for multi-gets in a thread pool

4.0.5

  • Timer around entire multi-get reads

4.0.4

  • Metrics for id services in multi-get

4.0.3

  • Naive implementation of multiget for the map db harness.

4.0.2

  • Add metrics around retrying after IOExceptions when flushing to hbase and apply some jittered backoff as retry attempts increase.

4.0.1

  • Removes bufferpool, garbagecollector, and memorypool metrics from static Metrics registry. If clients would like these metrics to be reported on they can be added manually.

4.0.0

  • Modifies how JmxReporter does object naming.

3.0.0

  • Updates metrics library to io.dropwizard.metrics instead of com.yammer.metrics. This will cause slight changes in the naming of metrics. Fixes bug where a checkAndPut operation checking for absence of a cell value is done with an empty array instead of a null value.

2.0.0

  • Non compatible update with existing data cubes. The DataCube constructor in previous versions with the useAddressPrefixByteHash parameter set to true is compatible with PREFIX_MODE.NO_ADDRESS_PREFIX, but is not compatible with the new mode PREFIX_MODE.MOD_ADDRESS_PREFIX because there was a bug since 1.4.0 that didn't implement this feature correctly.

1.5.0

  • Prevent getId calls from creating a new identifier
  • Add a getOrCreateId Method to the idmapper interface
  • Optimize id creation.

1.4.0

  • Add functionality to optionally add a hash in front of row keys, which permits users to ignore dimension order when considering performance.

1.3.0

  • Add on-success callback to the HBaseDbharness
  • Update many dependencies and increase minor version, 1.2.x -> 1.3.0
  • Support multiGet
  • Fix bug in Cas operations
  • Add Double op
  • Name the flusher threads better.
  • Allow for larger caches that will invalidate under memory pressure.
  • Expose cache metrics.
  • Redesign bucketer system for multiple buckets per dimension
  • Add release notes
  • Add instructions/workflow for doing a release
  • Add worked example about counting tweets, initial simple version
  • Include tests as a build artifact

1.2.3

  • Deprecated RollupFilter in favor of low-level Address/Batch access
  • Bug fix for overcounting in hbase backfill merger when cubes share a table
  • Bug fix for EnumSerializable returning incorrect values
  • Nullable dimensions
  • Add week and year buckets to HourDayMonthBucketer